SEM
methodology
_________________________________________________________________
Bollen,
K. A. (1989). Structural equations with
latent variables. Wiley Series in
Probability and Mathematical Statistics. New York: Wiley.
Bollen,
K.A., & Long, J.S. (eds.) (1993).
Testing structural equation models. Newbury Park, CA: Sage.
Bollen,
K.A. and R. Lennox. 1991. "Conventional Wisdom on Measurement: A
Structural Equation Perspective."
Psychological Bulletin, 110: 305-314.
available in .pdf format from Bollen's web
site at http://www.unc.edu/~bollen/semwork.html
Bollen,
K.A. (2002). Latent variables in psychology and the social sciences. Annual
Review of Psychology, 53, 605-634.
.pdf available at
http://psych.annualreviews.org/current.shtml
____________________________________________________
Bentler,
P.M., & Dudgeon , P. (1996).
Covariance structure analysis: Statistical practice, theory, and
directions. Annual Review of
Psychology, 47, 563-592.
full text available at:
http://psych.annualreviews.org/cgi/content/full/47/1/563
DMT: In experimental designs, the (linear) model
is assumed to be true, and we test null hypotheses about whether certain
parameters are equal to 0. In
non-experimental studies, like the ones we investigate using SEM, the model
itself is under dispute, along with the values of parameters. Nonzero and significant parameters are
interesting iff and only iff the model is "plausible" [as a description of the data generating
process (DGP)].
Bentler
highlights (p. 566) the "alarming" lack of attention to the normality
of variables in studies to which SEM is applied. The distribution of Xi (exogenous latent variables) and epsilon
(disturbance terms associated with latent factors) are important for
understanding normality but they are unknown.
The only information about distributional properties that we can
investigate comes from the indicators x.
The
choice of a discrepancy function (see Kaplan, pp. ???) (ML or GLS, for example)
influences model assessment. However,
because H naught strictly never holds, this is a problem. Sample size also influences the behavior of
the discrepancy function (p. 571).
Z tests
of parameters being equal to zero depend on estimates of the standard errors of
the parameters. This SE is an element
of the inverse of the [Fisher's] information matrix (see Freund, Kaplan)
He
focuses on the effects of choosing a discrepancy function, often some form of
ML estimator (which explicitly assumes multivariate normal distribution of the
measured and latent variables, but can also take the form of a WLS or GLS
estimator, which can be asymptotically distribution free (ADF).
Child
Development (February 1987, Vol. 58, No. 1) is a special issue devoted to SEM.
The journal is available at Bird
Library. Blackwell Publishers currently
manage the journal Child Development, and their current inventory dates to
1996. Earlier issues are often
available through Society for Research in Child Development (SCRD). According to Suzanne E Kelley, Academic
Permissions Mgr, Society for Research in Child Development (Phone:
734.998.6578 Fax: 734.998.6569 E-mail: sekelley@umich.edu), who emailed me on
3-29-02, the only place that still has a copy of the issue is:
Bell
& Howell Information & Learning (now Proquest, Inc.)
300
N. Zeeb Rd.
Ann
Arbor, MI 48106-1346
Phone:
800-521-0600
you
can only purchase by the entire volume; price for 1987 is $64.30.
_______________________________________________________________________
Bentler,
P.M. (1980). Multivariate analysis with latent variables: Causal modeling.
Annual Review of Psychology, 31, 419-456.
DMT:
generally cited in earlier articles as the key review of the field. It lays out the overall ideas with path
diagrams, and focuses on key areas of concern.
____________________________________________________
Glaser,
D. (2002). Structural Equation Modeling Texts: A primer for the beginner.
Journal
of Clinical Child Psychology, 31(4), 573-578.
____________________________________________________
Yuan,
K.-H., & Bentler, P. M. (1997). Mean and covariance structure analysis:
Theoretical
and practical improvements. Journal of the American Statistical
Association,
92, 767-774.
__________________________________________________________
Yuan,
K.H., Chan, W., & Bentler, P.M. (2000). Robust transformation with
applications to structural equation modelling. British Jounral of Mathematical
and Statistical Psychology, 53, 31-50.
Data sets in social and behavioural
sciences are seldom normal. Influential cases or outliers can lead to
inappropriate solutions and problematic conclusions in structural equation
modelling. By giving a proper weight to each case, the influence of outliers on
a robust procedure can be minimized. We propose using a robust procedure as a
transformation technique, generating a new data matrix that can be analysed by
a variety of multivariate methods. Mardia's multivariate skewness and kurtosis
statistics are used to measure the effect of the transformation in achieving
approximate normality. Since the transformation makes the data approximately
normal, applying a classical normal theory based procedure to the transformed
data gives more efficient parameter estimates. Three procedures for parameter
evaluation and model testing are discussed. Six examples illustrate the various
aspects with the robust transformation.
__________________________________________________________
Baumrind,
D. (1983). Specious causal attributions in the social sciences: The
reformulated stepping-stone theory of heroin use as exemplar. Journal of
Personality and Social Psychology, 45(6), 1289-1298.
__________________________________________________________
Cliff,
N. (1983). Multivariate Behavioral
Research, 18, 115-126.
Journal is available at Bizzell: BF39 /.M852
DMT: This is an early cautionary article,
qualitative in nature that reminds readers about the lack of equivalence
between correlation and causation. He
also defines and expands on the nominalistic fallacy with regard to latent
variables that he feels are given too much credence when they are discovered
via structural equation models. He
reminds readers that latent variables never "emerge;" they remain
forever latent.
________________________________________________________
______________________________________________________
Breckler,
S. J. (1990). Applications of covariance structure modeling in psychology:
Cause for Concern? Psychological Bulletin, 107, 260-273.
DMT
comments: Breckler's article begins
with a precise and excellent summary of the matrix models, including the
parameter matrix, and their dimensions.
He
outlines numerous concerns that relate to the practice of CSM. These include the problems with violating
the assumption of multivariate normality, and the need to use asymptotic
distribution free (ADF) estimation methods; problems with identification; the
nominalistic fallacy, that is the notion that latent variables actually exist
just because we can name them. He
argues (a la Baumrind, 1983) against using the term "causal
modeling," and reminds readers that covariance or correlational data
cannot imply causation.
He is
especially strong at showing the kinds of equivalent models that routinely
exist, and offers several examples of studies where existing equivalent models
have strongly different substantive interpretations than the published
ones. In particular, he offers dramatic
examples of equivalent models where directional effects are precisely reversed,
but produce implied (or fitted) covariance matrices and fit indices that are
identical with the published ones.
He
discusses model modification (what other authors call "specification
searches," and the problem with producing a model that takes advantage of
a particular dataset's peculiarities.
He points out that modification indices guide post hoc modification by
predicting the minimum decrease that will occur in the overall chi-square
statistic when a particular parameter is freed for estimation (hence with a
decrease of one degree of freedom.) He
strongly urges not only cross-validation, but better reporting of the various
modification steps that separate a priori models from those that are ultimately
published and advocated. He also offers
a list of possible and permissible model modifications.
__________________________________________________________
Estimation
__________________________________________________________
Bentler,
P.M. and Bonett,D.G. (1980) "Significance tests and goodness of fit in the
analysis of covariance structures", Psychological Bulletin, 88: 588-606.
Another classic and much-cited paper which
is a good introduction to the
problems
of assessing goodness of fit.
Bentler,
P.M. (1983). Some contributions to efficient statistics for structural models:
specification and estimation of moment structures. Psychometrika 48:493-517.
Widely quoted early source on
"Bentler-Weeks" models
Bentler
PM, Dijkstra T. 1985. Efficient estimation via linearization in structural
models. In Multivariate Analysis VI, ed. PR Krishnaiah,pp. 9-42. Amsterdam:
North Holland
Bentler,
P.M., & Yuan, K.-H. (1999). On adding a mean structure to a covariance
structure model. (UCLA Statistical
Series # 247)
Preprint available at:
preprints.stat.ucla.edu/247/247.pdf
Bollen,
1990, Overall fit in covariance structure models: Two types of sample size
effects. Psychological Bulletin, 107, 256-259.
Browne,
M.W. (1984). Asymptotically distribution-free methods for the analysis of
covariance structures. British Journal
of Math Stat Psychol 1984 May;37 ( Pt 1):62-83
Journal is available at Bizzell Library
Call BF1 / .B725
Enders,
C.K. (2001). A primer on maximum
likelihood algorithms available for use with missing data. Structural Equation Modeling, 8, 128-141.
Kano Y,
Berkane M, Bentler PM. 1990. Covariance structure analysis with heterogeneous
kurtosis parameters. Biometrika 77:575-85
Originators of a GLS estimator under
conditions of heterogenous kurtosis (HK)
Muthen,
B. (1993). Goodness of fit with categorical and other nonnormal variables. In
K.A. Bollen & J.S. Long (Eds.), Testing structural equation models (pp.
205-234). Newbury Park, CA: Sage.
Yuan,
K.-H., & Bentler, P. M. (1997). Mean and covariance structure analysis:
Theoretical
and practical improvements. Journal of the American Statistical
Association,
92, 767-774.
Yuan,
K.-H., & Bentler, P.M. (2000). A unified approach to structural equation
modeling with nonstandard samples. (UCLA Statistical Series # 276)
Preprint available at:
preprints.stat.ucla.edu/276/276.pdf
__________________________________________________________
__________________________________________________________
Kenny,D.A.
and Judd,C.M. (1984). Estimating the nonlinear and interactive effects of
latent variables, Psychological Bulletin, 96: 201-210.
Excellent multivariate course web site at:
http://www.psychol.ucl.ac.uk/teaching/c565.html
describes the article as : "A well
described account of an extension of LISREL models to allow for
interactions."
__________________________________________________________
Bentler
(1990), Comparative Fit Indices in Structural Models, Psychological Bulletin,
107, 238-246.
__________________________________________________________
Browne
MW. 1982. Covariance structures. In Topics in Applied Multivariate Analysis,
ed. DM Hawkins,pp. 72-141. Cambridge: Cambridge Univ. Press
Hawkins' text is available in OU's
Chem-Math lab:
Call # QA278 / T66
__________________________________________________________
Shipley,
Bill. (2000). Cause and Correlation in Biology: A User's Guide to Path
Analysis, Structural Equations, and Causal Inference. Cambridge, UK: Cambridge
University. 2000, 317 pages, $69.95 US
(hardcover).
http://www.callisto.si.usherb.ca:8080/bshipley/my%20book.htm
__________________________________________________________
Pearl,
J. (2000). Causality. Models, Reasoning, and Inference. Cambridge University.
Reviewed
by Bill Shipley (2000). Structural Equation Modeling, 7(4), 637-639.
__________________________________________________________
Spanos,
A. (1995). On theory testing in econometrics: Modeling wth non-experimental
data. Journal of Econometrics, 67, 189-226.
__________________________________________________________
Spanos,
A. (1986). Statistical Foundations of Econometric
Modelling</EM></a>. Cambridge University Press.
http://uk.cambridge.org/economics/catalogue/0521269121/default.htm
__________________________________________________________
Spirtes,
P., Glymour, C. & Scheines, R. (1993). Causation, Prediction, and Search.
Springer-Verlag.
_________________________________________________________
Bound,
Jeager and Baker, "Problems with instrumental variables estimation
when
the correlation between the instruments and the endogenous explanatory
variables
is weak," JASA, 90, 1995.
__________________________________________________________
Cohen,
P., Cohen, J., Teresi, J., Marchi, P. & Velez, N. (1990). Problems in the
measurement of latent variables in structural equation models. Applied
Psychological Measurement, 14, 183-186.
__________________________________________________________
MacCallum,
R.C., & Austin, J.T. (2000). Applications of structural equation modeling
in psychological research. Annual Review of Psychology. 51:201-26, 2000.
Ohio
State University, Department of Psychology, Columbus 43210-1222, USA.
maccallum.1@osu.edu
[Review]
[109 refs]
Abstract
This chapter presents a review of
applications of structural equation modeling (SEM) published in psychological
research journals in recent years. We focus first on the variety of research
designs and substantive issues to which SEM can be applied productively. We
then discuss a number of methodological problems and issues of concern that
characterize some of this literature. Although it is clear that SEM is a
powerful tool that is being used to great benefit in psychological research, it
is also clear that the applied SEM literature is characterized by some chronic
problems and that this literature can be considerably improved by greater
attention to these issues. [References: 109]
Full
text available:
http://psych.annualreviews.org/cgi/content/full/51/1/201
__________________________________________________________
Basic
textbooks:
Rex
Kline, "Principles and Practice of Structural Equation
Modeling."
http://www-psychology.concordia.ca/department/faculty/kline/Books/SEM1/sem1.html
Steiger,
J.H. (2001). Driving fast in reverse: The relationship between software
development, theory, and education in structural equation modeling. Journal of
the American Statistical Association, 96, 331-338.
Download .pdf version at
ftp://aep150666.psych.ubc.ca/JASA/Driving.pdf
DMT
comments: Steiger puts SEM in context
of broad class of CSM, where an observed covariance matrix sigma is modeled as
M(theta) where M is a comprehensible design matrix and theta is a vector of
estimable parameters.
He
outlines six areas where introductory texts shortchange practicioners. First among these is the techniques's
failure to produce unique models, and the almost necessary existence of
equivalent models. He next lists the
impenetrability of instances where nonlinear optimization techniques fail to
converge, and argues that most introductory texts skirt the issue by using
trivial examples. He also argues for
more stringent treatment of power and sample size, reporting techniques,
handling of the violation of assumptions, the use of standardized models, and
the status of standard errors when CSM is performed on correlation, not
covariance, matrices.
__________________________________________________________
Maruyama,
G., & McGarvey, B. (1980). Evaluating causal models: An application of
maximum likelihood analysis of structural equations. Psychological Bulletin,
87, 502-512.
Historically
interesting; includes nice definition of exogenous variables as ones whose
cause is unknown or uninteresting...
__________________________________________________________
Equivalent
models
1.
MacCallum,
R.C., Wegener, D.T., Uchino, B.N., & Fabrigar, L.R. (1993). The problem of
equivalent models in applications of covariance structure analysis.
Psychological
Bulletin, 114, 185-99.
For any
given covariance structure model, there will often be alternative models that
are indistinguishable from the original model in terms of goodness of fit to data.
The existence of such equivalent models is almost universally ignored in
empirical studies. A study of 53 published applications showed that equivalent
models exist routinely, often in large numbers. Detailed study of three
applications showed that equivalent models may often offer substantively
meaningful alternative explanations of data. The importance of the equivalent
model phenomenon and recommendations for managing and confronting the problem
in practice are discussed.
DMT
comments: The authors show that
equivalent models are a natural occurrence.
For example, in linear regression, the two models:
Y=XB and X=YB* have identical Rsquare
statistics when they are applied to any set of data. They differ only in their substantive interpretation, on the
logic of X or Y being the more logical choice for a dependent or independent
variable.
They state that ignoring the presence of
equivalent models is common, and liken it to ignoring the possibility of
confounding in ANOVA or similar designs.
The article primarily applies the
systematic findings of earlier authors (Stelzl, Lee and Hershberger) to studies
in various areas of psychological research.
These earlier authors showed that changes in the structural (not the
measurement) model result in equivalent models that fit any data set equally
well. That is, for any two models A and
B, for any sample covariance matrix S, Sigma hat A= Sigma hat B (where sigma
hat is the fitted covariance matrix).
2.
MacCallum,
R.C., & Brown, M.W. (1993). The use of causal indicators in covariance
structure models: Some practical issues.
Psychological Bulletin, 114, 533-541.
DMT: The authors distinguish between effect
indicators and causal indicators. The
former are the typical case, as in factor analysis, where indicators depend on
latent factors, along with a unique error term.
Causal
indicators, on the other hand, can be thought to cause the latent factors with
which they are associated. A typical
example is the relationship between indicators like income and education level
with a latent factor like SES. SES
can't be thought to cause income, more than income can be thought to cause
SES.
3.
Ed
Rigon discusses equivalent models on his SEM web page:
http://www.gsu.edu/~mkteer/equival.html
4.
Lee,
S., & Hershberger, S. (1990). A simple rule for generating equivalent
models in structural equation modeling. Multivariate Behavioral Research, 25,
313-334.
Journal is available at Bizzell: BF39 /.M852
5.
Stelzl,
I. (1986). Changing a causal hypothesis without changing the fit: Some rules
for generating equivalent path models. Multivariate Behavioral Research, 21,
309-31.
Journal
is available at Bizzell: BF39 /.M852
DMT: Stelzl derives four rules by which alternative
path models can be generated that will fit any data just as well as as a P0, or
primary postulated model.
For
example, he shows how the partial correlation method of obtaining estimates of
parameters, because it does not depend on the order, permits the switching of
the arrow of causation between many pairs of free parameters in a restricted or
confirmatory model.
__________________________________________________________
Anderson,
J. C., & Gerbing, D. W. (1988). Structural Equation modeling in practice :
A review and recommended two-step approach. Psychological Bulletin, 103(3),
411-423.
DMT: They distinguish between applications of SEM
for "theory testing and development, in which the goal is to proceed from
exploratory to confirmatory factor analysis, and the appropriate tool is a
method that explains the greatest amoung of covariance among observed
variables. On the other hand, when the
goal is prediction, methods that are based on principal components analysis,
and that explain the greatest amount of variance in a sample, are preferred.
__________________________________________________________
Crowley,
S.L., & Fan, X. (1997). Structural equation modeling: Basic concepts and
applications in personality assessment research. Journal of Personality
Assessment. 68, 508-531.
[Review]
[44 refs] Department of Psychology, Utah State University, USA.
Structural
equation modeling (SEM) has become an increasingly used methodological strategy
in psychology. Nevertheless, many psychologists continue to be unclear about
how to apply this analytic tool in their research. This article reviews SEM
from a conceptual perspective, particularly focusing on confirmatory factor
analysis. Additionally, the relation between SEM and other analytic techniques
(e.g., exploratory factor analysis) are addressed. A confirmatory factor
analytic example is presented and reviewed in detail. Finally, limitations of
SEM and other considerations are discussed. [References: 44] ref)
DMT
comments: Crowley and Fan point out
that overidentified models fit well, even though their large number of free
parameters (and therefore low df) belie a lack of theoretical rigor (p. 516).
They
choose to detail three problematic areas:
1.
multivariate normality, especially as it relates to the common practice of
MLE. They cite:
Fan, X.
(1996). SAS program to assess multivariate normality. Educational and
Psychological Measurement, 56, 668-674.
They
remark that the Satorra-Bentler Scaled Statistic accounts for non-normality in
the assessment of GOF. Alternatively,
ADF estimation methods exist for large samples.
2. target matrix, covariance or
correlation. Use of the correlation
matrix alters values of chi-square, GOF indices, and Standard errors. One alternative is always to use covariance
matrices, but report standardized solutions.
They
cite:
Cudeck (1989). Analysis of covariance
matrices using covariance structure models. Psychological Bulletin, 105,
317-327.
