Testing the Invariance of the Transtheoretical Model Utilizing a Population of Smokers

A crucial assumption in the development of a measure is that the measurement model is invariant. This means that the measurement model is the same in different populations and across different occasions. Utilizing data from a representative sample of smokers, we examined the invariance of the Decisional Balance Inventory and the Processes of Change. The sample consists of 4144 participants (44.3% male, 55.7% female, 95.7% White, mean age 40.7). After the number of factors for each scale was confirmed using Factor Analysis, the sample was partitioned into subgroups to test the invariance of the model. The scales were compared across two levels of gender (male and female), four levels of education (grade school, high school, college, and graduate school), two levels of race (white and black), and four levels of age (18-25, 26- 35, 36- 50, and 50+). Four different levels of invariance were evaluated. The first level, configural invariance, had no restrictions. The second level examined was the tau-equivalent model also known as weak factorial invariance and requires that the model loadings be proportional. The third level, pattern identity invariance requires the factor loadings to be equal. The fourth model tested is the most restrictive level also known strong factorial invariance. Strong factorial invariance requires that the loadings, error variances, means, and standard deviations be equal. The results indicated that the two TIM scales measure at high levels of invariance across the samples and across time. The research supports the premise that the Decisional Balance and Processes of Change scales can be used in a variety of populations and in longitudinal studies