Postpartum physical activity: measuring theory of planned behavior constructs

Abstract

Objectives: To produce evidence for factor validity and longitudinal invariance of scales used to examine the theory of planned behavior applied to physical activity.

Methods: Self-report questionnaires were administered at 3- (n = 267) and 12-months (n = 333) postpartum.

Results: A single-factor model fit data from the normative beliefs, perceived behavioral control, and behavioral beliefs scales. Attitude and control beliefs were found to be multidimensional. Longitudinal invariance of all scales was supported.

Conclusions: Each scale had strong validity evidence. Future research using these measures will help identify areas for intervention and reveal how changes in these constructs influence physical activity.

Figure 1

General model used to test longitudinal invariance for each scale. The model depicts the factor structure of a hypothetical 3-item scale measured at two time points (3 months postpartum and 12 months postpartum). Circles represent latent factors, while rectangles represent the observed items. In each model the item level error variance (“e” to left of item boxes) are allowed to correlate over time. This shows that the variance not explained by the model for a given item will be similar at the two time points. The covariance between the latent factor at 3- and 12-months postpartum is also estimated.

Figure 2

General form of the correlated factors and higher order model. Each model depicts the factor structure of a hypothetical 6-item scale measured at two time points (3 months postpartum [3M] and 12 months postpartum [12M]). Circles represent latent factors, while rectangles represent the observed items. In each model the item level error variance (“e” to left of item boxes) are allowed to correlate over time. This shows that the variance not explained by the model for a given item will be similar at the two time points. In the higher order model the two first order factors (factors 1 and 2) are modeled as indicators of the higher order factor.