Lee, S.Y. (1985). Analysis of covariance
and correlation structures. Computational Statistics and Data Analysis,
2,279-295.
3. Sample size rules of thumb:
a)
"just have 200" (Boomsma,
1987)
b) n= 5 or 10 for each estimated parameter
(Flyd and Widaman, 1995)
4. post-hoc model modification; there is a cost
associated with "capitalizing on idiosyncracies of data" (p. 527)
__________________________________________________________
Mulligan
S. (1998). Patterns of sensory
integration dysfunction: a confirmatory factor analysis. American Journal of Occupational
Therapy, 52(10):819-28, 1998
Nov-Dec. (36
Occupational
Therapy Department, School of Health and Human Services, University of New
Hampshire, Hewitt Hall, 4 Library Way, Durham, New Hampshire 03824-3563.
Objective.
This study evaluated a five-factor model of sensory integration dysfunction on
the basis of scores of children on the Sensory Integration and Praxis Tests
(SIPT). The purpose of the study was to determine a plausible model for
understanding sensory integration dysfunction. Method. The hypothesized model
of sensory integration dysfunction tested was derived from previous
multivariate analyses and consisted of five patterns of dysfunction, including:
bilateral integration and sequencing, somatosensory, somatopraxis, visuopraxis,
and postural ocular motor. Confirmatory factor analysis (CEA) of the SIPT
scores of 10,475 children and the scores of a subgroup of 995 children with
Learning disabilities were used to evaluate the model. Results. The CFA of the
hypothesized model indicated numerous weaknesses with it and, therefore, was
rejected. Exploratory factor analysis (EFA) was then performed with the same
data set to identify a better-fitting, more parsimonious model!
of sensory integration dysfunction. A
secondorder, four-factor model using generalized practic dysfunction as the
second-order factor and four first-order factors (dyspraxia, bilateral
integration and sequencing deficit, visuoperceptual deficit, somatosensory
deficit) were proposed. The CFA supported this model as the better-fitting
model. The proposed model held true when tested with the subgroup of children
with learning disabilities. Conclusions. The modified model of sensory
integration dysfunction proposed indicated that it was a good fit for the data
and improved on the initial model. Clinical implications of the findings relate
to the interpretation of SIPT scores and provide suggestions for test
development measuring sensory integration functions. The proposed model has applications
for occupational therapy intervention using sensory integration as the primary
frame of reference. (36 ref)
Subject:
mulligan used CFA for a dyspraxia instrument
_______________________________________________________________________
Longitudinal
Models:
_______________________________________________________________________
Bentler,
P.M., & Freeman, E.H. (1983). Tests for stability in linear structural
systems. Psychometrika, 48, 143-146.
Available
at Bird
_______________________________________________________________________
Collins,
L.M., & Sayer, A.G. (Eds.). (2001).
New methods for the analysis of change. Washington, DC: American
Psychological Association.
Abstract
from publicity materials: This volume
presents state-of-the-art methods explored by recognized authorities on the
analysis of change. Chapters highlight methods for estimating and evaluating
models of growth and change over time at the level of the individual; address
issues of measurement that are important in the analysis of change; point out
methods for separating intra-individual growth from some aspects of phenomena
that are stable over time; identify larger frameworks to integrate knowledge;
and provide methods for dealing with missing data. This volume of
methodological advances will influence a variety of disciplines from psychology
and sociology to education and economics.
442pp.
$49.95
available
at Bizzell: BF637 / c4 n47 2001
_______________________________________________________________________
Cudeck,
R. (1986). A note on structural models for the circumplex. Psychometrika, 51,
143-147.
Available
at Bird
________________________________________________________________________
Curran,
P.J., & Bollen, K.A. (2001). The best of both worlds: Combining
autoregressive and latent curve models. To appear in Collins, L. M. &
Sayer, A. G. (Eds.), New methods for the analysis of change, (pp.105-136).
Washington, DC: American Psychological Association.
________________________________________________________________________
Curran,
P.J. maintains a web page that deals with longitudinal data analysis:
http://www.unc.edu/~curran/references.html
______________________________________________________________________
Dumenci,
L, & Windle, M. (1996). A latent trait-state model of adolescent depression
using the center for epidemiologic studies-depression scale. Multivariate
Behavioral Research, 31, 313-330.
Journal
is available at Bizzell: BF39 /.M852
_______________________________________________________________________
Dumenci,
L, & Windle, M. (1998). A multitrait-multioccasion generalization of the
latent trait-state model: Description and application. Structural Equation
Modeling, 5, 391-410.
_______________________________________________________________________
Finch,
J.F., & Zautra, A.J. (1992). Testing latent longitudinal models of social
ties and depression among the elderly - A comparison of distribution-free and
maximum-likelihood-estimates with nonnormal data. Psycholgoy and Aging, 7,
107-118.
Journal
available at Bird
Also
available at Bizzell: BF 724.55 .A35
P79
_______________________________________________________________________
Finkel,
Steven E. (1995). Causal analysis with panel data. Thousand Oaks, CA: Sage.
Available as ebook through Bizzell Library
catalog:
http://libraries.ou.edu/eresources/catalog/
DMT:
Panel data include repeated measures on the same group of subjects. Finkel distinguishes between panels, which
include relatively few variables, consistently measured in two to five waves,
on the same group of subjects, and other data formats like repeated
cross-sections, "time series" that involve individuals, and trend studies
(in which the timing of measurements is inconsistent). He says the difference between panel studies
and pooled time series in less formal, although the latter tends to include a
greater number of measurements, and perhaps a smaller N.
Finkel (p.1) cites Menard (1991) with
three causal principles that we can add to Gollob and Reichardt's (1987)
below.
Finkel shows how a dependent variable or a
discrete inter-panel change can be modeled in terms of simultaneous or lagged
variables, including the dependent variable itself. He shows how models can be manipulated algebraically so that the
coefficients are interpretable in various ways, including as indicators of
equilibrium in the system. He shows how
continuous change models can be restated in terms of panel coefficients so that
"fundamental parameters of change" can be recovered.
p20
_______________________________________________________________________
Goldstein,
H., & McDonald, R.P. (1988). A general model for the analysis of multilevel
data. Psychometrika, 53, 455-467.
_______________________________________________________________________
Gollob,
H. F., & Reichardt, C. S. (1987). Taking account of time lags in causal
models. Child Development, 58, 80-92.
DMT:
The authors articulate "three principles of causal
effects." First, causes can only
come from prior variables. Second,
variables influence themselves through "autoregressive effects." Third, effect size depends on the length of
time that intervenes, and this time lag must be carefully described.
They point out that studies frequently
ignore autoregression, implicitly assuming that autoregressive effects equal
zero. They further aver that
cross-sectional designs necessarily ignore all three principles, and argue that
a preferable approach is a "latent longitudinal" model, in which
cross-sectional data are fit using a longitudinal model that posits latent
variables at some preexisting time t1.
To identify latent longitudinal models so that SEM software can estimate
their parameters, assumptions about the relationships among the latent
variables must be explicit. If the
assumptions are correct, the model's estimates are unbiased. If not, they are necessarily biased. They suggest running both cross-sectional
and latent longitudinal models. If both
models yield similar estimates, they can probably be trusted. If the models yield different estimates, the
latent longitudinal model's estimates are necessarily less untrustoworthy; they
may nevertheless be biased if the assumptions on which the model was identified
are not correct.
A drawing of a latent longitudinal model,
superimposed on a cross sectional model of x and y, is at
http://coph.ouhsc.edu/dthompso/web/sem/latlong.jpg
The authors presented a paper on this
topic in 1986 to the Society of Multivariate Experimental Psychology in
Atlanta, but I'm not sure where it went from there. The ideas may be present in some of Raykov's work at Fordham, and
in an article by Finch.
_______________________________________________________________________
Jöreskog,
K.G. (1970). Estimation and testing of simplex models. British Journal of
Mathematical and Statistical Psychology, 23, 121-145.
Joreskog (1979, p. 340) and Curran and
Bollen (2001, p. 112) attribute the "simplex model" to Guttman
(1954), in describing a "multiwave" model with a single
"fallible" variable, underlied by a latent factor at every time t.
Guttman, L. (1955). A generalized simplex
for factor analysis. Psychometrika, 20, 173-192.
Guttman, L. (1954). A new approach to
factor analysis: the radex. In P. F. Lazarsfeld (Ed.), Mathematical thinking in
the social sciences. New York: Columbia University Press.
Joreskog shows (p. 342) that a four-wave
model is necessarily identified. It
will have t(t+1)/2 non-redundant elements in the sample covariance matrix, and
will have 3t-3 = 3(t-1) free model parameters.
In general the free parameters are the (1) errors, (2) latent factors,
(3) errors in equations, and (4) regression
coefficients associated with each occasion. These would account for 4t free parameters, except that the first
occasion is represented by the model's only exogenous latent variable, which
has no associated regression coefficient or error in its structural equation. This leaves 4t-2 parameters. How do we get from 4t-2 to 3t-3 free
parameters?
Finally, if df= t(t+1)/1 - 3t-3 =
t^2-5t+6, then df > 0 as long as t>3.
_______________________________________________________________________
Jöreskog,
K.G. (1978). An econometric model for multivariate panel data. Annales de
ITNSEE, 30-31, 355-366.
Jöreskog,
K.G. (1979). Statistical estimation of structural models in
longitudinal-developmental investigations. In J.R. Nesselroade & P.B.
Baltes (Eds.), Longitudinal research in the study of behavior and development.
New York: Academic Press, 303-351.
Text
available at Bizzell: bf76.5 / L66
Some
credit Joreskog with sensitizing the modeling community to the issue of
"autoregressive effects."
_______________________________________________________________________
Jöreskog,
K.G. & Sörbom, D. (1977). Statistical models and methods for analysis of
longitudinal data. In D.J. Aigner & A.S. Goldberger (Eds.), Latent
variables in socio-economic models. Amsterdam: North Holland Publishing
Company, 235-285.
Jöreskog,
K.G., & Sörbom, D. (1979). Simultaneous analysis of longitudinal data from
several cohorts. Paper presented at the SSRC Conference on Analyzing
Longitudinal Data for Age, Period and Cohort Effects in Snowmass, Colorado,
June 18-20.
_______________________________________________________________________
Kenny,
D.A. and Zautra, A. (1995). The trait-state-error model for multiwave data.
Journal of Consulting and Clinical Psychology, 63:52-59.
Abstract:
Although
researchers in clinical psychology routinely gather data in which many
individuals respond at multiple times, there is not a standard way to analyze
such data. A new approach for the analysis of such data is described. It is
proposed that a person's current standing on a variable is caused by 3 sources
of variance: a term that does not change (trait), a term that changes (state),
and a random term (error). It is shown how structural equation modeling can be
used to estimate such a model.
DMT: The authors claim that available techniques
like those of West and Hepworth (1991; see below) require 50 or more serial
observations. They cite the need for
techniques that apply to relatively large N observed at relatively few points
in time.
They
relate their approach to
Multiple (and redundant) two-wave models
McCardle and Epstein's (1987, see below)
growth curve modeling
Pooled time series regression (they cite a
text by Dielman 1989, but I haven't looked for references in journals)
Their
latent variable approach considers measurement error among the predictors,
which is an advantage over time series regression.
They
refer to previous latent variable approaches that require more than indicator
per factor. Their approach handles
single indicators, as long as that indicator is measured at at least four time
points; then, the model is identified and estimable. (The idea that four time points are sufficient to identify a
multiwave model was shown by Joreskog (1979, p. 342).
The
authors now call their trait-state-error (TSE) model the STARTS model (stable
trait + autregressive trait + state) and describe it in a chapter in Collins
and Sayer (2001):
Kenny, D.A., & Zautra, A. Trait-state
models for longitudinal data. [Chapter]
_______________________________________________________________________
Labouvie,
E.W. (1981). The study of multivariate change structures: A conceptual
perspective. Multivariate Behavioral Research, 16, 23-25.
Journal
is available at Bizzell: BF39 /.M852
_______________________________________________________________________
Muthen,
B.O. (1991). Analysis of longitudinal data using latent variable models with
varying parameters. In Linda M. Collins and John L. Horn (eds.). Best Methods
for the Analysis of Change : Recent Advances, Unanswered Questions, Future
Directions. (pp. 1-17). Washington, DC:
American Psychological Association.
Available for purchase (APA Members/ Affiliates: $14.95;
Item # 4318071 ; ISBN: 1-55798-310-0
http://mirror.apa.org/books/4318071t.html
_______________________________________________________________________
Olsson,
U., & Bergman, L.R. (1977). A longitudinal factor model for studying change
in ability structure. Multivariate Behavioral Research, 12, 221-242.
Journal
is available at Bizzell: BF39 /.M852
_______________________________________________________________________
Raykov,
T. (1993). A structural equation model for measuring residualized change and
discerning patterns of growth or decline. Applied Psychological Measurement,
17, 53-71.
Raykov,
T. (1994). Studying correlates and predictors of longitudinal change using
structural equation modeling. Applied Psychological Measurement, 18, 63-77.
Raykov,
T. (1997). Simultaneous study of group and individual latent longitudinal
change patterns using structural equation modeling. Structural Equation Modeling, 4, 212-236.
Raykov,
T. (1997). Growth curve analysis of ability means and variances in measures
of fluid intelligence of older adults.
Structural Equation Modeling, 4, 283-319.
Raykov's
writings are collected at his home page:
http://www.fordham.edu/psychology/faculty/raykov.html
_______________________________________________________________________
Rogosa,
D., (1979), Causal models in longitudinal research. In J.R. Nesselroad and
P.D., Baltes (Eds.) Longitudinal Research in the Study of Behaviour and
Development. New York, Academic Press, 263-302.
Text
available at Bizzell: bf76.5 / L66
Rogosa,
D., (1995), Myths and methods: Myths about longitudinal research plus
supplemental questions. In J. M. Gottman (Ed.) The analysis of change. New
Jersey: LEA.
Available
at Bizzell: BF 39 .A476 1995
Rogosa unleashes stinging criticism of
SEM, and charges that LISREL should not be used with longitudinal data. One major problem is that SEM produces
parameter estimates that reflect only the time-structuring of the data, and
otherwise ignore information from the sample covariances.
The article summarizes many of his papers
over the years, many of them critical of approaches to longitudinal data
analysis like simplex models or cross-lagged two-panel approaches.
Rogosa maintains a web site where he
shares the data that he uses in "exhibit 1," his LISREL example:
http://www.stanford.edu/~rag/Myths/myths.html
_______________________________________________________________________
Rovine,
M.J., & Molenaar, P.C.M.
(2001). A structural equations
modeling approach to the general linear mixed model. In L. Collins & A. Sayer (Eds.), New Methods for the Analysis
of Change (pp. 65-96). Washington, DC: American Psychological
Association.
Argues that certain mixed models are
satisfactorily modeled as SEM, in particular because of its ability to include
"patterned" covariance structures for the covariance. This is an advantage over typical approaches
that, like repeated measures analysis of variance, must assume a very simple
covariance structure for errors.
Rutter,
M., (1988), Longitudinal data in the study of causal process: some uses and
some pitfalls. In M Rutter (ed.), Studies of psychosocial risk; the power of
longitudinal data. Cambridge: CUP.
Available
at Bizzell: BF 713.5 .S78 1988
_______________________________________________________________________
Sörbom,
D. (1975). Detection of correlated errors in longitudinal data. British Journal
of Mathematical and Statistical Psychology, 28, 138-151.
Stoolmiller,
M. (1995). Using latent growth curve models to study developmental processes.
In J. M. Gottman (Ed.), The Analysis of Change (pp. 103-138). Mahwah, NJ:
Lawrence Earlbaum Associates.
Werts,
C.E., Linn, R.L., & Jöreskog, K.G. (1977). A simplex model for analyzing
academic growth. Educational and Psychological Measurement, 37, 745-756.
Werts,
C.E., Linn, R.L., & Jöreskog, K.G. (1978). Reliability of college grades
from longitudinal data. Educational and Psychological Measurement, 38, 89-95.
West,
S.G., & Hepworth, J.T. (1991). Statistical issues in the study of temporal
data: daily experiences. [Review] Journal of Personality. 59(3):609-62, 1991
Abstract
This article reviews statistical issues that
arise in temporal data, particularly with respect to daily experience data.
Issues related to nonindependence of observations, the nature of data
structures, and claims of causality are considered. Through the analysis of
data from a single subject, we illustrate concomitant time-series analysis, a general
method of examining relationships between two or more series having 50 or more
observations. We also discuss detection of and remedies for the problems of
trend, cycles, and serial dependency that frequently plague temporal data, and
present methods of combining the results of concomitant time series across
subjects. Issues that arise in pooling cross-sectional and time-series data and
statistical models for addressing these issues are considered for the case in
which there are appreciably fewer than 50 observations and a moderate number of
subjects. We discuss the possibility of using structural equation modeling to
analyze dat!
a
structures in which there are a large number (e.g., 200) of subjects, but
relatively few time points, emphasizing the different causal status of
synchronous and lagged effects and the types of models that can be specified
for longitudinal data structures. Our conclusion highlights some of the issues
raised by temporal data for statistical models, notably the important roles of
substantive theory, the question being addressed, the properties of the data,
and the assumptions underlying each technique in determining the optimal
approach to statistical analysis. [References: 112]
Williams,
L.J., & Podsakoff, P.M. (1989). Longitudinal field methods for studying
reciprocal relationships in organizational behaviour. In: Cummings, L.L.,
Straw, B.M. (Eds.) Research in
Organizational Behaviour, 11, 247-292.
Journal
available at Bizzell: HM 131 .R7 v11
_____________________________________________________________________
Zapf,
D., Dormann, C., & Frese, M. (1996). Longitudinal studies in organizational
stress research: A review of the literature with reference to methodological
issues. Journal of Occupational Health and Psychology, 1, 145-169.
NOT
AVAILABLE IN OU LIBRARIES
__________________________________________________________
_______________________________________________________________________
LATENT
GROWTH CURVE MODELING
Latent
growth curve models estimate each individual's initial status (in the form of
an intercept, and rate of growth (in the form of a slope), and they estimate
population (group) intercepts and slopes in the form of latent variables.
Repeated
measures AOV is a form of a latent growth curve model.
Curran
lists references at:
http://www.unc.edu/~curran/references.html
_______________________________________________________________________
Acock,
A.C., & Li, F. (1999). Latent growth curve analysis: A gentle introduction.
Available in pdf format at:
www.orst.edu/Dept/hdfs/acock/lgcgeneral.pdf
Li, F,
& Acock, A.C. (1999). Latent curve analysis: A manual for research data
analysts.
available in .pdf format at:
www.orst.edu/Dept/hdfs/acock/lgcmanual.pdf
This manual includes tips for getting
started with a variety of software packages that perform SEM.
Mentions Hatcher, L. (1996). Using SAS
PROC CALIS: A step-by-step introduction for beginners. Structural Equation Modeling, 3 (2),
176-192.
