It is recognized that two analytical approaches can be utilized in longitudinal couple studies within hierarchical linear modeling. First, a 3-level model can be created where repeated assessments (Level-1) are nested within the individual (Level-2) that are nested within the couple (Level-3) [
4]. However, we have chosen to utilize the more common approach, which is to nest the individual repeated assessments (Level-1) within the couple (Level-2) ([
4]; S. [
43]). Therefore, we used the OpDes Program for power estimation of hierarchical linear models (S. W. [
44]) to calculate the sample size needed for our analyses. Specifically, with a frequency of 4 measurement occasions, a duration of 6 months, within-person variance of 1.0, a growth rate of 1.0, and a moderate effect size (.40), a total of 200 couples (i.e., 100 couples per condition) are needed to show a significant adherence to MVPA as measured via accelerometry. The effect size represents the low-end of findings from prior intervention research with this demographic [
17,
21,
22,
34,
38], yet it is clearly in the clinically meaningful range (D.E.R. [
65]). These studies showed mean increases of moderate to vigorous physical activity of 80 min per week, which is over half of the recommended weekly activity for public health [
62]. Our sample size includes a potential 25% attrition rate similar to the longitudinal study (thus total recruitment
N = 267. The attrition in the prior trial was actually 15% [
48], but we sought to oversample to accommodate the active component of this experimental trial compared to the prior passive prospective design. Our over-sampling procedures account for attrition due to second pregnancy or other possible reasons for drop out such as break-up, moving away, etc. The prediction-based research will be examined by group condition as well as via the collapsed sample for mediation analyses. Considering an average of five predictor IVs (TPB model), and using a small-medium effect size (f
2 = .10) we will have sufficient power (.80) to evaluate these predictors at an alpha of .05. Our longitudinal study also supported the use of a small-medium effect size as an appropriate criterion [
47]. Finally, the evaluation of physiological outcomes of participants across time will follow a 2 (condition) × 2 (time) interaction. The proposed sample size is, therefore, more than adequate to ensure sufficient statistical power for the physiological measurements.