Estimating individual influences of behavioral intentions: An application of random-effects modeling to the theory of reasoned action.

Abstract

Methods are proposed and described for estimating the degree to which relations among variables vary at the individual level. As an example of the methods, M. Fishbein and I. Ajzen's (1975; I. Ajzen & M. Fishbein, 1980) theory of reasoned action is examined, which posits first that an individual's behavioral intentions are a function of 2 components: the individual's attitudes toward the behavior and the subjective norms as perceived by the individual. A second component of their theory is that individuals may weight these 2 components differently in assessing their behavioral intentions. This article illustrates the use of empirical Bayes methods based on a random-effects regression model to estimate these individual influences, estimating an individual's weighting of both of these components (attitudes toward the behavior and subjective norms) in relation to their behavioral intentions. This method can be used when an individual's behavioral intentions, subjective norms, and attitudes toward the behavior are all repeatedly measured. In this case, the empirical Bayes estimates are derived as a function of the data from the individual, strengthened by the overall sample data. Psychological theories often posit relations among variables that can vary depending on the individual. Thus, it may be stated that on average, a given variable X influences another variable Tto a certain degree, but that the degree of this relationship is not a constant but varies to some degree in the population of individuals. In other words, a relationship between two variables may be quite strong in some individuals, whereas for others it may be weak or nonexistent. In clinical research, for example, the effect of a given therapeutic agent for treating, say, depression is thought to vary to some degree from individual to individual; in some individuals, the therapeutic agent, given at the same dose and under the same conditions, is effective whereas for others it is not. Most traditional statistical methods, on the other hand, provide estimation of relations among variables on a group and not on an individual level. A

How to assess social regulation of stress and emotions in daily family life? A computer-assisted family self-monitoring system (FASEM-C)

“…Multilevel models (or random coefficient models, see Longford (1995) or for an example Hedeker et al (1996)) allow the hierarchical structure of the family selfobservation data to be modelled and controlled where there are at least three different levels: within families there are individuals, and within individuals we have multiple observations. Therefore a three-level model was chosen, allowing the intercept to vary at each level.…”

Effect of nicotine on negative affect among more impulsive smokers.

“…For example, repeated-measures analysis of variance models assume that repeated measurements over time are equally correlated, with constant variance. Random-effects regression models with autoregressive errors, on the other hand, assume that one measurement will be more strongly correlated with measures at points closer in time and less strongly correlated with measures that are temporally distant (Gibbons et al, 1993;Hedeker, Flay, & Petraitis, 1996).…”

An Introduction to Growth Modeling

“…Again, it is feasible that the drug effects on outcome are stronger for endogenous than nonendogenous patients. Although these possibilities will not be considered here, an example of an HLM allowing such individual variation in relationships is described by Hedeker, Flay, and Petraitis (1996). Fitting the present model yields the results given in Table 12.6.…”