With
data available at:
http://www.ori.org/~fuzhongl/data/home.htm
_______________________________________________________________________
Chou,
C-P, Bentler PM, Pentz MA. (1998). Comparison of two statistical approaches to
study growth curves: The multilevel model and latent curve analysis. Structural
Equation Modeling, 5, 247-266.
_______________________________________________________________________
Duncan,
T. E., & Duncan, S. C. (1995). Modeling the processes of development via
latent variable growth curve methodology. Structural Equation Modeling, 2,
187-213.
_______________________________________________________________________
McArdle,
J.J., & Epstein, D. (1987). Latent growth curves within developmental
structural equation models. Child Development, 58, 110-133.
Abstract:
Uses structural equation modeling to combine traditional ideas from
repeated-measures analysis of variance (ANOVA) with some traditional ideas from
longitudinal factor analysis. A longitudinal model that includes correlations,
variances, and means is described as a latent growth curve model. When merged
with repeated-measures data, this technique permits the estimation of
parameters representing both individual and group dynamics. The statistical
basis of this model allows hypothesis testing of various developmental ideas,
including models of alternative dynamic functions and models of the sources of
individual differences in these functions. Aspects of these latent growth
models are illustrated with a set of longitudinal Wechsler Intelligence Scale
for Children (WISC) data from 204 children (tested just prior to 1st grade and
retested in the 3rd and 5th grades) and by using the linear structural
relationships V computer program.
DMT: relate their model to Joreskog's
(1979) autoregression model, and include mean structure to the covariance
structure so that the results are as interpretable as those of ANOVA.
_______________________________________________________________________
Meredith,
W. & Tisak, J. (1990). Latent curve analysis. Psychometrika, 55, 107-122.
_______________________________________________________________________
Muthén,
B. & Curran, P. (1997). General longitudinal modeling of individual
differences in experimental designs: a latent variable framework for analysis
and power estimation. Psychological Methods, 2, 371-402.
_______________________________________________________________________
Rogosa,
D. R., Brandt, D., & Zimowski, M. (1982).
A growth curve approach to the measurement of change. Psychological Bulletin, 92, 726-748.
_______________________________________________________________________
Suhr,
D.D. (2001). Proc GLM or Proc CALIS? Proceedings of the 26th annual SAS Users
Group International (SUGI26), Long Beach, CA, April 22-25, 2001.
Outline
differences between the two procedures and illustrates using a repeated
measures AOV approach.
She
offers a brief summary of the similarities and differences between
"traditional" techniques, like AOV and regression, and SEM. She suggests that path analysis is
appropriate to measured variables, and includes regression. She highlights SEM's flexibility, including
its ability to include measurement error, something that regression must assume
to be nonexistent.
She
describes a "default model" that proc glm tests: the model assumes to
be linear and constant, and demarcated by single unit differences in time,
which is in turn illustrated in dummy variables (T=1,2,... t).
_______________________________________________________________________
Walker,
A.J., Acock, A.C., Bowman, S.R., & Li, F. (1996). Amount of care given and
caregiving satisfaction: a latent growth curve analysis. Journal of Gerontology
(B Psychological Sciences and Social Sciences, 51(3), 130-142.
Abstract:
We
examined the wear-and-tear hypothesis using data from 4 annual interviews with
130 (128 White) middle-aged daughters caring for their physically impaired,
elderly mothers. We formulated a latent growth curve model hypothesizing that
increases in the amount of care given by daughters caused a decrease in
caregiving satisfaction, independent of caregiving duration. We found
considerable individual variability and change in both caregiving satisfaction and
the amount of care given in univariate latent growth curve analyses. Contrary
to the wear-and-tear hypothesis, a multivariate latent growth curve analysis
revealed duration of caregiving had no effect on either initial caregiving
satisfaction or change in satisfaction. An elaborated wear-and-tear model was
supported, however. The mechanism for decline in satisfaction is an increase in
the amount of care given.
COMPARING
HLM AND SEM-LGC MODELS
_______________________________________________________________________
MacCallum,
R.C., Kim, C., Malarkey, W., & Kielcolt-Glaser, J (1997). Studying
multivariate change using multilevel models and latent curve models.
multivariate Behavioral Research, 32, 215-253.
Willett,
J.B., & Sayer, A.G. (1994). Using covariance structure analysis to detect
correlates and predictors of individual change over time. Psychological
Bulletin, 116, 363-381.
Recently,
methodologists have shown how two disparate conceptual arenas - individual
growth modeling and covariance structure analysis - can be integrated. The
integration brings the flexibility of covariance analysis to bear on the
investigation of systematic interindividual differences in change and provides
another powerful data-analytic tool for answering questions about the
relationship between individual true change and potential predictors of that
change. The individual growth modeling framework uses a pair of hierarchical
statistical models to represent (a) within-person true status as a function of
time and (b) between-person differences in true change as a function of
predictors. This article explains how these models can be reformatted to
correspond, respectively, to the measurement and structural components of the
general LISREL model with mean structures and illustrates, by means of worked
example, how the new method can be applied to a sample of longitudinal panel
data.
DMT
comments: These authors explain how
LISREL parameter matrices can be constructed so that growth parameters become
estimable. Their approach takes
advantage of SEM's ability to flexibly model a variety of covariance
structures, including heteroskedastic and autocorrelated errors.
A
disadvantage of SEM for this purpose is its requirement that data be "time
structured," that is, observations must occur in the same
"waves" for each participant.
Although the waves need not be equally or systematically spaced, the
existence of "missing values," or individually distinct patterns of
timing of observations, is problematic.
Alternative approaches that forgive a lack of
"time-structuring" include hierarchical linear models (HLM).
__________________________________________________________
Rabe-Hesketh
S. Yang SY. Pickles A.
Title
Multilevel models for censored and latent responses
[Review]
Source
Statistical Methods in Medical Research.
10(6):409-427, 2001 Dec.
Abstract
Multilevel models were originally developed
to allow linear regression or ANOVA models to be applied to observations that
are not mutually independent. This lack of independence commonly arises due to
clustering of the units of observations into 'higher level units' such as
patients in hospitals. In linear mixed models, the within-cluster correlations
are modelled by including random effects in a linear model.
In this paper, we discuss generalizations of
linear mixed models suitable for responses subject to systematic and random
measurement error and interval censoring.
The first example uses data from two
cross-sectional surveys of schoolchildren to investigate risk factors for early
first experimentation with cigarettes. Here the recalled times of the
children's first cigarette are likely to be subject to both systematic and
random measurement errors as well as being interval censored. We describe
multilevel models for interval censored survival times as special cases of
generalized linear mixed models and discuss methods of estimating systematic
recall bias.
The second example is a longitudinal study
of mental health problems of patients nested in clinics. Here the outcome is
measured by multiple questionnaires allowing the measurement errors to be
modeled within a linear latent growth curve model. The resulting model is a
multilevel structural equation model. We briefly discuss such models both as
extensions of linear mixed models and as extensions of structural equation
models. Several different model structures are examined.
An important goal of the paper is to place a
number of methods that readers may have considered as being distinct within a
single overall modeling framework. [References: 66]
Reprint
available from:
Rabe-Hesketh S Univ London Kings Coll, Inst Psychiat, Dept Biostat & Comp
London SE5 8AF England
Summary:
Latent growth curve model
__________________________________________________________________________
Examples
of latent growth curve models
Roth
DL. Haley WE. Owen JE. Clay OJ. Goode KT.
Latent
growth models of the longitudinal effects of dementia caregiving: A comparison
of African American and white family caregivers
Psychology
& Aging. 16(3):427-436, 2001 Sep.
Self-report
measures of depression, physical health symptoms, and life satisfaction were
collected over a 2-year period from 197 family caregivers of dementia patients
and 218 noncaregivers (controls). Latent growth models were used to compare
changes across time for African American and White caregivers, with gender,
age. and socioeconomic status serving as covariates. Results indicated that
White caregivers sustained higher levels of elevated depression and decreasing
life satisfaction over time compared with African American caregivers. Both
groups of caregivers reported increases in physical symptoms over time. These
results indicate worsening difficulties over time for many White caregivers.
African American caregivers show more resilience on measures of depression and
life satisfaction but are still vulnerable to increases in physical symptoms
over time. Implications for additional research and clinical intervention are
discussed. [References: 45]
Reprint
available from Roth DL
Univ
Alabama, Dept Psychol, 415 Campbell
Hall, Birmingham, AL ,35294
__________________________________________________________________________
McCarty
HJ. Roth DL. Goode KT. Owen JE. Harrell L.
Donovan K. Haley WE.
Longitudinal
course of behavioral problems during Alzheimer's disease: linear versus
curvilinear patterns of decline.
Journals
of Gerontology Series A-Biological Sciences & Medical Sciences. 55(4):M200-6, 2000 Apr.
Department
of Psychology, University of Alabama at Birmingham, 35294, USA.
BACKGROUND:
Patients with Alzheimer's Disease (AD) are commonly assumed to experience a
linear decline in behavioral functioning that parallels progressive cognitive
decline. However, some researchers have suggested that specific behavioral
problems either decline at different rates or improve in late dementia.
METHODS: The present analyses examined 150 AD patients at an initial
assessment, 61 of whom were also evaluated annually on two additional
occasions. Measures of cognitive impairment and behavioral problems were
obtained. RESULTS: Cross-sectional results indicated curvilinear associations
between dementia severity and certain behavioral problems (forgetful behaviors,
and emotional and impulsive behaviors). Longitudinal analyses further indicated
trends for curvilinear rates of behavioral disturbance across time, with some
problem areas showing improvement as AD progresses through the most severe
stages. CONCLUSIONS: Even though Alzheimer's disease is a progressive dem!
entia
characterized by increasing cognitive deterioration, it appears to be
inaccurate to expect behavioral functioning to show the same linear decline
across time.
HIERARCHICAL
LINEAR MODELS
Bryk,
A.S., & Raudenbush, S.W. (1987). Application of hierarchical linear models
to assessing change. Psychological Bulletin, 101, 147-158.
Bryk,
A.S., & Raudenbush, S.W. (1988). Toward a more appropriate
conceptualization of research on school effects: A three-level hierarchical
linear model. American Journal of
Education, 97, 65-108.
Bryk,
A.S., & Raudenbush, S.W. (1992). Hierarchical linear models: Applications
and data analysis methods. Newbury Park, CA: Sage.
Raudenbush,
S.W. (2001). Comparing personal trajectories and drawing causal inferences from
longitudinal data. Annual Review of Psychology, 52, 501-25.
The first part of this article (available
online at http://psych.annualreviews.org/cgi/reprint/52/1/501.pdf)
Is nearly identical to his chapter in the
following reference (Collins and Sayer, 2001).
___________________________________________________________________________
Raudenbush,
S.W. (2001). Toward a coherent framework for comparing trajectories of individual
change. In L. Collins & A. Sayer
(Eds.), New Methods for the Analysis of Change
(pp. 35-64). Washington, DC:
American Psychological Association.
DMT:
Raudenbush distinguishes between hierarchical linear models (which include
multilevel models and random coefficient models) and SEM (which include latent
variables models and latent curve models).
He notes that SEM assumes time-structured data when they are applied to
longitudinal data. Different subjects
must be observed at the same times, and for equal numbers of times. Each subject must have the same covariance
matrix. Given this restriction on the
structure of the data (y), the researcher can choose from among a very wide
range of structural models that describe structural parameters (theta,
generically, in Raudenbush's notation).
For example, the theorized covariance structure of the errors, both in
measurement and in the structural equations, can take many forms.
HLM can
be applied to data that is not time-structured; different subjects can have
different numbers of observations, and the timing of different observations
need not be uniform among subjects.
However, the variety of models that can be applied this less structured
data is less rich than with the variety of SEM models that can be applied to
more structured data.
___________________________________________________________________________
Raudenbush
SW. Hierarchical linear models to study the effects of social context on
development. In: JM Gottman (ed.) The Analysis of Change. Mahwah, NJ: Lawrence
Erlbaum, 1995, pp. 165-
Raudenbush
points out (p. 167) that "the essential statistical problem is that when
time-series designs vary across subjects, efficient estimation of linear model
parameters requires some type of iterative procedure."
___________________________________________________________________________
Jennrich,
R.I., & Schluchter, M.D. (1986). Unbalanced repeated-measures models with
structured covariance matrices. Biometrics, 4,805-820.
"Random
coefficients model," an approach to using reported versus complete data,
as long as reported data are missing at random.
__________________________________________________________
Graham JW.
Hofer SM. Piccinin AM.
Institution
College of Health and Human Development,
Pennsylvania State University, University Park 16802-6504, USA.
Title
Analysis with missing data in drug
prevention research. [Review] [27 refs]
Source
NIDA Research Monograph. 142:13-63, 1994.
Abstract
Missing data problems have been a thorn in
the side of prevention researchers for years. Although some solutions for these
problems have been available in the statistical literature, these solutions
have not found their way into mainstream prevention research. This chapter is
meant to serve as an introduction to the systematic application of the missing
data analysis solutions presented recently by Little and Rubin (1987) and
others. The chapter does not describe a complete strategy, but it is relevant
for (1) missing data analysis with continuous (but not categorical) data, (2)
data that are reasonably normally distributed, and (3) solutions for missing
data problems for analyses related to the general linear model in particular,
analyses that use (or can use) a covariance matrix as input. The examples in
the chapter come from drug prevention research. The chapter discusses (1) the
problem of wanting to ask respondents more questions than most individuals can
answer; (2) the problem of attrition and some solutions; and (3) the problem of
special measurement procedures that are too expensive or time consuming to
obtain for all subjects. The authors end with several conclusions: Whenever
possible, researchers should use the Expectation-Maximization (EM) algorithm (or
other maximum likelihood procedure, including the multiple-group structural
equation-modeling procedure or, where appropriate, multiple imputation, for
analyses involving missing data [the chapter provides concrete examples]); If
researchers must use other analyses, they should keep in mind that these others
produce biased results and should not be relied upon for final analyses; When
data are missing, the appropriate missing data analysis procedures do not
generate something out of nothing but do make the most out of the data
available; When data are missing, researchers should work hard (especially when
planning a study) to find the cause of missingness and include the cause in the
analysis models; and Researchers should sample the cases originally missing
(whenever possible) and adjust EM algorithm parameter estimates accordingly.
[References: 27]
__________________________________________________________
Hoyle RH.
Smith GT.
Institution
Department of Psychology, University of
Kentucky, Lexington 40506-0044.
Title
Formulating clinical research hypotheses as
structural equation models: a conceptual overview. [Review] [46 refs]
Source
Journal of Consulting & Clinical
Psychology. 62(3):429-40, 1994 Jun.
Abstract
Structural equation modeling is a
comprehensive, flexible approach to research design and data analysis. Although
in recent years there has been phenomenal growth in the literature on technical
aspects of structural equation modeling, relatively little attention has been devoted
to conceiving research hypotheses as structural equation models. The aim of
this article is to provide a conceptual overview of clinical research
hypotheses that invite evaluation as structural equation models. Particular
attention is devoted to hypotheses that are not adequately evaluated using
traditional statistical models. [References: 46]
__________________________________________________________
Fisher JD.
Fisher WA.
Institution
Department of Psychology, University of
Connecticut, Storrs 06269-1020.
Title
Changing AIDS-risk behavior. [Review] [170
refs]
Source
Psychological Bulletin. 111(3):455-74, 1992 May.
Abstract
This article contains a comprehensive,
critical review of the acquired immunodeficiency syndrome (AIDS)-risk-reduction
literature on interventions that have targeted risky sexual behavior and
intravenous drug use practices. A conceptually based, highly generalizable
model for promoting and evaluating AIDS-risk behavior change in any population
of interest is then proposed. The model holds that AIDS-risk reduction is a
function of people's information about AIDS transmission and prevention, their
motivation to reduce AIDS risk, and their behavioral skills for performing the
specific acts involved in risk reduction. Supportive tests of this model, using
structural equation modeling techniques, are then reported for populations of
university students and gay male affinity group members. [References: 170]
__________________________________________________________
__________________________________________________________
Morris RJ.
Bergan JR. Fulginiti JV.
Institution
College of Education, Division of
Educational Psychology, University of Arizona, Tucson 85721.
Title
Structural equation modeling in clinical
assessment research with children. [Review] [53 refs]
Source
Journal of Consulting & Clinical
Psychology. 59(3):371-9, 1991 Jun.
Abstract
The use of structural equation modeling has
gained increased interest in recent years in the social and behavioral
sciences. This article reviews the basic tenets of structural modeling in
relation to issues in research and practice involving clinical assessment and
compares this approach with more traditional psychometric approaches to the
validation of assessment instruments with children. Arguments for and against
the inclusion of nonexperimental variables in causal studies aimed at
establishing construct validity are also discussed. An illustrative example of
the application of structural equation modeling in clinical assessment research
is provided, and a comparison is made between this approach and traditional
psychometric procedures. Implications and suggestions for the use of structural
modeling are discussed for both the practitioner and the clinical researcher.
[References: 53]
__________________________________________________________
Raykov T.
Tomer A. Nesselroade JR.
Institution
Center for Developmental and Health Research
Methodology, College of Health and Human Development, Pennsylvania State
University.
Title
Reporting structural equation modeling
results in Psychology and Aging: some proposed guidelines. [Review] [41 refs]
Source
Psychology & Aging. 6(4):499-503, 1991 Dec.
Abstract
Structural equation modeling (SEM) is now
widely used in social and behavioral science research. SEM provides the
possibility of fitting, and evaluating the fit, of well-specified, theoretical
models to empirical data--more generally, of testing elaborated psychological
theories. The options available to users of these approaches are many and
varied. Popular SEM computational software packages, such as LISREL and EQS,
provide a large amount of information, and there is some uncertainty as to what
should be routinely reported. A series of guidelines are proposed for reporting
SEM results in articles submitted to Psychology and Aging. The suggested
guidelines ask authors using SEM methodology to provide important analysis
information that will enable readers to evaluate the findings. [References: 41]
__________________________________________________________
Gritz ER.
DiMatteo MR. Hays RD.
Institution
Division of Cancer Control, Jonsson
Comprehensive Cancer Center, University of California, Los Angeles 90024.
Title
Methodological issues in adherence to cancer
control regimens. [Review] [49 refs]
Source
Preventive Medicine. 18(5):711-20, 1989 Sep.
Abstract
A six-factor model provides a heuristic
framework for understanding adherence behavior: (1) effective provider
communication; (2) rapport with provider; (3) client's beliefs and attitudes;
(4) client's social climate and norms; (5) behavioral intentions; and (6)
supports for and barriers to adherence. Four classes of methodological issues are
discussed. Recruitment may be affected by the sociodemographics of the target
population, biomedical variables, population size and location, and patient
sources utilized. Interventions can be structured to maximize enrollment,
participation and long-term retention, and adherence to the regimen promoted
with behavioral methodology. Measurement of adherence optimally includes
multiple measures at multiple time points, a well-defined focus and unit of
adherence, well-constructed response options, and multiple sources of
information. Sample size calculations and interpretation of clinical trial
results are affected by adherence rates. Mult!
ivariate
analytic techniques, such as structural equation modeling, make it possible to
specify models depicting hypothesized structural relationships between
different theoretical constructs and to evaluate the plausibility of these
models. [References: 49]
__________________________________________________________
Francis DJ.
