A scale to measure college student relationship involvement

College Student Journal, March, 2006 by Mark Whatley, Gelena M. Little, David Knox

The Relationship Involvement Scale was created to allow individuals to identify the degree to which they are involved in a relationship. The developed scale includes norms, reliability, and validity and was completed by a total of 306 undergraduates at two southern universities. While there were no significant differences between women and men on perceived level of involvement, Whites were significantly more involved than non-Whites and seniors significantly more than freshmen, sophomores, and juniors. Limitations and future uses of the scales are suggested.

**********

"We have been hanging out together," "We have a thing for each other," and "We are an item" are words individuals say to friends and family to express that they are involved in a relationship. But "exactly how involved is the couple?" we sometimes wonder. For most people, involvement is an important consideration in romantic relationships. Over 90% of Americans agree lack of devotion (e.g., infidelity) is wrong (Treas & Giesen, 2000). However, what people consider high devotion or involvement in a monogamous relationship greatly depends on how involvement is defined. Furthermore, people may dishonestly respond in socially desirable ways that makes finding solid statistics on relationship involvement a challenging enterprise.

This study developed a scale to measure involvement in romantic relationships. Literature inquiries yielded no past measures for this subject. The Level of Relationship Involvement Scale (see Appendix A) was developed to empirically identify the degree to which individuals in a developing relationship are involved. Previous attempts included as scale by King and Christensen (1983) who identified six levels, the last of which was living together, being engaged or married. We felt that developing this scale would assist future researchers and counselors in a number of areas, such as exploring compatibility issues, commitment anxiety, and areas of weakness in romantic relationships.

Development of the Relationship Scale

The Relationship Scale was developed by identifying 60 statements commonly thought to reflect involvement in a relationship. These statements were then examined through an exploratory factor analysis using SPSS version 11.0. The analysis allows one to establish the underlying relationship among a set of items (Pedhazur & Schmelkin, 1991; Tabachnick & Fidell, 1989, 2001). The items comprising a scale are inferred to indicate the unobservable or latent construct of interest (e.g., how committed the individual is to a partner). In the role of scale development, factor analysis is generally utilized to select sets of items that are indicative of the underlying construct the researcher is trying to assess (e.g., Glick & Fiske, 1996). The objective of factor analysis, then, is to discern the minimum number of factors that are compatible with the data.

The factor analysis used the maximum likelihood method of extraction and varimax rotation. Maximum likelihood minimizes the discrepancy between the population and sample covariance matrix, thereby maximizing the fitting function. Varimax rotation maximizes the factor loading variance and is the most commonly used method (e.g., Tabachnick & Fidell, 1989). Factor loadings are indicative of the relationship between a measure or item and the underlying factor. Factor loadings are seen as meaningful only when the value of the factor loading is equal to or exceeds .30.

The results of the factor analysis indicated 13 factors with eigenvalues greater than 1.0. Simply stated, an eigenvalue is a measure of the variance accounted for by a given factor (Pedhazur & Schmelkin, 1991; Tabachnick & Fidell, 1989). After inspection of the scree plot, a one-factor solution consisting of 10 items were deemed appropriate. An 10 item solution allows greater flexibility in research applications.

The coefficient of variation for Factor I was .29. The coefficient of variation allows a comparison and assessment of the amount of variation that exists in a measure (Howell, 1992). The higher the value the more variation exists, and the greater the variation the greater the ability of a measure to discriminate between groups. The reliability (internal consistency) was .91. Factor loadings and items are presented in Table 1.

Participants' scores on the Relationship Involvement Scale were correlated with participants' response to two questions. First, participants were asked to indicate how long they were in their present relationship. Those scoring higher on the relationship involvement scale had been in a relationship significantly longer than those scoring lower on the scale, r(214) = .23, p <. 01. Second, those not currently in a relationship were asked to divulge how long their last relationship lasted. The last relationship of those scoring higher on the relationship involvement scale had a significantly longer duration than those scoring lower on the scale, r(85) = .29, p <. 01. These two analyses provided initial scale validation of participants who were more involved in the relationship with their partner stayed in that relationship longer.