Sociology, Department of


Date of this Version



Appendix doi:10.32873/unl.dc.oth.007


Copyright 2020 McQuillan, J., A. L. Greil, A. Rybinsk, S. Tiemeyer, K. M. Shreffler, and C. Warner Colaner.


Barnett, Marshall, Raudenbush, and Brennan (1993) provide an approach to crosssectional couple data that increases the power and precision of the estimates and allows the simultaneous analysis of both partners. Following Barnett et al (1993), we constructed two parallel versions of the distress scale for each partner by matching standard deviations of individual items. We randomly assigned each of the 10 items to an A or B group (each group had five items). This procedure resulted in two parallel subscales of distress for each partner. Overall, the subscales had equal amounts of error variance. The level 1 data were “stacked” with four rows of data for each couple and columns containing the partner ID, the individual ID, an indicator for women, an indicator for men, and a column for the distress subscales. The level 2 (couple level) data contained one row for each couple and columns for all of the other variables.