Institution
Department of Psychology, University of
Houston, TX 77204-5341.
Title
An introduction to structural equation
models. [Review] [35 refs]
Source
Journal of Clinical & Experimental
Neuropsychology. 10(5):623-39, 1988
Oct.
Abstract
This paper provides an overview of
structural equation models, and their potential for advancing
neuropsychological theory and practice. Four topics are covered: (1) an
overview of the various classes of models, and an introduction to the
terminology and diagrams used to describe them, (2) an outline of the steps
involved in applying structural equation modeling to any research problem, (3)
an overview of the information used in assessing model fit, and a discussion of
the role of significance tests in structural models, and (4) an outline of the
advantages and disadvantages of structural equation models, and their potential
contribution to neuropsychology. The paper is intended to help researchers (1)
assess the relevance of these advanced statistical techniques to their own
research, and (2) begin the process of successful application. [References: 35]
__________________________________________________________
Hser,
Y.I., Shen, H., Chou, C.P., Messer, S.C., & Anglin, M.D. (2001). Analytic
approaches for assessing long-term treatment effects. Examples of empirical
applications and findings. Evaluation
Review, 25(2), 233-262.
DMT: The authors examine current methodological
issues that relate to studies that evaluate substance abuse intervention and
treatment programs. They discuss
several approaches, turning first to the autoregressive panel model, which they
consider a modification of structural equation modeling (p. 239), then to
latent growth curve models, which they consider both a special case of SEM (p.
239) and multilevel or hierarchical linear models (p.243). They also point types of data where survival
analysis or event history analysis is relevant. They discuss a latent transition model, which is essentially a
Markov model in which transition probabilities are calculated, and time-series
analysis.
A
disadvantage of SEM is that models are difficult to identify unless they
measure outcomes at relatively few time points. Additionally, if the autoregressive coefficients (stability)
among factors is high, collinearities (correlations) are also high,
longitudinal or cross-lagged effects are difficult to detect.
Latent
growth curve modeling offsets these disadvantages, and can be accomplished
through SEM, or through hierarchical modeling.
The disadvantage of the former is that the model is difficult to
construct, and estimates may not converge.
Hierarchical models, in contrast, are easily understood and efficient in
estimating model parameters. However,
SEM can test path models, including mediational variables, multiple change
processes, and multiple groups.
They
provide some good references for latent growth curve modeling. They make the point that no method is
superior. Often the research question
drives the choice of technique.
Alternatively, each approach gives the researcher a different view of
the potential in the data, so that we might consider how different approaches
might help one reformulate the research question.
Journal available at Bizzell: Call HM1 /
.E8
__________________________________________________________
2
studies on "adolescent problem behavior" that include clearly
longitudinal samples
1.
Ary DV.
Duncan TE. Biglan A. Metzler CW.
Noell JW. Smolkowski K.
Institution
Oregon Research Institute, Eugene 97403,
USA. dennis@ori.org
Title
Development of adolescent problem behavior.
Source
Journal of Abnormal Child Psychology. 27(2):141-50, 1999 Apr.
Abstract
The developmental model of adolescent
antisocial behavior advanced by Patterson and colleagues (e.g., Patterson,
Reid, & Dishion, 1992) appears to generalize the development of a diverse
set of problem behaviors. Structural equation modeling methods were applied to
18-month longitudinal data from 523 adolescents. The problem behavior construct
included substance use, antisocial behavior, academic failure, and risky sexual
behavior. Families with high levels of conflict were less likely to have high
levels of parent-child involvement. Such family conditions resulted in less
adequate parental monitoring of adolescent behavior, making associations with
deviant peers more likely. Poor parental monitoring and associations with
deviant peers were strong predictors of engagement in problem behavior. These
constructs accounted for 46% of the variance in problem behavior. Although
association with deviant peers was the most proximal social influence on
problem behavior, parental monitoring and family factors (conflict and
involvement) were key parenting practices that influenced this developmental
process.
Journal available at OUHSC-TULSA
AVAILABLE ONLILNE AFTER 2000
2.
Ary
DV. Duncan TE. Duncan SC.
Hops H.
Institution
Oregon Research Institute, Eugene 97403,
USA.
Title
Adolescent problem behavior: the influence
of parents and peers.
Source
Behaviour Research & Therapy. 37(3):217-30, 1999 Mar.
Abstract
This paper presents evidence that the
Patterson et al. (1992) model of development of antisocial behavior in children
generalizes to the development of a wide array of problem behaviors during
later adolescence and that youth antisocial behavior, high-risk sexual
behavior, academic failure and substance use form a single problem behavior
construct. Structural equation modeling methods were applied to 24-month
longitudinal data from 204 adolescents and parents. The model fit the data
well, accounting for 52% of the variance in adolescent problem behavior.
Specifically, families experiencing high levels of conflict were more likely to
have low levels of parent-child involvement. These family conditions were
related to poor parental monitoring and association with deviant peers one year
later. Poor parental monitoring and associations with deviant peers were strong
proximal predictors of engagement in an array of problem behaviors at two-year
follow-up.
__________________________________________________________________________
Fergusson
DM. Woodward LJ. Horwood LJ.
Institution
Christchurch Health and Development Study,
Christchurch School of Medicine, New Zealand. david.fergusson@chmeds.ac.nz
Title
Childhood peer relationship problems and
young people's involvement with deviant peers in adolescence.
Source
Journal of Abnormal Child Psychology. 27(5):357-69, 1999 Oct.
Abbreviated
Source
J Abnorm Child Psychol. 27(5):357-69, 1999 Oct.
Abstract
Using prospective longitudinal data from the
Christchurch Health and Development Study, this article examined the
relationship between children's peer relationship problems in middle childhood
and their subsequent risk of forming deviant peer affiliations in adolescence.
The analysis proceeded in three steps. First, a structural equation model
demonstrated a moderate association between early peer relationship problems
and later deviant peer affiliations (r = .27). Second, the model was extended
to include a latent variable measure of early conduct problems. This analysis
revealed that when the confounding effects of concurrently measured conduct
problems were taken into account, peer relationship problems in middle
childhood were no longer significantly related to young people's choice of
deviant peers in adolescence. Third, the model was further extended to include
lagged variables, permitting an examination of possible reciprocal
relationships between early conduct problems and peer relationship problems.
Results suggested that both early peer relationship problems and adolescent
deviant peer involvement are symptomatic of early child behavioral adjustment.
The implications of these findings for explanations of deviant peer selection
are discussed.
__________________________________________________________________________
APPLIED
INVESTIGATIONS THAT USE CROSS-LAGGED VARIABLES IN THEIR DESIGNS
One
class of longitudinal models performed with latent variables involved a
cross-lagged design. Kivimaki and
colleagues (2000) attribute this approach to longitudinal panel studies to R.D.
Shingles, but his name is not in the medical literature. I'm not sure if anyone
has used more than a pair of variables but the idea is to correlate them
themselves and with each other over time to understand how the relationship
changes and to assess the true direcion of cause and effect. Nested models can be compared to see whether
causes are in a particular direction or, more generally, bidirectional. (DMT)
Examples
(many others are available through medline, including studies of depression
among older people)
Rueter,
M.A., & Conger, R.D. (1998). Reciprocal influences between parenting and
adolescent problem-solving behavior. Developmental Psychology, 34, 1470-82.
Abstract
This investigation evaluated the hypothesis
that the development of either effective or disruptive adolescent
problem-solving behavior is reciprocally associated with the child-rearing
strategies of parents. Longitudinal data collected over 3 time points from a
large sample of families were analyzed at 1-year and 2-year measurement
intervals by using structural equation modeling. Parent and adolescent behavior
was assessed by independent observers. Reciprocal parent--adolescent
interactions occurred primarily in the presence of disruptive adolescent
behavior. Analyses involving positive adolescent behavior produced
unidirectional effects from parent behavior to adolescent behavior. Also,
reciprocal associations were most evident when the 2-year measurement interval
was used.
DMT: The study measures indicators of two
constructs (parental childrearing practices and adolescent problem-solving
behavior) three times, models the constructs generally as
"cross-lagged," then uses differences in chi-square to compare four
specific and nested models. The
baseline model posits autoregressive effects but no others. One model that is nested within the baseline
model but less restricted permits "bi-directional" lagged
effects. The other two models posit
uni-directional lagged effects in opposite directions, that is, from parent to
child, then from child to parent.
Kivimaki,
M., Feldt, T., Vahtera, J., & Nurmi, J. (2000). Sense of coherence and health:
Evidence from two cross-lagged longitudinal samples. Social Science &
Medicine, 50(4), 583-97.
Abstract
We explored the stability of sense of
coherence (SOC) and the relationship between SOC and health in two cross-lagged
longitudinal samples by using structural equation modeling. In Study 1,
comprising 577 municipal male and female employees, SOC was found to be stable
in both sexes. In women, SOC significantly predicted sickness absences in the
4-year follow-up period. A low SOC, but not a high SOC, was associated with
health prospects. Surprisingly, SOC did not influence sickness absences among
men. Study 2 further tested the relationship between SOC and health in 320 male
technical designers. Although SOC was cross-sectionally associated with
psychological and somatic health complaints, it did not predict later health
complaints in a 5-year follow-up. Thus, the present study supports the
dispositional character of SOC in both sexes and its predictive validity among
women. No support was found to the salutogenic status of SOC and an unexpected
gender difference was revealed. (85
ref)
Summary: Includes some references to methodology in
longitudinal studies.
Newsom
JT. Knapp JE. Schulz R. (1996). Longitudinal analysis of specific domains of
internal control and depressive symptoms in patients with recurrent cancer.
Health Psychology, 15(5), 323-31.
Abstract
The relation between perceptions of control
and depressive symptoms was examined in a longitudinal study of patients with
recurrent cancer. Five domains of control (self-blame, control over cancer
onset, control over symptoms, control over the course of the illness, and
overall control over life events) were found to be independent of one another.
In cross-sectional analyses, depression symptomatology was negatively
correlated with illness course control, symptom control, and overall control.
Cross-lagged longitudinal analyses using structural equation modeling suggested
only onset control and overall control were significantly associated with
depressive symptomatology over the 8-month interval. Greater baseline onset
control predicted greater follow-up depression, whereas higher baseline
depression predicted lower follow-up overall control. The importance of
developing and using domain-specific measures of control and investigating the
association of control and adjustment !
in
longitudinal analysis are discussed.
Bullers,
S., Cooper, M.L., & Russell, M. (2001). Social network drinking
and adult alcohol involvement: a longitudinal exploration of the direction of
influence. Addictive Behaviors, 26(2), 181-99.
Abstract
Past research shows consistent associations
between individuals' drinking patterns and the drinking patterns of their
social network members. This association has usually been attributed to the
influence of social networks on individual behavior. Recent studies concerning
adolescent drinking behavior suggest that such associations may be due, in
part, to selection effects in which individuals form social ties with those who
have drinking habits similar to their own. The present study used longitudinal
data and structural equation modeling to compare the selection and influence
effects among a large representative sample of adults. Results suggested that
both selection and influence affect the association between individual and
network drinking patterns among adults, but that social selection effects are
substantially stronger than social influence effects. A cross-lagged structural
equation model with a normed fit index of .975, showed that the path indicating
peer influence had a coefficient of .069 (P<.01), whereas the path
indicating network selection had a coefficient of .193 (P<.01). Comparisons
across age, race, sex, and marital status groups revealed similar results, with
stronger selection than influence effects for all groups.
Summary: well-written explication of the use of SEM
coefficients. Thoughtful writing on
nature of latent variables.
__________________________________________________________________________
Statistical
models in clinical epidemiology
Loeb M,
Walter SD, McGeer A, Simor AE, McArthur MA, Norman G (1999). A comparison of
model-building strategies for lower respiratory tract infection in long-term
care. J Clin Epidemiol 1999 Dec;52(12):1239-48
Five
strategies for creating predictive models of lower respiratory tract infection
in residents of long-term care facilities were compared. A linear judgment
model was derived by administering clinical vignettes to physicians who
indicated the risk of infection based on the presence or absence of five
predictor variables. A model based on physician consensus was created using the
same variables. Three models based on empirical data (logistic regression,
proportional hazards, and recursive partitioning) were created from a
"derivation" sample of data from a cohort study of lower respiratory
tract infections in nursing homes using the five predictor variables. All
models were applied to a validation set and compared using receiver operating
characteristic (ROC) curves. The data-derived and consensus models showed the
highest discriminative ability while the linear judgment model showed inferior
performance.
__________________________________________________________________________
Bast
J. Reitsma P.
Institution
Paedologisch Instituut-Vrije Universiteit,
Amsterdam, The Netherlands.
Title
Analyzing
the development of individual differences in terms of Matthew effects in
reading: results from a Dutch Longitudinal study.
Source
Developmental Psychology. 34(6):1373-99, 1998 Nov.
Abstract
The Matthew effect hypothesis provides a
theoretical framework to describe the development of individual differences in
reading ability. The model predicts an increase of individual differences in
reading. Reciprocal relationships between reading and other factors seem to
cause these increasing differences. This longitudinal study of 3 years was
concerned with uncovering the existence and causes of increasing individual differences
in reading in the early elementary grades. Data were analyzed within a
structural equation modeling framework. The results clearly indicate increasing
individual differences for word recognition skills. For reading comprehension,
no such effects could be established for this limited time period. More
important, some evidence for interactive relationships between reading and
other cognitive skills, behaviors, and motivational factors, hypothesized to
cause increasing differences between readers, was found.
__________________________________________________________________________
Floyd
FJ. Gilliom LA. Costigan CL.
Institution
Psychology Department, University of North
Carolina at Chapel Hill 27599-3270, USA. ffloyd@email.unc.edu
Title
Marriage and the parenting alliance:
longitudinal prediction of change in parenting perceptions and behaviors.
Source
Child Development. 69(5):1461-79, 1998 Oct.
Abstract
The study evaluates how marriage and the
parenting alliance affect parenting experiences over time. Couples (N = 79)
with school-age children who have mental retardation completed self-report and
observational measures of marriage, the parenting alliance, and parenting
attitudes and behaviors at 2 periods, 18-24 months apart. Longitudinal
structural equation modeling demonstrated significant effects of marital
quality on changes over time in self-reports of perceived parenting competence
for both the mothers and the fathers, and in observed negative mother-child
interactions. Also, in all cases, the parenting alliance mediated the effects
of marriage on parenting experiences. There was little evidence of reciprocal
causation in which parenting variables predicted change in the quality of
marriage and the parenting alliance. Interactions involving child age suggested
that teenagers as opposed to younger children were more reactive to negative
features of their parents' marital functioning and parenting alliance.
Implications are discussed regarding stable but negative marital functioning
and regarding possible differences in mothers' and fathers' parenting in the
context of marital distress.
__________________________________________________________________________
Cole DA.
Peeke L. Dolezal S. Murray N.
Canzoniero A.
Institution
Department of Psychology, University of
Notre Dame, Indiana 46556, USA. cole.1@nd.edu
Title
A longitudinal study of negative affect and
self-perceived competence in young adolescents.
Source
Journal of Personality & Social
Psychology. 77(4):851-62, 1999 Oct.
Abstract
In a 4-wave, 2-year longitudinal design, the
authors obtained measures of negative affect (NA) and self-perceived competence
from 220 boys and 216 girls who were 7th graders at the beginning of this
study. NA was operationalized as the common dimension underlying self-reports
of depressive symptoms, anxiety symptoms, and negative emotions. Self-perceived
competence consisted of 2 higher order constructs: a well-behaved/good-student
factor and an attractive/athletic/popular factor. Structural equation modeling
revealed very high stability estimates for all constructs. Nevertheless,
self-perceived competence in the attractive/athletic/popular domain predicted
changes in NA. Conversely, NA predicted changes in self-perceived competence in
the well-behaved/good-student domain. The primacy of NA versus self-cognitions
depends, in part, on the type of self-cognitions being examined.
__________________________________________________________________________
Hultsch
DF. Hertzog C. Small BJ.
Dixon RA.
Institution
Department of Psychology, University of
Victoria, British Columbia, Canada. dfh@uvic.ca
Title
Use it or lose it: engaged lifestyle as a
buffer of cognitive decline in aging?.
Source
Psychology & Aging. 14(2):245-63, 1999 Jun.
Abstract
Data from the Victoria Longitudinal Study
were used to examine the hypothesis that maintaining intellectual engagement
through participation in everyday activities buffers individuals against
cognitive decline in later life. The sample consisted of 250 middle-aged and
older adults tested 3 times over 6 years. Structural equation modeling
techniques were used to examine the relationships among changes in lifestyle
variables and an array of cognitive variables. There was a relationship between
changes in intellectually related activities and changes in cognitive
functioning. These results are consistent with the hypothesis that
intellectually engaging activities serve to buffer individuals against decline.
However, an alternative model suggested the findings were also consistent with
the hypothesis that high-ability individuals lead intellectually active lives
until cognitive decline in old age limits their activities.
____________________________________________________________________
__________________________________________________________________________
Duits
AA. Duivenvoorden HJ. Boeke S.
Taams MA. Mochtar B. Krauss XH.
Passchier J. Erdman RA.
Institution
Department of Medical Psychology and Psychotherapy,
Erasmus University Rotterdam, The Netherlands.
Title
A structural modeling analysis of anxiety
and depression in patients undergoing coronary artery bypass graft surgery: a
model generating approach.
Source
Journal of Psychosomatic Research. 46(2):187-200, 1999 Feb.
Abstract
The present study is a longitudinal study
designed to explore structural relationships between anxiety, depression,
personality, and background factors (e.g., gender, age, and complicated medical
characteristics) in patients undergoing coronary artery bypass graft (CABG)
surgery. At two timepoints before and two after CABG, 217 patients completed
self-report questionnaires. To explore structural relationships, the structural
equation modeling (SEM) method was applied. Using the model-generating
approach, a model was developed, providing a good fit. The structural
relationships revealed, in particular, the key position of neuroticism, which
was related to both pre- and postoperative anxiety and depression. Relationships
between anxiety and depression over time, both intra- and interrelationships,
were relatively weak. Relationships between anxiety and depression at the same
points in time were relatively strong, with preoperative depression leading to
preoperative anxiety, and postoperative anxiety leading to postoperative
depression. To provide a useful framework for development of intervention
strategies, further research is needed to evaluate the plausibility of the
final structural model.
__________________________________________________________________________
_________________________________________________________________
Mason
WA. Windle M.
