Work

A Comparison of Latent Growth Models for Constructs Measured by Multiple Indicators

Year: 2005
Type: dissertation
Abstract: Latent growth modeling (LGM) of composites of multiple items (for example, \nmeans or sums of items) has been frequently used to analyze the growth of latent \nconstructs. However, composites are only equivalent to latent constructs if the items’ \nfactor loadings are equal to one and there is no measurement error (Bollen & Lennox, \n1991). In this study, the adequacy of using univariate LGM to model composites of \nmultiple items, as well three other alternative methods were evaluated through a Monte \nCarlo simulation study. The four methods evaluated in this study were the univariate \nLGM, the univariate LGM with fixed error variances, the univariate LGM with the \ncorrection for attenuation, and the curve-of-factors model (McArdle, 1988; Tisak and \nMeredith, 1990). \nThis simulation study manipulated the number of items per construct, the number \nof measurement times, the sample size, the reliability of the composites, the invariance of \nitem parameters, and whether the items were essentially tau-equivalent or essentially \ncongeneric. One thousand datasets were simulated for each of the conditions. \nThe results indicate that using univariate LGM with composites of multiple items \nonly produces unbiased parameter estimates and standard errors if the items are \nessentially tau-equivalent. The univariate LGM with fixed error variances performed \nidentically to the univariate LGM. The univariate LGM with the correction for \nattenuation produced unbiased parameter estimates when the items were essentially tauequivalent, \nbut produced negatively biased estimates of standard errors. \nThe curve-of-factors model was found to be the most appropriate method to \nanalyze the growth of latent constructs measured by multiple items. The curve-of-factors \nmodel was able to provide unbiased parameter estimates and standard errors under all \nconditions evaluated in this study. However, with sample sizes of 100 or 200, a large \npercentage of chi-square statistics were positively biased and the fit indices indicated \ninadequate model fit. \nThis study’s recommendation is that the curve-of-factors model should be \npreferred to analyze the growth of latent variables measured by multiple items, but the \nuse of sample sizes larger than 200 is strongly recommended to help ensure that adequate \nfit statistics and fit indices are obtained for appropriate models.
Author: Walter Lana Leite
Language: English
Open Access status: green
OpenAlex ID: W7027062502
Cites: 87
Cited by: 2
Related to: 10