Page 528 - Encyclopedia of Nursing Research
P. 528
STRUCTURAl eQUATIOn mODelIng n 495
dyad risk assessment profile to determine relationships. It is used interchangeably with
patient needs and caregiver concerns prior to the terms causal modeling, covariance struc-
discharge. Joan grant documented the effec- ture modeling, and lISRel modeling. The S
tiveness of her problem-solving intervention theoretical issues are discussed in the entry
in reducing stroke caregiver depression and on Causal Modeling. A description of the ana-
improving caregiver perceived health (grant, lytic issues when programs such as lISRel
elliott, Weaver, Bartolucci, & giger, 2002). or eQS are used will ensue.
Rosemarie King has also been funded to eval- Sem techniques are extremely flexible.
uate the effectiveness of her problem-solving most models of cause can be estimated. In
intervention for stroke caregivers. Tamilyn some models, the causal flow is specified only
Bakas was funded to develop and pilot test the between the latent variable and its empirical
Telephone Assessment and Skill-Building Kit, indicators, such as in a factor analysis model.
which has shown evidence of content validity This is known as confirmatory factor analy-
and satisfaction in stroke caregivers (Bakas sis. In other models, causal paths among the
et al., 2009). The Telephone Assessment and latent variables also are included.
Skill-Building Kit program is currently being Conducting a confirmatory factor analy-
tested in a larger randomized controlled clin- sis with Sem has many advantages. With
ical trial. linda pierce has been funded to test Sem, the analyst can specify exactly which
her intervention titled, “The Caring Web” for indicators will load on which latent variables
stroke caregivers, which has been found to (the factors), and the amount of variance in
reduce emergency department visits and hos- the indicators not explained by the latent
pital readmissions of stroke survivors (pierce, variable (due to error in either measure-
Steiner, Khuder, govoni, & horn, 2009). All ment or model specification) is estimated.
of these studies show great potential toward Correlations between latent variables and
improving the care and well-being of families among errors associated with the indicators
of stroke survivors. can be estimated and examined. Statistics
now is a very fruitful time for nurses to that describe the fit of the model with the data
conduct research in the area of stroke and allow the analyst to evaluate the adequacy
stroke caregivers. With stroke being a lead- of the factor structure, make theoretically
ing cause of serious, long-term disability in appropriate modifications to the structure
the United States, it is imperative that nurses based on empirical evidence, and test the
take the lead in developing programs that change in fit caused by these modifications.
improve the care of stroke survivors and Thus, confirmatory factor analysis provides
their family members. a direct test of the hypothesized structure of
an instrument’s scales.
Tamilyn Bakas An advantage of using Sem to estimate
Staci S. Wuchner models containing causal paths among the
latent variables is that many of the regression
assumptions can be relaxed or estimated. For
example, with multiple regression, the ana-
Structural equation lyst must assume perfect measurement (no
measurement error); however, with Sem,
modeling measurement error can be specified and the
amount estimated. In addition, constraints
can be introduced based on theoretical expec-
Structural equation modeling (Sem) is tations. For example, equality constraints, set-
used to describe theoretical and analytic ting two or more paths to have equal values,
techniques for examining cause-and-effect are useful when the model contains cross-