Title
Family, religious, school and peer
influences on adolescent alcohol use: A longitudinal study
Source
Journal of Studies on Alcohol. 62(1):44-53,
2001 Jan.
Abstract
Objective: In this study, the cross-temporal
relationship between family social support and adolescent alcohol use was
examined. A primary aim was to investigate the mechanisms through which family
social support affects drinking among youth. Another aim was to examine
reciprocal relationships among the study variables. Method: Four-wave (with
6-month intervals) panel survey data collected From 840 middle adolescent boys
(n = 443) and girls (n = 397) attending a suburban school district in western
New York were analyzed using structural equation modeling with maximum
likelihood estimation, Results: Analyses revealed that family social support
was indirectly associated with decreased alcohol consumption among the
respondents, primarily through variables measuring religiosity, school grades
and peer alcohol use. In addition, adolescent alcohol use was directly
associated with subsequent increases in peer alcohol use and later decreases in
school performance. Results also showed that receiving good grades in school
predicted moderate increases in family social support. Conclusions: The
findings of this study are discussed in terms of the interrelationships that
exist among multiple socializing influences and alcohol use among adolescents.
[References: 65]
Summary: Longitudinal study
Reprint available from:
Mason WA
Univ Alabama, Dept Psychol
415 Campbell Hall,1530 3rd Ave S
Birmingham, AL 35294
USA
Univ Alabama, Dept Psychol
Birmingham, AL 35294
USA
______________________________________________________________________
Structural
models with Sullivan MD. Kempen
GIJM. Van Sonderen E. Ormel J. (2000). Models of health-related
quality of life in a population of community-dwelling Dutch elderly
Quality
of Life Research. 9(7):801-810, 2000.
Objective:
Though health-related quality of life (HRQoL) is now commonly measured as an
outcome in clinical trials, the relationships between its components remain
unclear. The relation of physical symptoms, physical function, and
psychological symptoms to each other and to overall quality of life is of
special interest. Method: Cross-sectional data from 5279 community-dwelling
elders who participated in the Groningen Longitudinal Aging Study were analyzed
using structural equation modeling techniques. Three models were examined. One
"Linear" model included: number of chronic medical conditions,
physical symptoms, physical functioning, activity interference, social
function, perceived health and overall quality of life in a simple linear
progression. Another 'non-linear' model included these variables, but allowed
effects between non-adjacent variables. A third 'non-linear' model included
these variables plus anxiety and depressive symptoms. Results: The Linear Model
did not!
satisfactorily account for the observed data
[X-2(15(df)) = 2946.96]. so the saturated Non-Linear Model, incorporating paths
between non-adjacent components. is described. When anxiety and depressive
symptoms were added to this Non-Linear Model, they fit best in a position
mediating the relation between perceived health and overall quality of life
[X-2(5(df)) = 136.78]. Conclusions: Overall quality of life appears to be
related to symptom status as directly as it is related to functional status.
Anxiety and depressive symptoms appear to mediate the relation between general
health perceptions and overall quality of life. Quality of life measures should
therefore include assessments of physical and psychological symptom severity as
well as functional status if they are to truly reflect what matters to
patients. The disability-adjusted life year (DALY) measure used by the WHO may
inadequately reflect the effect of symptoms on patient's quality of life.
[References: 29]
Summary: quality of life ; WHO; disablement
______________________________________________________________________
Carmelli
D. Kelly-Hayes M. Wolf PA.
Swan GE. Jack LM. Reed T.
Guralnik JM. (2000). The contribution of genetic influences to measures
of lower-extremity function in older male twins. Journals of Gerontology Series
A-Biological Sciences & Medical Sciences.
55(1):B49-53, 2000 Jan.
Center
for Health Sciences, SRI International, Menlo Park, California 94025, USA.
doritc@unix.sri.com
Tests
of balance, gait, and endurance were administered to 95 monozygotic (MZ) and 92
dizygotic (DZ), white male twins aged 68 to 79 years who had been born in the
United States. Within-twin-pair correlations were calculated for each
individual task and for an overall summary performance score. These were
subjected to structural equation modeling to determine the contributions of
genetic and environmental influences to individual differences in performance
scores. MZ intraclass correlations were significant and greater than DZ
correlations for the 8-foot walk and the repeated chair stands task, but not
for the standing balance task. The heritability of the lower-extremity summary
score was 57%, of which 39% was due to additive genetic effects and 18% due to
nonadditive effects. In addition, we found that genetic influences contributed
primarily to twin similarity in the poorest quartile of performance, whereas
shared environmental influences contributed to twin similarity in the best
quartile.
______________________________________________________________________
Sousa
KH. Chen F.
Institution
Assistant Professor, College of Nursing,
Arizona State University, Tempe, AZ.
Title
Health-Related quality of life theory and
structural equation modeling... 34th Annual Communicating Nursing Research
Conference/15th Annual WIN Assembly, "Health Care Challenges Beyond 2001:
Mapping the Journey for Research and Practice," held April 19-21, 2001 in
Seattle, Washington.
Source
Communicating Nursing Research, 34(9):290, 2001 Spring.
_________________________________________________________________
Marsh
HW. Yeung AS. (1998). Top-down, bottom-up, and horizontal models:
the direction of causality in multidimensional, hierarchical self-concept
models. Journal of Personality & Social Psychology. 75(2):509-27, 1998 Aug.
Faculty of Education, University of Western
Sydney at Macarthur, Campbelltown, New South Wales, Australia.
h.marsh@uws.edu.au
A new
structural equation modeling approach to questions of the direction of causal
flow between global and specific multidimensional measures of self-concept (SC)
in two 2-wave, longitudinal studies demonstrated that (a) higher order factors
were unable to explain relations among first-order factors at Time 1 (T1), at
Time 2 (T2), or between T1 and T2; (b) T1 global SC had little effect on
specific SC factors at T2 (a top-down model), but specific factors at T1 had
even less effect on T2 global SC (a bottom-up model); and (c) many specific
factors were more stable than global factors, but higher order factors were
most stable. Results provide little support for top-down, bottom-up, or
reciprocal models, instead arguing for a horizontal model in which each T2 SC
factor is primarily a function of the matching T1 SC. This casts further doubt
on the usefulness of hierarchical representations of SC.
__________________________________________________________________
Rotheram-Borus
MJ. Stein JA. Lin YY.
Institution
AIDS Institute and Department of Psychiatry,
University of California, Los Angeles, USA. rotheram@ucla.edu
Title
Impact of parent death and an intervention
on the adjustment of adolescents whose parents have HIV/AIDS.
Source
Journal of Consulting & Clinical
Psychology. 69(5):763-73, 2001 Oct.
Abbreviated
Source
J Consult Clin Psychol. 69(5):763-73, 2001 Oct.
Abstract
The impact of parental death and the
efficacy of a coping-skills intervention were examined on the adjustment of 211
adolescent children of parents with HIV/AIDS (PWH) over a 2-year period. During
the follow-up period, 35% of the PWH died. Using longitudinal structural
equation model, controlling for prior measures of adjustment at baseline, the
authors found that children of deceased PWH reported significantly more
emotional distress and problem behaviors 2 years later. Youth randomized with
their parent to a coping-skills intervention reported significantly fewer
problem behaviors and sexual partners 2 years later. Also, adolescents were
better-adjusted 2 years later when their parents had reported less emotional
distress and less severe physical health symptoms at baseline. Female
adolescents reported more emotional distress at baseline and at 2 years than
males; male adolescents reported more problem behaviors at baseline than the
females.
__________________________________________________________________
McAuley
E. Blissmer B. Marquez DX.
Jerome GJ. Kramer AF. Katula J.
Institution
University of Illinois at Urbana-Champaign,
Urbana, Illinois 61801.
Title
Social relations, physical activity, and
well-being in older adults.
Source
Preventive Medicine, 31(5):608-17, 2000 Nov. (47 ref)
Abstract
Background. A randomized controlled trial
was conducted to examine: (a) the effect of two physical activity modes on
changes in subjective well-being (SWB) over the course of a 12-month period in
older, formerly sedentary adults (N = 174, M age = 65.5 years) and (b) the role
played by physical activity participation and social support in changes in SWB
over time. Method. Participants were randomized into either an aerobic activity
group or a stretching and toning group. Structural equation modeling was
employed to conduct multiple sample latent growth curve analyses of individual
growth in measures of SWB (happiness, satisfaction with life, and loneliness)
over time. Results. A curvilinear growth pattern was revealed with well-being
significantly improving over the course of the intervention followed by
significant declines at the 6-month follow-up. Subsequent structural analyses
were conducted showing that frequency of exercise participation was a
significant predictor of improvement in satisfaction with life, whereas social
relations were related to increases in satisfaction with life and reductions in
loneliness. Improvements in social relations and exercise frequency also helped
to buffer the declines in satisfaction with life at follow-up. Conclusions. It
appears that social relations integral to the exercise environment are
significant determinants of subjective well-being in older adults. Findings are
discussed in terms of how physical activity environments might be structured to
maximize improvements in more global well-being constructs such as satisfaction
with life. Copyright 2000 American Health Foundation and Academic Press. (47 ref)
Summary: "Multiple sample latent growth curve
analysis" ; longitudinal
__________________________________________________________-
Peirce RS.
Frone MR. Russell M. Cooper ML.
Mudar P.
Institution
Research Institute on Addictions, Buffalo,
New York 14203, USA.
Title
A longitudinal model of social contact,
social support, depression, and alcohol use.
Source
Health Psychology. 19(1):28-38, 2000 Jan.
Abstract
The longitudinal relations among contact
with one's social network (social contact), perceived social support,
depression, and alcohol use were examined. An integrative model was developed
from affect regulation theory and theories of social support and dysfunctional
drinking. Data were obtained from a random sample of 1,192 adults. The 3-wave
panel model was tested using structural equation modeling analysis. Results
revealed that (a) social contact was positively related to perceived social support;
(b) perceived social support was, in turn, negatively related to depression;
and (c) depression was, in turn, positively related to alcohol use for 1 of 2
longitudinal lags. There was partial support for the feedback hypothesis that
increased alcohol use leads to decreased contact with family and friends.
Although the results generally supported the authors' hypotheses, the
significant coefficients in the model were generally small in size.
Comment: How do the covariance structures differ
between panel studies and true longitudinal studies that involve the same
subjects over time. Is it useful to
look at the relationship???????
_________________________________________________________
Sieving
RE. Perry CL. Williams CL.
Institution
Division of General Pediatrics and
Adolescent Health, University of Minnesota School of Medicine, Minneapolis, MN
55455.
Title
Do friendships change behaviors, or do
behaviors change friendships? Examining paths of influence in young
adolescents' alcohol use.
Source
Journal of Adolescent Health, 26(1):27-35, 2000 Jan. (38 ref)
Abstract
PURPOSE: This study examined support for
models of peer influence, which postulates that young adolescents whose friends
use alcohol will also engage in that behavior, and of peer selection, whereby
young adolescents seek out friends whose drinking behavior is similar to their
own. METHODS: Data for this study are from 1804 adolescents participating in
Project Northland, a school- and community-based alcohol use prevention trial.
Using latent variable structural equation modeling, a series of models examined
directions of influence between participant alcohol use and friend drug use
over three points in Grades 7, 8, and 9. RESULTS: Findings indicated that higher
levels of friends' drug use led to increased participant alcohol use. The
reverse-order relationship (i.e., greater participant involvement in alcohol
leading to more drug use among friends) was not supported by these data.
Finally, best-fitting models supported the notion that both participants'
alcohol use and the alcohol and other drug use of friends were highly stable
over time. CONCLUSIONS: Similarity in drinking behavior among adolescent
friends may be more related to processes of peer influence than to processes of
peer selection. Findings support the utility of alcohol use prevention programs
that equip younger teens with skills to resist peer influences to use
alcohol. (38 ref)
Summary:
longitudinal study
___________________________________________________________
MODEL
FIT
Bentler,
P. M. (1990). Comparative fit indexes in structural models. Psychological
Bulletin, 107, 238-246.
Bollen,
1990, Overall fit in covariance structure models: Two types of sample size
effects. Psychological Bulletin, 107, 256-259.
Gerbing,
D.W., & Anderson, J.C. (1993). Monte Carlo evaluations of goodness-of-fit
indices for structural equation models. In K.A. Bollen, & J.S. Long (eds.),
Testing structural equation models. Newbury Park, CA: Sage.
Hu, L.,
& Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance
structure analysis: Conventional criteria versus new alternatives. Structural
Equation Modeling, 6(1), 1-55.
Hu,
L.-T., & Bentler, P. (1995). Evaluating model fit. In R. H. Hoyle (Ed.),
Structural Equation Modeling. Concepts, Issues, and Applications (pp.76-99).
London: Sage.
Marsh,
H. W., Balla, J. R., & McDonald, R. P. (1988). Goodness of fit indexes in
confirmatory factor analysis: The effect of sample size. Psychological Bulletin,
103, 391-410.
Steiger,
J.H. (1989). EZPATH: A supplementary module for SYSTATand
SYGRAPH. Evanston, IL: SYSTAT.
Tanaka,
J.S. (1993). Multifaceted conceptions of fit in structural equation models. In
K.A. Bollen, & J.S. Long (eds.), Testing structural equation models.
Newbury Park, CA: Sage.
__________________________________________________________________
Evers AWM.
Kraaimaat FW. van Riel
PLCM. Bijlsma JWJ.
Title
Cognitive, behavioral and physiological reactivity
to pain as a predictor of long-term pain in rheumatoid arthritis patients
Source
Pain. 93(2):139-146, 2001 Aug.
Abstract
A heightened reactivity to pain is assumed
to play a significant role in the maintenance and exacerbation of pain in
patients: with chronic pain. In a prospective study involving 95 rheumatoid
arthritis (RA) patients, the relative contribution of self-reported cognitive,
behavioral and physiological components of pain reactivity were examined for a
change in pain within 1 year. Regression analyses indicated that self-reported
physiological reactivity predicted an increase in clinical and self-reported
pain after I year, but not cognitive and behavioral reactivity. Neither disease
activity nor neuroticism mediated or moderated the relationship of pain
reactivity to long-term pain. However, structural equation modeling revealed
that neuroticism directly affected physiological reactivity to pain, which in
turn was the only significant predictor for subsequent pain. The results of
this study underline the crucial role of physiological pain reactivity for
exacerbation of pain in RA patients and are indicative for a symptom-specific
pattern of physiological pain reactivity that is sustained by psychological
predisposition and respondent learning processes. (C) 2001 International
Association for the Study of Pain. Published by Elsevier Science B.V. All
rights reserved. [References: 71]
Summary:
longitudinal model???
Institution
Reprint available from:
Evers AWM
Univ Nijmegen, Dept Med Psychol
POB 9101 NL-6500 HB Nijmegen
Netherlands
Univ Nijmegen, Dept Med Psychol
NL-6500 HB Nijmegen Netherlands
Univ Nijmegen, Med Ctr St Radboud, Dept
Rheumatol
NL-6500 HD Nijmegen Netherlands
Univ Utrecht, Med Ctr, Dept Theumatol &
Clin Immunol
Utrecht
Netherlands
__________________________________________________________________
Labouvie
E. Pinsky I.
Title
Substance use and driving: the coexistence
of risky and safe behaviors
Source
Addiction. 96(3):473-484, 2001 Mar.
Abstract
Aims. Two risky behaviors (driving after
drinking/getting drunk, riding with drinking drivers) and two safe behaviors
(deciding not to drive under the influence of alcohol (DUI), preventing someone
else from DUI) were examined in relation to use frequency and friends' DUI to
determine whether individuals tend to engage in both types of behaviors.
Design. Self-report questionnaires were administered to a random sample of 1233
young adults in New Jersey (USA) on two occasions (mean age 21 and mean age
28). Structural equation modeling was used to assess the goodness of fit of a
hypothesized model of cross-sectional and longitudinal relationships. Findings.
Relationships between the four behaviors were strongly positive for men and
women at both occasions and were substantially accounted for by use frequency
and friends' DUI. At the later age, however, it was necessary to add
non-recursive pathways to the model, which were different for men and women. Conclusions.
Findings suggest that (1) riding with drinking drivers plays an important role
in the maintenance of the other behaviors and (2) most individuals vacillate
between risky and safe behaviors indicating that drinking contexts are best
viewed as risky decision-making situations requiring individuals to choose
between riskier and safer courses of action. [References: 51]
Summary:
SEM used to assess GOF of a hypothesized model of cross-sectional and
longitudinal relationships. Non-recursive model.
Institution
Reprint available from:
Labouvie E
Rutgers State Univ, Ctr Alcohol Studies
607 Allison Rd
Piscataway, NJ 08854 USA
Rutgers State Univ, Ctr Alcohol Studies Piscataway, NJ 08854 USA
UMDNJ, Robert Wood Johnson Med Sch, Dept
Psychiat New Brunswick, NJ 08903
__________________________________________________________
Mesman J.
Bongers IL. Koot HM.
Institution
Erasmus University Rotterdam, The
Netherlands.
Title
Preschool developmental pathways to preadolescent
internalizing and externalizing problems.
Source
Journal of Child Psychology & Psychiatry
& Allied Disciplines. 42(5):679-89,
2001 Jul.
Abstract
The present study investigated longitudinal
pathways from specific early preschool behavioral problems (ages 2-3 years) to
internalizing and externalizing problems in preadolescence (ages 10-11 years),
and the role of social problems at school entry (ages 4-5 years) in such
pathways. Path analyses were performed using both parent and teacher reports in
a sample of 251 to 346 children from the general population, depending on the
availability of parent and teacher data at each time of assessment. Structural
equation modeling revealed homotypic internalizing and externalizing pathways,
predictions from early preschool externalizing problems to later internalizing
problems, and negative predictive paths from early internalizing problems to
externalizing problems in preadolescence. Cross-informant predictions spanning
8 years were found between parent-reported aggression and overactivity at ages
2-3 years and teacher-reported externalizing problems at ages 10-11 years.
Further, results showed that boys' pathways were more complex and showed
greater predictive validity than pathways for girls, and that social problems
at school entry played a significant role in pathways to internalizing
problems, but only for boys. The results are discussed from a developmental
psychopathology perspective.
__________________________________________________________
Nyamathi
AM. Stein JA.
Institution
University of California, Los Angeles,
Office of the Dean, School of Nursing, Center for the Health Sciences, Los
Angeles, CA 90024.
Title
Assessing the impact of HIV risk reduction
counseling in impoverished African American women: a structural equations
approach.
Source
AIDS Education & Prevention, 9(3):253-73, 1997 Jun. (73 ref)
Abstract
We assessed changes in cognitive,
psychological, and risky behavior latent variables after traditional or specialized
AIDS education after 2 years using structural equation modeling (SEM) in a
sample of impoverished at-risk African American women (N = 300). Both groups
reported significant improvement at 2 years in their self-esteem and social
resources. They also reported less threat perception, avoidant coping,
emotional disturbance, HIV risk behavior, and drug use behavior. There was an
advantage to specialized group membership. When compared with the traditional
group at 2 years, women in the specialized group reported enhanced social
resources, reduced emotional distress, less use of an avoidant coping style,
and less drug use. We discuss advantages of culturally sensitive HIV risk
reduction programs and the importance of connecting women with social services in
their communities. (73 ref)
Summary: assessing changes over time.
___________________________________________________________
Gonschorek
AS. Lu IL. Halliwill JR. Beightol
LA. Taylor JA. Painter JA.
Warzel H. Eckberg DL.
