An empirical approach for constructing scales for behavioral marital assessment devices
The present study examined several facets of generalizability, two method of scale construction, and the cross-sample stability of data obtained from a 24 category marital observational coding system. At approximately two months following the birth of their first child, two samples of married couples were video taped while discussing two types of tasks. Undergraduate research assistants coded the video taped samples using the Marital Coding System, a variation of the Marital Interaction Coding System where some behaviors were refined and some new behaviors were added to capture couples emotionalsupportive behavior, as well as, their problem solving behavior. The results of the generalizability studies showed that the MCS codes were heterogeneous for the between samples and between tasks analyses. However, the between obervers analysis showed an acceptable degree of homogeneity. The comparison of an a priori approach with an empirical approach for constructing scales indicated that each method had a variety of strengths and weaknesses when applied to data obtained from an observational coding system. However, examination of the results of both methods showed that some overlap existed between them. thereby, indicating some validity for the conceptual between meaning of the code categories. The overlap between the two methods of scale construction was most prominent for the code categories classified as negative verbal behavior. Negative verbal behavior also demonstrated the most consistency across samples and tasks for the cross-validational study. This finding becomes even more impressive given the apparent heterogeneity of variance between the samples and tasks. Finally, the relationship among the different facets of this study and their implications for future research with marital communication were discussed.