Title
Influence
of respiratory motor neurone activity on human autonomic and haemodynamic
rhythms
Source
Clinical Physiology. 21(3):323-334, 2001
May.
Abstract
Although humans hold great advantages over
other species as subjects for biomedical research, they also bring major
disadvantages. One is that among the many rhythmic physiological signals that
can be recorded, there is no sure way to know which individual change precedes
another, or which change represents cause and which represents effect. In an
attempt to deal with the inherent complexity of research conducted in intact
human subjects, we developed and used a structural equation model to analyse
responses of healthy young men to pharmacological changes of arterial pressure
and graded inspiratory resistance, before and after vagomimetic atropine. Our
model yielded a good fit of the experimental data, with a system weighted R-2
of 0.77, and suggested that our treatments exerted both direct and indirect
influences on the variables we measured. Thus, infusions of nitroprusside and
phenylephrine exerted all of their direct effects by lowering and raising
arterial pressure; the changes of R-R intervals, respiratory sinus arrhythmia
and arterial pressure fluctuations that these drugs provoked, were indirect consequences
of arterial pressure changes. The only direct effect of increased inspiratory
resistance was augmentation of arterial pressure fluctuations. These results
may provide a new way to disentangle and understand responses of intact human
subjects to experimental forcings. The principal new insight we derived from
our modelling is that respiratory gating of vagal-cardiac motor neurone firing
is nearly maximal at usual levels of arterial pressure and inspiratory motor
neurone activity. [References: 41]
Summary:
assessing change over time, and causality in intact subjects
Institution
Reprint available from:
Eckberg DL
Hunter Holmes McGuire Dept Vet Affairs Med Ctr, Dept Biostat
1201 Broad Rock Blvd Richmond, VA 23249 USA
Hunter Holmes McGuire Dept Vet Affairs Med
Ctr, Dept Biostat, Richmond, VA
,23249 USA
Berufsgenossenschaftliches Unfallkrankenhaus
Hamb, Dept Neurol, Hamburg Germany
Hunter Holmes McGuire Dept Vet Affairs Med
Ctr, Dept Physiol & Med
Richmond, VA 23249 USA
Virginia Commonwealth Univ, Med Coll
Virginia, Richmond, VA 23298 USA
Mayo Clin,
Rochester, MN USA
Hebrew Rehabil Ctr Aged, Boston, MA 02131 USA
Harvard Univ, Sch Med, Boston, MA
USA
Univ Magdeburg, Dept Physiol, D-39106 Magdeburg Germany
Lusk
SL. Ronis DL. Hogan MM.
Institution
University of Michigan School of Nursing,
Room 3182, 400 N Ingalls, Ann Arbor, MI 48109-0482.
Title
Test of the health promotion model as a
causal model of construction workers' use of hearing protection.
Source
Research in Nursing & Health, 20(3):183-94, 1997 Jun. (35 ref)
Abstract
OBJECTIVE: Because individual action in
using hearing protection is a useful method to reduce noise exposure when
engineering controls cannot, the purpose of this study was to test Pender's
health promotion model (HPM) as a causal model to understand the use of hearing
protection by construction workers. This model, derived from social learning
theory, attempts to explain individuals' participation in health-promoting
behaviors and posits that cognitive-perceptual factors influence
health-promoting behavior. DESIGN: Not given. SETTING: Classroom. POPULATION:
Midwestern carpenters, operating engineers, and plumbers/pipefitters were
recruited through trade unions and trade group associations to participate in
the study. A total of 359 persons were included in theses analyses. The vast
majority were male and Caucasian, the mean age was 33 years old. INTERVENTIONS:
Written questionnaires were completed by the participants in classroom settings
as they attended journeyman and refresher courses provided by their trade
groups. Noise exposure was assessed with a series of questions about time on
the job exposed to "high noise" over the last week, month, and 3
months. Perceived control of health, the individual's perception of the meaning
of health, the individual's conception of current health and perceived
self-efficacy were all measured. Perceived barriers, the real or perceived
impediments to engaging in the behavior, was measured by the Barriers to Use of
Hearing Protection scale developed for this program of research. MAIN OUTCOME
MEASURE(S): In this test of the health promotion model, a comparison of the
results from the testing of the theoretical model and the exploratory model
proved useful. In both models, three cognitive-perceptual factors, benefits
(value of use), barriers, and self-efficacy, had significant direct paths to
use, with the expected signs. The HPM's ability to predict behavior has been
supported. However, further tests of the HPM are needed, especially as not all
of the components of the HPM have been fully tested. In the current study,
interpersonal influences, which had not been tested with factory workers, were
found to be strong predictors of hearing protection use. RESULTS/CONCLUSIONS:
Though the HPM is a causal model, structural equation modeling with
cross-sectional data provides no evidence about the directions of causality
among variables. Some alternative models, including a model with all causal
paths reversed, would fit the data equally well. Longitudinal studies and tests
of theory-based interventions, such as the one we are currently conducting, are
needed to provide evidence of the directions and causal nature of these paths.
[CINAHL abstract] (35 ref)
Summary:
Abstract includes discussion of causal implications of cross-sectional data
when used with SEM
_______________________________________________
Vollebergh
WAM. Iedema J. Bijl RV.
de Graaf R. Smit F. Ormel J.
Title
The structure and stability of common mental
disorders - The NEMESIS Study
Source
Archives of General Psychiatry.
58(6):597-603, 2001 Jun.
Abstract
Background: We analyzed the underlying
latent structure of 12-month DSM-III-R diagnoses of 9 common disorders for the
general population in the Netherlands. In addition, we sought to establish (1)
the stability of the latent structure underlying mental disorders across a
1-year period (structural stability) and (2) the stability of individual
differences in mental disorders at the level of the latent dimensions
(differential stability).
Methods: Data were obtained from the first
and second measurement of the Netherlands Mental Health Survey and Incidence
Study (NEMESIS) (response rate at baseline: 69.7%, n=7076; 1 year later, 79.4%,
n=5618). Nine common DSM-III-R diagnoses were assessed twice with the Composite
International Diagnostic Interview with a time lapse of 1 year. Using
structural equation modeling, the number of latent dimensions underlying these
diagnoses was determined, and the structural and differential stability were
assessed.
Results: A 3-dimensional model was
established as having the best fit: a first dimension underlying substance use
disorders (alcohol dependence, drug dependence); a second dimension for mood
disorders (major depression, dysthymia), including generalized anxiety
disorder; and a third dimension underlying anxiety disorders (simple phobia,
social phobia, agoraphobia, and panic disorder). The structural stability of
this model during a 1-year period was substantial, and the differential
stability of the 3 latent dimensions was considerable.
Conclusions: Our results confirm the
3-dimensional model for 12-month prevalence of mental disorders. Results
underline the argument for focusing on core psychopathological processes rather
than on their manifestation as distinguished disorders in future population
studies on common mental disorders. [References: 35]
Summary: longitudinal?? Assesses stability of latent
constructs over time.
Institution
Reprint available from:
Vollebergh WAM
Netherlands Inst Mental Hlth & Addict,
Trimbos Inst
POB 725,
NL-3500 AS Utrecht, Netherlands
Netherlands Inst Mental Hlth & Addict,
Trimbos Inst
NL-3500 AS Utrecht, Netherlands
Social & Cultural Planning Off
The Hague,
Netherlands
Univ Groningen, Dept Social Psychiat
Groningen,
Netherlands
_______________________________________________
Harford
TC. Muthen BO.
Title
The dimensionality of alcohol abuse and
dependence: A multivariate analysis of DSM-IV symptom items in the National
Longitudinal Survey of Youth
Journal
of Studies on Alcohol. 62(2):150-157, 2001 Mar.
Abstract
Objective: This article examines the factor
structure of 22 symptom items used to configure the criteria of DSM-IV
(Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition) alcohol
abuse and dependence and relates the factor structure to background
characteristics.
Method: Data for this study were drawn from
the National Longitudinal Study of Labor Market Experience in Youth (NLSY). The
symptom items were related to the covariates using the statistical technique of
structural equation modeling generalized to dichotomous outcomes. The present
model is a special case of structural equation modeling, a multiple causes and
multiple indicators (MIMIC) model, in which one or more latent variables (i.e.,
alcohol abuse and dependence) intervene between a set of observed background
variables predicting a set of observed response variables (i.e., DSM-IV symptom
items).
Results: The results of the structural
equation analysis provide further support for two dimensions underlying the
DSM-IV symptom items. Although the two-factor dimension bore a strong
resemblance to the DSM-IV conceptions of abuse and dependence, there were
notable differences in the item content of the symptom items for each
dimension. The dependence dimension drew upon items related to the abuse
criteria for continued drinking despite social problems and recurrent drinking
resulting in failure to fulfill role obligations. The abuse dimension drew upon
items related to the abuse criterion for hazardous drinking and the dependence
criterion for larger amounts over time. The two factors were shown to have
different relationships to the background variables. Alcohol dependence tvas
related to family history of alcoholism and educational status. Age was not
related to dependence and inversely related to alcohol abuse.
Conclusions: Findings from this study
replicate the two-dimensional model for DSM-IV criteria found in other studies
and provide further support for the validity of alcohol dependence in general
population samples. A major implication of the factor structure in the present
study relates to the different classification of cases that would otherwise be
obtained with DSM-IV criteria. These departures were shown to affect abuse,
which retained only 40% of DSM-IV diagnoses, more strongly than dependence,
which retained 91% of DSM-IV diagnoses. [References: 31]
Publication
Type
Article
Institution
Reprint available from:
Harford TC
Univ Calif Los Angeles, Grad Sch Educ &
Informat Studies
Los Angeles, CA 90024
USA
Boston Univ, Sch Publ Hlth, Dept Social
& Behav Sci
Boston, MA 02215
USA
______________________________________________________________________________
Harford
TC. Muthen BO.
Institution
Social and Behavioral Sciences Department,
Boston University School of Public Health, Massachusetts, USA.
Title
Adolescent and young adult antisocial
behavior and adult alcohol use disorders: a fourteen-year prospective follow-up
in a national survey.
Source
Journal of Studies on Alcohol. 61(4):524-8, 2000 Jul.
Abbreviated
Source
J Stud Alcohol. 61(4):524-8, 2000 Jul.
Abstract
OBJECTIVE: Data from the National
Longitudinal Survey of Youth (NLSY) are used to examine the association between
antisocial behaviors (ASB) reported in youth (15-22 years old) and alcohol use
disorders (AUD) 14 years later in a large (N = 7,326) representative national
sample. METHOD: Structural equation modeling generalized to dichotomous
outcomes was used to assess the associations between latent variables of ASB
with latent variables of AUD and background variables. RESULTS: Exploratory
factor analysis of 17 ASB items yielded three factors having clear
interpretations with the literature-property offenses, person offenses and
illicit substance involvement. When examined in the context of the multivariate
structural equation model, several independent associations between ASB and AUD
symptoms and covariates were found. Although there were significant and
independent effects for each ASB factor on each of the alcohol use disorder
factors, the strength of the association was strongest for the effects of early
illicit substance involvement on alcohol abuse and dependence. CONCLUSIONS:
Both illicit substance involvement and delinquency other than illicit substance
involvement reported in 1980 were associated with alcohol use disorders 14
years later.
______________________________________________________________________________
O'Boyle CA.
Henly SJ. Larson E.
Institution
Division of Disease Prevention and Control,
Minnesota Department of Health, USA.
Title
Understanding adherence to hand hygiene
recommendations: the theory of planned behavior.
Source
American Journal of Infection Control. 29(6):352-60, 2001 Dec.
Abstract
BACKGROUND: Most health care workers (HCWs)
are aware of the rationale for hand hygiene procedures, yet failure to adhere
to guidelines is common. Little is known about factors that motivate HCWs to
practice hand hygiene. PURPOSE: The purposes of this study were to (1) estimate
adherence to hand hygiene recommendations; (2) describe relationships among
motivational factors, adherence, and intensity of nursing unit activity; and
(3) test an explanatory model for adherence to hand hygiene guidelines based on
the theory of planned behavior (TPB). METHOD: A longitudinal, observational
design was used to collect data from 120 registered nurses employed in critical
care and postcritical care units. Nurses provided information about
motivational factors and intentions and a self-report of the proportion of time
they followed guidelines. At least 2 weeks later, the nurses' hand hygiene
performance was observed while they provided patient care. Structural equation
modeling was used to test the TPB-based model. RESULTS: Rate of adherence to
recommendations for 1248 hand hygiene indications was 70%. The correlation
between self-reported and observed adherence to handwashing recommendations was
low (r = 0.21). TPB variables predicted intention to handwash, and intention
was related to self-reported hand hygiene. Intensity of activity in the nursing
unit, rather than TPB variables, predicted observed adherence to hand hygiene
recommendations. CONCLUSIONS: The limited association between self-reported and
observed hand hygiene scores remains an enigma to be explained. Actual hand
hygiene behavior may be more sensitive to the intensity of work activity in the
clinical setting than to internal motivational factors.
______________________________________________________________________________
Lonczak HS.
Huang B. Catalano RF. Hawkins JD.
Hill KG. Abbott RD. Ryan JAM.
Kosterman R.
Title
The social predictors of adolescent alcohol
misuse: A test of the social development model
Source
Journal of Studies on Alcohol.
62(2):179-189, 2001 Mar.
Abbreviated
Source
J. Stud. Alcohol. 62(2):179-189, 2001 Mar.
Abstract
Objective: This study was conducted to
investigate the ability of the social development model (SDM) to predict
alcohol misuse at age 16 and to investigate the ability of the SDM to mediate
the effects of alcohol use at age 14 on alcohol misuse at age 16.
Method: The sample of 807 (411 males) is
from the longitudinal panel of the Seattle Social Development Project which, in
1985, surveyed all consenting fifth-grade students from 18 elementary schools
serving high-crime neighborhoods in Seattle, Washington. Alcohol use was
measured at age 14, predictors of alcohol misuse were measured at age 15 and
alcohol misuse was measured at age 16. Structural equation modeling was used to
examine the fit of the model to the data.
Results: All factor loadings were highly
significant and the measurement model achieved a good fit with the data
(Comparative Fit Index [CFI] = 0.93). A sec end-order structural model fit the
data well (CFI = 0.91) and also explained 45% of the variance in alcohol misuse
at age 16. The SDM partially and significantly mediated the direct effect of
age-14 alcohol use on age-16 alcohol misuse.
Conclusions: The risk and protective
processes specified by the SDM serve as potential targets for the prevention or
reduction of adolescent alcohol misuse. [References: 42]
Institution
Reprint available from:
Catalano RF
Univ Washington, Sch Social Work, Social Dev
Res Grp, Dept Educ Psychol
9725 3rd Ave NE,401
Seattle, WA 98115
USA
Univ Washington, Sch Social Work, Social Dev
Res Grp, Dept Educ Psychol
Seattle, WA 98115
USA
______________________________________________-
Gross
D. Conrad B. Fogg L. Wothke W.
Institution
Rush Univ Coll Nurs, 1743 W Harrison St Rm
301 SSH, Chicago IL 60612-3824.
Title
A longitudinal model of maternal
self-efficacy, depression, and difficult temperament during toddlerhood.
Source
Research in Nursing & Health, 17(3):207-15, 1994 Jun. (34 ref)
Abstract
The purpose of this study was to test a
model of maternal self-efficacy during toddlerhood using a longitudinal
sequential design. Participants were 126 mothers of 1-year olds (Cohort 1) and
126 mothers of 2-year olds (Cohort 2) who completed questionnaires measuring
maternal self-efficacy, depression, and perceived difficult toddler temperament
three times over 1 year. Data were analyzed using structural equation modeling
and maximum likelihood estimation. Findings support a model whereby (a) the
more depressed the mother feels, the more likely she is to rate her toddler's
temperament as difficult, (b) the more difficult the child's temperament is
perceived to be, the lower the mother's estimates of her parenting
self-efficacy, (c) the lower the mother's self-efficacy, the greater her
depression, and (d) the more depressed the mother feels at one point in time,
the more likely she is to remain depressed 6 months later. Implications of the
findings are discussed as they relate to self-efficacy theory and nursing
intervention with parents of difficult toddlers. (34 ref)
____________________________________________________________
Galaif
ER. Stein JA. Newcomb MD. Bernstein DP.
Title
Gender differences in the prediction of
problem alcohol use in adulthood: Exploring the influence of family factors and
childhood maltreatment
Source
Journal of Studies on Alcohol.
62(4):486-493, 2001 Jul.
Abstract
Objective: The purpose of this study was to
contrast men and women in prospective relationships among family-oriented and
alcohol-related variables obtained during adolescence, childhood physical,
emotional and sexual abuse collected retrospectively, and later adult problem
alcohol use. Method: In structural equation models, early family processes
(support/bonding, parent drug-use problems, parental divorce and childhood
maltreatment) and prior alcohol use simultaneously predicted adult problem
alcohol use at two later time points in a longitudinal community sample of 426
(305 female) adults. Results: Significant relationships were found among family
processes, childhood maltreatment and problem alcohol use within time and
longitudinally for both men and women. Greater family support/bonding during
adolescence predicted less problem alcohol use in adulthood. Men and women who
experienced sexual abuse as a child reported more problem alcohol use in
adulthood. Problem alcohol use was stable across time. Men reported more
problem alcohol use in adolescence and adulthood, and women reported more early
sexual abuse. These results contradict those that find no significant
relationships between childhood abuse and subsequent alcohol-related problems.
Parental drug use problems during the participant's adolescence did not
directly predict problem alcohol use in adulthood. The relationship was more
indirect in that parental drug use was associated with family-related
concomitants that in turn were significant predictors of more problem alcohol use
in adulthood. Conclusions: The strong stability for problem alcohol use across
the three time periods is a signal that alcohol use in adolescence should not
be ignored; furthermore, family dynamics need attention in addressing problem
alcohol use. [References: 27]
Summary:
longitudinal study
Institution
Reprint available from:
Galaif ER
Univ Calif Los Angeles, Dept Psychol
Los Angeles, CA USA
Univ Calif Los Angeles, Drug Abuse Res
Ctr Los Angeles, CA 90095 USA
_______________________________________________________________________
Maia
JAR. Lefevre J. Claessens A. Renson R. Vanreusel
B. Beunen G.
Institution
Faculty of Sports Sciences and Physical
Education, University of Porto, Porto, Portugal. E-mail: jmaia@fcdef.up.pt.
Title
Tracking of physical fitness during
adolescence: a panel study in boys.
Source
Medicine & Science in Sports &
Exercise, 33(5):765-71, 2001 May. (51 ref)
Abstract
Purpose: To investigate the tracking in
physical fitness (PF) viewed as a whole, a multidimensional trait of the
subject, and to establish the stability of each factor of PF in adolescence
from the perspective of a panel study using the structural equation modeling
approach. Methods: From a sample of 454 boys followed from 12 to 18 yr of age
of the Leuven Growth Study, we considered only three consecutive measurement
occasions with a mean age of 12.76, 14.69, and 17.73 yr. Physical fitness was
evaluated by means of a battery composed of the following tests: plate tapping,
sit and reach, vertical jump, arm pull, leg lifts, bent arm hang, and shuttle
run. Structural equation models were fitted to the data, namely autoregressive
models with latent variables. These models were used to quantify the tracking
of PF as a whole and also of the individual marker variables of fitness.
Results: Stability estimates of PF as a whole are rather high, beta21 = 0.86
and beta32 = 0.68, with an explained variance of 74% and 73%, respectively.
Tracking coefficients represented by disattenuated autocorrelations among the
fitness factor gave high results: r1,2 = 0.86; r1,3 = 0.78; and r2,3 = 0.85.
Conclusions: Physical fitness as a whole is highly stable in adolescent years
and very predictable from early years. The same is observed for each factor of
fitness. Moreover, autoregressive models within the context of structural
equation modeling are better suited than simple Pearson or Spearman
autocorrelations to study the tracking problem of PF. (51 ref)
Summary: Longitudinal panel study with autoregressive
correlation matrix; three measurements over time.
__________________________________________________
Zsembik BA.
Peek MK.
Institution
Associate Professor, Department of
Sociology, University of Florida, PO Box 117330, Gainesville, FL 32611-7330;
zsembik@soc.ufl.edu.
Title
Race differences in cognitive functioning
among older adults.
Source
Journals of Gerontology Series
B-Psychological Sciences & Social Sciences, 56B(5):S266-74, 2001 Sep.
(59 ref)
Abstract
OBJECTIVES: Explaining race differences in
cognitive functioning in later life continues to challenge researchers. This
study was an attempt to incorporate the clinical literature, emphasizing
biological correlates of cognitive functioning, and the social research
literature, emphasizing social inequalities and consequent health outcomes, in
the examination of sources of race differences in cognitive functioning in
older adults. METHODS: With data from Wave 1 of the Assets and Health Dynamics
of the Oldest Old survey, the authors used structural equation models (LISREL
8.30) to estimate the direct effects of race on cognitive functioning and
indirect effects through social and biological risk factors for the total
sample (N = 5,955). RESULTS: Race had a direct association with cognitive
functioning. Race also had indirect effects on cognitive functioning through
social risk factors-education and health insurance. There did not appear to be
indirect effects of race through bi!
ological
risk factors. DISCUSSION: The direct and indirect effects of race through
social risk factors attest to the importance of examining different ways
through which race can influence cognitive functioning of older adults. This
research also emphasizes the need for researchers to investigate more closely
race differences in dimensions of cognitive functioning and cognitive
functioning over time. (59 ref)
Summary: Used LISREL to clarify direct and indirect
effects
__________________________________________________
Peek
MK. Lin N. (1999). Age differences in the effects of network
composition on psychological distress.
Social Science & Medicine, 49(5):621-36, 1999 Sep. (52 ref)
Dept of Health Promotion and Gerontology,
University of Texas Medical Branch 301 University Blvd, Galveston, TX
77555-1028; mkpeek@utmb.edu.
The
main goal of this research is to better understand age differences in the
effects of social networks on mental health. Using a social network approach to
revise the convoy of social support model (Kahn and Antonucci, 1980), we
examine specifically how two aspects of social support networks (kin
composition and convoy dimensions) influence psychological distress for older
and younger samples (18-59 and 60+). We hypothesize that kin composition will
influence distress in general but especially for the younger sample, while two
competing hypotheses for the convoy of social support model are tested. Using
data from a three-wave panel health study, structural equation models (LISREL
8.20) indicate that a greater proportion of kin in the perceived support
network and the presence of family members in the inner circle of the convoy
significantly reduce distress, primarily for the younger sample. Implications
for the convoy model are discussed.
(52 ref)
___________________________________________
Peek,
Coward, Peek
J
Gerontol B Psychol Science Soc Sci,
1998, 53B: S127-136.
Used SE
model
__________________________________________________
Authors
Kalichman SC. Rompa D. DiFonzo K. Simpson D.
Austin J. Luke W. Kyomugisha F. Buckles J.
Institution
Center for AIDS Intervention Research
(CAIR), Medical College of Wisconsin.
Title
HIV treatment adherence in women living with
HIV/AIDS: research based on the Information-Motivation-Behavioral Skills model
of health behavior.
Source
Journal of the Association of Nurses in AIDS
Care, 12(4):58-67, 2001 Jul-Aug. (24 ref)
Abstract
Close and consistent adherence to anti-HIV
medication regimens is necessary to achieve the maximum benefit of these
potentially effective treatments. The authors examined cognitive and behavioral
factors associated with HIV treatment adherence in a convenience sample of 112
women, 72 of whom were currently taking HIV treatments at the time of the
study. Women completed confidential surveys and interviews to assess
HIV-related health status, treatment regimens, and cognitive behavioral
characteristics derived from the Information-Motivation-Behavioral Skills model
of health promotion behaviors. Results showed that women who had missed at
least one dose of their HIV medications in the past week reported lower
intentions (motivation) to remain adherent and lower adherence self-efficacy
(skills). Structural equation modeling showed that motivational and
skills-building factors significantly predicted the number of medication doses
missed. However, treatment-related information did not predict treatment
adherence. In addition, women who had missed a dose of medication in the past
week were more likely to have ever used devices and strategies to remind them
of doses, but were no more likely to currently use such strategies.
Interventions that enhance treatment adherence motivation and build adherence
skills may help improve HIV treatment adherence in women receiving anti-HIV therapies. (24 ref)
Summary: May make use of parameter estimates
______________________________________________________
Resnick B.
Zimmerman S. Orwig D. Furstenberg A. Magaziner J.
Institution
University of Maryland School of Nursing,
Room 375, 655 West Lombard Street, Baltimore, MD 21201
(bresnick@umaryland.edu).
Title
Model testing for reliability and validity
of the Outcome Expectations for Exercise Scale.
Source
Nursing Research, 50(5):293-9, 2001 Sep-Oct.
(40 ref)
Abstract
BACKGROUND: Development of a reliable and
valid measure of outcome expectations for exercise appropriate for older adults
will help establish the relationship between outcome expectations and exercise.
Once established, this measure can be used to facilitate the development of
interventions to strengthen outcome expectations and improve adherence to
regular exercise in older adults. OBJECTIVES: Building on initial psychometrics
of the Outcome Expectation for Exercise (OEE) Scale, the purpose of the current
study was to use structural equation modeling to provide additional support for
the reliability and validity of this measure. METHODS: The OEE scale is a
9-item measure specifically focusing on the perceived consequences of exercise
for older adults. The OEE scale was given to 191 residents in a continuing care
retirement community. The mean age of the participants was 85 6.1 and the majority were female (76%),
White (99%), and unmarried (76%). Using structural equation modeling,
reliability was based on R2 values, and validity was based on a confirmatory
factor analysis and path coefficients. RESULTS: There was continued evidence
for reliability of the OEE based on R2 values ranging from .42 to .77, and
validity with path coefficients ranging from .69 to .87, and evidence of model
fit (X2 of 69, df = 27, p < .05, NFI = .98, RMSEA = .07). CONCLUSION: The
evidence of reliability and validity of this measure has important implications
for clinical work and research. The OEE scale can be used to identify older
adults who have low outcome expectations for exercise, and interventions can
then be implemented to strengthen these expectations and thereby improve
exercise behavior. (40 ref)
Summary: reliability and validity, using confirmatory
factor analysis and path coefficients.
_________________________________________-
Logsdon MC.
Usui W.
Institution
School of Nursing, University of Louisville.
Title
Psychosocial predictors of postpartum
depression in diverse groups of women.
Source
Western Journal of Nursing Research, 23(6):563-74, 2001 Oct. (40 ref)
Abstract
The purpose of this study was to test the
extent to which a causal model developed from a theoretical formulation of
postpartum depression was consistent with data collected from three groups of
postpartum women. In this cross-sectional, correlational design, the samples
consisted of primarily middle-class, Caucasian mothers of term infants and
preterm infants, and low-income, African American mothers of term infants. Instruments
included the CES-D Depression instrument, the Postpartum Support Questionnaire,
Rosenberg's Self-Esteem instrument, and a question regarding closeness to
partner. The causal model was tested with structural equation modeling.
Importance of support, support received, and closeness to partner were
significant predictors of both self-esteem and depression. Predictors of
postpartum depression are the same across diverse samples of women, as proposed
in the causal model. (40 ref)
Summary: Validation of a causal model
_______________________________________________________________
Authors
Martin M.
Grünendahl M. Martin P.
Institution
German Centre for Research on Ageing,
Bergheimer Strasse 20, 69115 Heidelberg, Germany. E-mail: mmartin@dzfa.uni-heidelberg.de.
Title
Age difference in stress, social resources,
and well-being in middle and older age.
Source
Journals of Gerontology Series
B-Psychological Sciences & Social Sciences, 56B(4):P214-22, 2001 Jul.
(46 ref)
Abstract
The present study examines the
interrelationships among the constructs of social resources, stress, and
well-being in middle-aged and older adults. Two samples of 489 middle-aged
adults (41-43 years) and 449 older adults (61-63 years) from the Interdisciplinary
Longitudinal Study of Adult Development were compared with respect to the
availability of social resources, levels of stress, and well-being. The data
were used to construct separate structural equation models explaining the
influence of stress and social resources on well-being in the two groups. The
results indicate higher levels of health-related stress and similar levels of
social resources in the older group. Structural equation models and examination
of total and indirect effects confirmed that a stress-suppression model has the
best fit in explaining the interrelationships between stress, social resources,
and well-being. There was a strong effect of stress on well-being, no direct
effect of social resources on well-being, and a mediating effect of stress on
well-being. The relative contributions of stress and resources to well-being
were comparable between age groups.
(46 ref)
Summary: Longitudinal model???
_______________________________________________________________
Zhou Q.
O'Brien B. Soeken K.
Institution
Research Clinical Specialist, Inova Fairfax
Hospital, Women's Service, 3300 Gallows Road, Falls Church, VA 22042
(qiuping@zhou@inova.com).
Title
Rhodes Index of Nausea and Vomiting -- Form
2 in pregnant women: a confirmatory factor analysis.
Source
Nursing Research, 50(4):251-7, 2001 Jul-Aug.
(24 ref)
Abstract
BACKGROUND: Despite widespread application
of Rhodes Index of Nausea and Vomiting-Form 2 (INV2) in practice and research,
empirical analyses have not been consistently performed to verify the a priori
factors that guided the subclass construction of the symptoms. OBJECTIVES: To
examine the dimensional structure of Rhodes INV in a sample of pregnant women.
METHOD: Data were collected from 152 pregnant women who were experiencing some
degree of nausea and vomiting during early pregnancy and analyzed using
structural equation modeling techniques. Five competing measurement structures
were tested and compared. The structure (model) that provided the closest fit
to the data was selected and relationships (factor loadings) between the
constructs and indicators were established. RESULTS: The model fitting the data
the closest was a three-factor structure measuring nausea, vomiting, and
retching as three separate, but correlated dimensions. The factor loadings were
high (0.73-0.96) and significant (p < .001). The model treating nausea and
vomiting as a one-factor concept as well as the model including two factors
named symptom occurrence and symptom distress did not fit the data. CONCLUSION:
Rhodes INV2 is a valid measurement tool if subscales are formed to reflect the
multidimensional structure of nausea and vomiting in pregnancy. (24 ref)
Summary: comparison of several structural models
____________________________________________________________
Weiss
MR. Kimmel LA. Smith AL.
Institution
Kinesiology Program, University of Virginia,
Charlottesville, VA 22903.
Title
Determinants of sport commitment among
junior tennis players: enjoyment as a mediating variable.
Source
Pediatric Exercise Science, 13(2):131-44, 2001 May. (29 ref)
Abstract
This study examined determinants of junior
tennis players' motivation to continue involvement using the sport commitment
model as the theoretical framework (20). Based on the sport enjoyment
literature, a version of the original sport commitment model (i.e., all
determinants directly predict commitment) and a revised model where enjoyment
was a mediator of the relationships between determinants and level of
commitment were tested. Tennis players (N= 198; ages 10-18 years) completed
self-report questionnaires on the constructs of interest. Hypothesized
relationships among variables were tested using structural equation modeling.
Results provided support for both the original and mediational models, with
enjoyment exerting the strongest effect on tennis commitment in both models. An
alternative model was tested where both direct and indirect effects through
enjoyment on commitment were specified. The alternative model was accepted as
most theoretically appealing because determinants of commitment and sources and
consequences of sport enjoyment were accounted for within the larger conceptual
model. (29 ref)
_______________________________________________________________
Svedberg P.
Lichtenstein P. Pedersen NL.
Institution
Department of Medical Epidemiology,
Karolinska Institutet, Box 281, SE-171 77 Stockholm, Sweden. E-mail:
Pia.Svedberg@mep.ki.se.
Title
Age and sex differences in genetic and
environmental factors for self-rated health: a twin study.
Source
Journals of Gerontology Series
B-Psychological Sciences & Social Sciences, 56B(3):S171-8, 2001 May.
(36 ref)
Abstract
OBJECTIVES: Self-rated health has been shown
to be a predictor for future health status and mortality. The purpose of this
study was to investigate age-group and sex differences in genetic and
environmental sources of variation for self-rated health. METHODS: A sample of
twins from the Swedish Twin Registry participated in a computer-assisted
telephone interview with assessment of self-rated health. Structural equation
model analyses on 1,243 complete twin pairs provided estimates of genetic and
environmental components of variance. RESULTS: Individual differences primarily
reflected individual specific environmental influences at all ages. The
increase in total variance across age groups was primarily due to genetic
influences in the age groups 45--74 years and greater environmental influences
in the oldest age group (>74). No significant sex differences were found in
variance components. DISCUSSION: Genetic variance in the two middle age groups
(45--74) could reflect genetic susceptibility to age-dependent illnesses not
yet expressed in the youngest group. The findings suggest that it might be more
fruitful to explore the origins of individual differences for self-rated health
in the context of an individual's age and birth cohort rather than in the
context of sex. (36 ref)
_______________________________________________________________
Gutierrez PM. Rodriguez PJ. Garcia P.
Institution
Northern Illinois University, Dept of
Psychology, 1425 West Lincoln Highway, DeKalb, IL 60115-2892.
Title
Suicide risk factors for young adults:
testing a model across ethnicities.
Source
Death Studies, 25(4):319-40, 2001 Jun.
(47 ref)
Abstract
A general path model based on existing
suicide risk research was developed to test, factors contributing to current
suicidal ideation in young adults. A sample of 67.3 undergraduate students
completed a packet of questionnaires containing the Beck Depression Inventory,
Adult Suicidal Ideation Questionnaire, and Multi Attitude Suicide Tendency
Scale. They also provided information on history of suicidality and exposure to
attempted and completed suicide in others. Structural equation modeling was
used to test the fit of the data to the hypothesized model. Goodness-of fit
indices were adequate and supported the interactive effects of exposure,
repulsion by life, depression, and history of self-harm on current ideation.
Model, fit for three subgroups based on race/ethnicity (i.e., White, Black, and
Hispanic) determined that repulsion by life and depression function differently
across groups. Implications of these findings for current methods of suicide
risk assessment and future research are discussed in the context of the
importance of culture. (47 ref)
_______________________________________________________________
Mahon NE.
Yarcheski A.
Institution
Professor, College of Nursing, Rutgers, The
State University of New Jersey.
Title
Outcomes of depression in early adolescents.
Source
Western Journal of Nursing Research, 23(4):360-75, 2001 Jun. (44 ref)
Abstract
The purpose of this study was to examine the
relationship between depressed mood and depressive symptomatology and the
influence of both variables on perceived social support, interpersonal
conflict, general well-being, and perceived health status in 144 early
adolescents. The subjects responded to instruments measuring the study
variables in classroom settings. Two bivariate regression structural equation
models were examined using the LISREL 7 computer software program. In the
health model, depressed mood had a direct positive effect on depressive
symptoms and a direct negative effect on well-being and on perceived health
status. Depressive symptoms had a direct negative effect on well-being and on
perceived health status. In the interpersonal model, depressed mood had a
direct positive effect on depressive symptoms and on conflict and a direct
negative effect on social support. Depressive symptoms had a direct positive
effect on conflict and a direct negative effect on social support. (44 ref)
_______________________________________________________________
Yarcheski A. Mahon NE.
Institution
Professor, College of Nursing, Rutgers
University.
Title
A causal model of depression in early
adolescents.
Source
Western Journal of Nursing Research, 22(8):879-94, 2000 Dec. (56 ref)
Abstract
The purpose of this study was to test the
extent to which a causal model developed from a theoretical formulation of
depression was consistent with data obtained from early adolescents, age 12 to
14. In this cross-sectional correlational design, the final sample consisted of
225 adolescents who responded to a demographic data sheet and instruments
measuring depression, self-esteem, state anxiety, and perceived stress in
classrooms. The causal model was tested via the LISREL 7 program, using a
maximum likelihood structural equation model. The results yielded a chi-square
(l, N = 225) = .71, p =.401, indicating a good fit of the model to the data.
Perceived stress had the strongest direct, indirect, and total effect on
depression in early adolescents. Contrary to expectation, self-esteem did not
have a direct effect on depression, and girls did not report higher levels of
depression than did boys. (56 ref)
_______________________________________________________________
Boland
CS.
Title
Social support and spiritual well-being:
empowering older adults to commit to health-promoting behaviors.
Source
Journal of Multicultural Nursing &
Health, 6(3):12-23, 2000 Fall. (50 ref)
Abstract
Objectives: Purpose was to examine factors
that may enhance level of commitment by older adults to tile practice of
health-promoting behaviors. A proposed expansion of Pender's Revised Health
Promotion Model (HPM) was tested. Methods: A theoretical model was developed
and tested using Structural Equation Modeling in a sample of adults aged 65 and
older (N595) from communities in four states. Results: Selected Personal
Factors, Social Support (SS) and Spiritual-Well-Being (SWB) had direct and
indirect influence on level of commitment and were more relevant to the
performance of one health-promoting behavior than to that of another. The
combined effect of SS and SWB on commitment and specific behaviors was more
efficacious than individually. Recommended additions of spiritual factors to
Pender's HPM were supported Conclusions: Findings add to our understanding of
older adult health-promoting practices by underscoring the importance of social
and spiritual interactions for enco!
uraging
health behaviors. Inclusion of recommended spiritual factors would enhance the
HPM's effectiveness as a guide to investigate and explain influences in older
adult decisions to engage in HPBs and would facilitate design of more powerful,
culture-specific and individualized interventions to facilitate commitment to
healthy behaviors. (50 ref)
____________________________________________________________
Schnoll RA.
Harlow LL. Brower L.
Institution
Assistant Member, Fox Chase Cancer Center,
Cheltenham, Pennsylvania.
Title
Spirituality, demographic and disease
factors, and adjustment to cancer.
Source
Cancer Practice: a Multidisciplinary Journal
of Cancer Care, 8(6):298-304, 2000
Nov-Dec. (51 ref)
Abstract
PURPOSE: The purpose of this study was to
examine the relationship between demographic-disease variables, spirituality,
and psychosocial adjustment in a heterogeneous sample of patients with cancer.
DESCRIPTION OF STUDY: Participants (N = 83)
accrued through the Rhode Island Hospital and the American Cancer Society
completed questionnaires, and structural equation modeling was used to examine
the relationships among disease and demographic factors, spirituality, and
psychosocial adjustment to cancer.
RESULTS: Of five models tested, a
mediational model received the strongest support (chi-square(35)-66.61; P =
.005; comparative fit index = .90; root mean square error of approximation =
.09), explaining 64% of the variance in psychosocial adjustment. Being a woman,
having a longer illness duration, and having a lower disease stage were related
to greater levels of purpose in life and religious/existential beliefs, which,
in turn, were associated with higher levels of family and social adjustment and
psychological health.
CLINICAL IMPLICATIONS: The results indicate
that spirituality can influence how patients with cancer adjust to their
diagnosis and treatment and, thus, support the need for interventions that
target spirituality to promote psychosocial adjustment in this population. (51 ref)
Summary:
cancer rehabilitation; mediational model
____________________________________________________________
Resnick
B. Zimmerman SI. Orwig D.
Furstenberg A. Magaziner J.
Institution
University of Maryland School of Nursing,
655 West Lombard Street, Room 375, Baltimore, MD 21201. E-mail: bresnick@umaryland.edu.
Title
Outcome Expectations for Exercise Scale:
utility and psychometrics.
Source
Journals of Gerontology Series
B-Psychological Sciences & Social Sciences, 55B(6):S352-6, 2000 Nov.
(42 ref)
Abbreviated
Source
J GERONTOL B PSYCHOL SCI SOC SCI. 55B(6):S352-6, 2000 Nov. (42 ref)
Abstract
OBJECTIVES: The purpose of this study was to
develop a measure of outcome expectations for exercise specifically for the
older adult (The Outcome Expectations for Exercise [OEE] Scale), and to test
the reliability and validity of this measure in a sample of older individuals.
This scale was developed based on Bandura's theory of self-efficacy and the
work of prior researchers in the development of measures of outcome
expectations. METHODS: The OEE scale, which was completed during a face-to-face
interview, was tested in a sample of 175 residents in a continuing care
retirement community. RESULTS: There was support for the internal consistency
of the OEE scale (alpha coefficient of .89), and some support for reliability
based on a structural equation modeling approach that used R2 estimates,
although less than half of these were greater than 0.5. There was evidence of
validity of the measure based on: (a) a confirmatory factor analysis in which
the model fit the data (normed fit index [NFI] = .99, root mean square error of
approximation [RMSEA] - .07, chi2/df = 2.8); (b) support for the hypothesis
that those who exercised regularly had higher OEE scores than those who did not
(F = 31.3, p < .05, eta squared = .15); and (c) a statistically significant
relationship between outcome expectations and self-efficacy expectations (r =
.66). DISCUSSION: This study provides some initial support for the reliability
and validity of the OEE scale. Outcome expectations for exercise were related
to exercise behavior in the older adult, and the OEE scale can help identify
older adults with low outcome expectations for exercise. Interventions can then
be implemented to help these individuals strengthen their outcome expectations,
which may subsequently improve exercise behavior. (42 ref)
Summary:
reliability
____________________________________________________________
Pérez-Escamilla R. Cobas JA.
Balcazar H. Benin MH.
Institution
Department of Nutritional Sciences,
University of Connecticut, Storrs, CT 06269-4017.
Title
Specifying the antecedents of breast-feeding
duration in Peru through a structural equation model.
Source
Public Health Nutrition, 2(4):461-7, 1999 Dec. (33 ref)
Abstract
OBJECTIVE: To examine the effects of
socioeconomic status and biocultural variables (planned pregnancy, prenatal
care, timing of initiation of breast-feeding and caesarean section delivery) on
breast-feeding duration in Peru using structural equation models. DESIGN AND
SETTING: Structural equation models were analysed with LISREL using data from
the 1991-92 Peruvian Demographic and Health Survey. SUBJECTS: Models were
tested among 6,020 women whose last child was born within 5 years of the survey
and among 2,711 women whose last child was born 2-5 years preceding the survey.
RESULTS: Unplanned pregnancy and socioeconomic status had a negative influence
on breast-feeding duration. Prenatal care was positively associated with the
timing of breast-feeding initiation in both samples and with breast-feeding
duration in the whole sample. The timing of breast-feeding initiation was
inversely associated with breast-feeding duration only in the sample of older
children. CONCLUSIONS: These results imply that an unplanned pregnancy, a
delayed breast-feeding initiation, and higher socioeconomic status are risk
factors for an earlier discontinuation of breast-feeding through complex
mechanisms involving direct and indirect effects. (33 ref)
____________________________________________________________
Fontana A.
Schwartz LS. Rosenheck R.
Institution
Northeast Program Evaluation Center (182),
Veterans Affairs Medical Center, 950 Campbell Ave, West Haven, CT 06516.
Title
Posttraumatic stress disorder among female
Vietnam veterans: a causal model of etiology.
Source
American Journal of Public Health, 87(2):169-75, 1997 Feb. (29 ref)
Abstract
OBJECTIVES: The Vietnam and Persian Gulf
wars have awakened people to the realization that military service can be
traumatizing for women as well as men. This study investigated the etiological
roles of both war and sexual trauma in the development of chronic posttraumatic
stress disorder among female Vietnam veterans. METHODS: Data from the National
Vietnam Veterans Readjustment Study for 396 Vietnam theater women and 250
Vietnam era women were analyzed with structural equation modeling. RESULTS: An
etiological model with highly satisfactory fit and parsimony was developed.
Exposure to war trauma contributed to the probability of posttraumatic stress
disorder in theater women, as did sexual trauma in both theater and era women.
Lack of social support at the time of homecoming acted as a powerful mediator
of trauma for both groups of women. CONCLUSIONS: Within the constraints and
assumptions of causal modeling, there is evidence that both war trauma and
sexual trauma are powerful contributors to the development of posttraumatic
stress disorder among female Vietnam veterans. (29 ref)
____________________________________________________________
SEM and
quality of life (HRQL)
1.
Watkins
KW. Connell CM. Fitzgerald JT. Klem L. Hickey T. Ingersoll-Dayton B. (2000). Effect of
adults' self-regulation of diabetes on quality-of-life outcomes. Diabetes
Care, 23(10):1511-5, 2000 Oct. (30 ref)
Department
of Health Promotion and Education, School of Public Health, University of South
Carolina, Columbia, SC 29208. E-mail: kwatkins@sph.sc.edu.
OBJECTIVE:
To examine the relationships among cognitive representations of diabetes,
diabetes-specific health behaviors, and quality of life using Leventhal and
Diefenbach's self-regulation model of illness (Leventhal H, Diefenbach M: The
active side of illness cognition. In Mental Representation in Health and
Illness. SkeltonJA, Croyle RT, Eds. New York, Springer-Verlag, 1991, p.
247-272). RESEARCH DESIGN AND METHODS: This research involved secondary
analysis of a mailed survey completed by 296 adults (ages 20-90 years).
Structural equation modeling was conducted to investigate relationships among
cognitive representations, diabetes-specific health behaviors, and quality of
life. Model differences by diabetes type were also investigated. RESULTS:
Findings indicated that certain cognitive representation constructs were
related to increased diabetes-specific health behaviors, decreased sense of
burden, and positive quality-of-life outcomes. Individuals levels of
understanding!
of diabetes and their perceptions of control
over diabetes were the most significant predictors of outcomes. However,
diabetes-specific health behaviors were related to an increased sense of burden
that was negatively associated with quality of life. Multigroup analyses
indicated that this self-regulatory model provided a good fit for individuals
with type 1 diabetes, those with type 2 diabetes who take insulin, and those
with type 2 diabetes who do not take insulin. CONCLUSIONS: These findings
advance what is known about cognitive representations of illness and the
self-regulation of diabetes as well as the relationships between cognitive
representations of illness, quality of life, and behavioral factors. In
particular, results from this study suggest the need for further study to
address ways of reducing the burden of diabetes associated with health
behaviors and decreased quality of life.
(30 ref)
Publication
Type
Journal Article. Research. Tables/Charts.
2.
Siddiqui
O. Ali MW. (1999). Linear structural equation model in
analyzing quality-of-life-data from clinical trials. Journal of
Biopharmaceutical Statistics.
9(4):661-81, 1999 Nov.
Center
for Economics Research Unit, Georgetown University Medical Center, Washington,
DC 20007, USA.
Assessment
of quality of life (QOL) in clinical trials becomes a challenging task from the
viewpoint of clinical biostatistics. The responses of the items for measuring
QOL indices usually vary widely from patient to patient and from time to time.
Measurement errors might be present in the responses of the items, and they
might be correlated. Hence, in analyzing QOL data, the usual assumption that
there are no measurement errors in responses is too liberal. Because the QOL
indices are likely to be correlated, separate analysis of each index might not
be efficient from the point of view of statistical methodology. We apply linear
structural equation modeling (LISREL) in assessing the QOL data obtained from a
clinical trial. A basic premise of the LISREL approach is that the abstract
concepts (latent constructs) that are not directly measurable can be studied.
LISREL is a statistical procedure for conceiving and testing structural
hypotheses that cannot be tested adequately with other statistical procedures.
It allows us to specify relations between unobserved and observed variables
while controlling for measurement errors and correlations among both the
measurement errors and the latent constructs.
3.
Newsom
JT. Schulz R. (1996). Social support as a mediator in the relation
between functional status and quality of life in older adults. Psychology &
Aging. 11(1):34-44, 1996 Mar.
University
Center for Social and Urban Research, University of Pittsburgh, Pennsylvania
15620, USA.
The
relations among physical functioning, social support, depressive symptoms, and
life satisfaction were examined in a national sample of 4,734 adults age 65 and
older. Regression analyses were used to examine the relative importance of
objective and subjective support measures in understanding the relation between
physical impairment and quality of life. Impairment was associated with fewer
friendship contacts, fewer family contacts, less perceived belonging support,
and less perceived tangible aid, but only measures of perceived support
predicted depressive symptomatology. A structural equation modeling approach
was then used to explore the mediational role of perceived social support in
the relation between impairment and quality of life variables. Results are
consistent with the hypothesis that lower reported social support is an important
reason for decreases in life satisfaction and increases in depressive symptoms
found among older adult populations. Implications for understanding the role of
social support in attenuating the effects of physical disability in older
adults are discussed.
4.
Mitchell
WG. Scheier LM. Baker SA. (1994). Psychosocial, behavioral, and medical outcomes in children with
epilepsy: a developmental risk factor model using longitudinal data.
Pediatrics. 94(4 Pt 1):471-7, 1994 Oct.
Department
of Neurology, University of Southern California School of Medicine, Childrens
Hospital, Los Angeles 90027.
OBJECTIVE.
We studied factors predicting the risk of adverse long-term psychosocial,
behavioral, and medical outcomes in children with epilepsy. METHODS. Children
(N = 157, 4.5 to 13 years) were enrolled in a prospective longitudinal study
when first seen. Potential subjects were excluded if they were moderately or
severely mentally retarded, had motor or sensory handicaps interfering with
testing, or did not speak either English or Spanish. MEASURES. To develop risk
predictors, we collected information regarding the child's medical and seizure
history, cognitive functioning, and behavior problems, and family functioning.
Children and their families were followed for a minimum of 18 months, then
underwent reassessment of medical status, parent's attitudes toward epilepsy,
and the child's behavioral and cognitive functioning. Data were analyzed by
confirmatory factor analysis to develop baseline factors (Sociocultural Risk,
Seizure Risk, and Behavior Problems) and outcome fac!
tors
(Medical/Seizure Problems, Parent's Negative Attitudes Toward Epilepsy, and
Behavior Problems), followed by structural equation modeling to determine
across-time causal effects. Eighty-eight subjects completed all baseline and
outcome measures. RESULTS. Among significant across-time effects, Medical
Outcome was predicted by Seizure Risk. An increased number of stressful life
events predicted better Medical Outcome. Low acculturation increased Parent's
Negative Attitudes and was associated with increased Behavior Problems at
baseline. Behavior Problems were stable across time. It is interesting that IQ
did not affect any of the outcomes, although its effect may have been mediated through
other baseline measures. CONCLUSIONS. Seizure history was the best predictor of
ongoing medical difficulties, whereas the most important causes of ongoing
parental anxiety and negative attitudes toward epilepsy were sociocultural.
Variation in medical or attitudinal outcomes was not influe!
nced by
either the child's IQ or reported behavioral problems. These findings suggest
that to alter attitudes toward epilepsy, programs should be tailored to the
sociocultural background of the family. Studies of quality of life of children
with epilepsy should include appropriate sociocultural measures.
_______________________________________________________________________
GJ
Botvin's name comes up as one who uses SEM to study longitudinal data, especially
as it relates to prevention. Several
examples:
1.
Scheier LM.
Botvin GJ. Griffin KW.
Institution
Department of Public Health, Institute for
Prevention Research, Weill Medical College of Cornell University, New York, New
York 10021, USA. lmscheie@med.cornell.edu
Title
Preventive intervention effects on
developmental progression in drug use: Structural equation modeling analyses
using longitudinal data.
Source
Prevention Science. 2(2):91-112, 2001 Jun.
Abstract
This study examined the plausibility of the
gateway hypothesis to account for drug involvement in a sample of middle school
students participating in a drug abuse, prevention trial. Analyses focused on a
single prevention approach to exemplify intervention effects on drug
progression. Improvements to social competence reduced multiple drug use at 1-
and 2-year follow-ups. Specific program effects disrupted drug progression by
decreasing alcohol and cigarette use over 1 year and reducing cigarette use
over a 2-year period. Controlling for previous drug use, alcohol was integrally
involved in the progression to multiple drug use. Subgroup analyses based on
distinctions of pretest use/nonuse of alcohol and cigarettes provided partial
support for the gateway hypothesis. However, evidence also supported alternate
pathways including cigarette use as a starting point for later alcohol and
multiple drug use. Findings underscore the utility of targeting more than one
gateway substance to prevent escalation of drug involvement and reinforce the
importance of social competence enhancement as an effective deterrent to
early-stage drug use.
2.
Epstein JA.
Griffin KW. Botvin GJ.
Institution
Institute for Prevention Research, Cornell
University Medical College, New York, New York 10021, USA.
Title
Role of
general and specific competence skills in protecting inner-city adolescents
from alcohol use.
Source
Journal of Studies on Alcohol. 61(3):379-86, 2000 May.
Abstract
OBJECTIVE: The purpose of this longitudinal
investigation was to test whether higher levels of general competence are
linked to greater refusal assertiveness that is, in turn, related to less
subsequent alcohol use among inner-city adolescents. METHOD: A large sample of
students attending 22 middle and junior high schools in New York City
participated. Students completed surveys at baseline, at 1-year follow-up and
at 2-year follow-up (N = 1,459; 54% female). The students self-reported alcohol
use. decision-making skills, self-efficacy and refusal assertiveness. Teams of
three to five data collectors administered the questionnaire following a
standardized protocol. The data were collected in school during a regular
40-minute class period. RESULTS: According to the tested structural equation
model, Decision Making (beta = .07, p < .05) and Self-Efficacy (beta = .24,
p < .001) predicted higher Refusal Assertiveness and this greater
assertiveness predicted less drinking at the 2-year follow-up (beta = -.21, p
< .001). Earlier drinking predicted 2-year follow-up drinking (beta = .40, p
< .001). Goodness-of-fit indices were excellent (chi2 = 1107.9, 238 df, N =
1,438, p < .001; NFI = .93, NNFI = .94, CFI = .95). CONCLUSIONS: The tested
model had a good fit and was parsimonious and consistent with theory. This
research highlights the importance of addressing decision-making skills,
self-efficacy and refusal assertiveness within adolescent alcohol prevention
programs.
3.
Epstein
JA. Griffin KW. Botvin GJ.
Institution
Institute for Prevention Research, Cornell
University Medical College, 411 East 69th St, KB 201, New York, NY 10021;
jepstein@mail.med.cornell.edu.
Title
A model of smoking among inner-city
adolescents: the role of personal competence and perceived social benefits of
smoking.
Source
Preventive Medicine, 31(2 part 1):107-14, 2000 Aug. (48 ref)
Abstract
BACKGROUND: Based on current trends, smoking
will remain a major public health problem in the 21st century. Effective
smoking prevention approaches offer the best hope for decreasing the rise in
adolescent smoking rates. Competence enhancement approaches to smoking
prevention are among the most successful. Yet, there is not a full
understanding of how effective prevention approaches work. This study tests
whether a deficiency in competence (poor decision-making skills and low
personal efficacy) is linked to acquiring beliefs in the perceived benefits of
smoking and whether these perceived benefits are then related to subsequent
smoking. METHODS: A sample of 1459 students attending 22 middle and junior high
schools in New York City participated. Students completed surveys at baseline,
1-year follow-up and 2-year follow-up during a regular class period. They
self-reported smoking, decision-making skills, personal efficacy and beliefs in
the perceived benefits of smoking. RESULTS: The tested structural equation
model had a good fit and was parsimonious and consistent with the theory
underlying the competence approach to smoking prevention. CONCLUSIONS: This
research highlights the importance of addressing decision-making skills,
personal efficacy, and beliefs in the social benefits of smoking within
adolescent smoking prevention programs. Copyright 2000 American Health
Foundation and Academic Press. (48
ref)
4.
Epstein JA.
Griffin KW. Botvin GJ.
Institution
Institute for Prevention Research, Cornell
University Medical College, New York, NY 10021, USA.
jepstein@mail.med.cornell.edu
Title
Competence skills help deter smoking among
inner city adolescents.
Source
Tobacco Control. 9(1):33-9, 2000 Mar.
Abstract
OBJECTIVE: To test whether higher levels of
general competence are linked to more frequent use of refusal assertiveness
that is in turn related to less subsequent smoking among inner city
adolescents. METHODS: Longitudinal study conducted during three year middle
school or junior high school period. A sample of 1459 students attending 22
middle (ages 11-14 years) and junior high (ages 12-15 years) schools in New
York City participated. Students completed surveys at baseline, one year follow
up, and two year follow up. The students self reported smoking, decision making
skills, personal efficacy, and refusal assertiveness. Teams of three to five
data collectors administered the questionnaire following a standardised
protocol. These data were collected in school during a regular 40 minute class
period. RESULTS: Based on the tested structural equation model, decision making
and personal efficacy (that is, general competence) predicted higher refusal
assertiveness and this greater assertiveness predicted less smoking at the two
year follow up. The tested model had a good fit and was parsimonious and
consistent with theory. CONCLUSIONS: Adolescent smoking prevention programmes
often teach refusal skills in order to help youth resist peer pressure to
smoke. The present findings suggest that teaching general competence skills as
well may help to reduce smoking because youth with better personal efficacy and
decision making skills are better able to implement smoking refusal strategies.
______________________________________________________________________