1. Introduction
Relationship quality is the most frequently studied construct in close relationships [
1]. Following the initial studies by Hamilton in 1929 and Terman in 1938, a large number of studies with respect to relationship quality have been carried out in order to identify relationship quality, its determinants and consequences [
2]. Relationship quality is not only investigated by psychologists, but also by researchers from the field of sociology, family studies, and other disciplines [
3,
4].
Relationship quality involves individuals in a wide range of relationships, including those who live together, who are married, and who have a romantic relationship. However, due to its legal association, it is easier to identify married couples, which is why most relationship quality studies are carried out with them [
5]. In addition to this, there are studies on relationship quality between married and unmarried couples. In a recent study [
6] examining the relationship quality of cohabiting and married couples, direct married couples’ relationship quality was reported highest, whereas it was reported the lowest with couples who cohabit and do not plan to marry. Marrieds who premaritally cohabited and cohabiters with plans to marry remained undifferentiated and indicated an intermediate level of relationship quality.
Relationship quality is a determinant of quality of life and well-being [
7,
8]. In a meta-analysis conducted with 126 articles and more than 72,000 participants, examining the relationship between marital quality and physical health revealed that high marital quality was associated with better health and lower mortality rate [
9]. In order to determine whether physical and mental health associated with marriage also applied to college students, a study with 1621 dating college students was conducted [
10]. It was found that individuals in committed relationships were less likely to be overweight and less likely to experience mental health problems. But there were no significant differences between single students and students in romantic relationships. Moreover, relationship quality has proven to have several effects on children [
11].
According to Fletcher et al. [
12], for researchers who want to measure relationship quality it is very challenging to decide which scale to use, as there are many measurement tools and the terms used to define the measured subject differ. Some of the self-report measurement tools used to measure relationship quality are as follows: Marital Adjustment Test [
13], the Dyadic Adjustment Scale [
14], the Quality of Marriage Index [
15], the Kansas Marital Satisfaction Scale [
16], the Golombock–Rust Inventory of Marital State [
17], the Relationship Assessment Scale [
18], the Perceived Relationship Quality Components [
12], and Relationship Quality Scale [
19].
The quality of relationship is studied by using several terms such as “satisfaction, adjustment, happiness, companionship, success” [
3,
4,
20]. For this reason, there is still no consensus on how these terms will be defined or measured [
2]. According to Fincham and Rogge [
3], researchers who work on marriage start from two basic approaches. First, the
relationship or
interpersonal approaches typically focus on interaction patterns such as companionship, communication, or conflict and is preferably called adjustment. Second, the
intrapersonal approach involves the couples’ subjective evaluations on marriage and works with terms such as marital satisfaction or marital happiness. Even though the authors find both approaches valuable, they have criticized them for their problems with measurements. According to Fincham and Rogge [
3], researchers have worked on the same subject despite the conceptual distinctions they made when introducing their research. For instance, The Marital Adjustment Test and the Dyadic Adjustment Scale, which are the two most commonly used measures of relationship quality, include items that evaluate both intrapersonal and interpersonal processes. In sum, relationship quality is traditionally conceptualized as unidimensional, and a total score is obtained. According to Fincham and Rogge [
3], evaluating relationship quality unidimensionally blurs this important phenomenon and over-simplifies the theories.
Measurements of relationship quality with the spouse or the romantic partner are important both in understanding the nature of the relationships and in arranging the intervention programs that will help to make the relationships more positive. Self-report scales that have proven reliability and validity in determining relationship quality are essential measurement tools for researchers. In Turkey, some of these tools are used to examine marital or dating relationships (e.g., Dyadic Adjustment Scale, Marital Adjustment Test). Even if the relevant measurement tools provide valuable information on relationships in Turkey, in all the cases, relationships are regarded as unidimensional. However, as mentioned above, relationships have both positive and negative qualities at the same time, in other words, they are bidimensional.
The aim of this research is to adapt the Positive–Negative Relationship Quality scale [
21], which measures the quality of relations in two dimensions, into Turkish. Additionally, this study aims to test different factorial models including orthogonal factors model, correlated factors model, one-factor model, and bifactor model. Bifactor model is especially important to see whether the relationship quality can be approved as a two-dimensional structure by different latent variables rather than as proposed by Rogge et al. with item-response theory.
In order to determine the validity and reliability of the scale, two studies were performed. The first was carried out with a student sample and had the main purpose to examine the construct validity of the measurement tool in the Turkish sample with an exploratory analysis. In the second study, confirmatory findings and criterion validity studies were added. Analyses in both studies were performed using IBM SPSS statistics 23 (IBM Corp. Released 2015 IBM SPSS Statistics for Windows, Version 23.0. Armonk, NY, USA) and Lisrel 8.80 (Scientific Software International, Inc. Released 2006 Lisrel for Windows, Version 8.80. Lincolnwood, IL, USA).
4. Discussion
The aim of this study was to adapt the PN-RQ scale into Turkish and to determine the validity and reliability of the scale. Exploratory factor analysis results, obtained from both the first and the second study, revealed that the relationship quality can be regarded as a bifactorial structure and that there is also a negative relationship between these factors. For the criterion-related validity of the correlation of PN-RQ with DAS, which is frequently used in the literature, the Cronbach’s alpha internal consistency coefficients, obtained in both the first and second studies, provide evidence for the reliability. All of the findings in the Turkish sample are in line with the findings obtained in the original scale study.
In addition to the findings of the exploratory factor analysis, the identification of the bifactor model, as the best model among the various confirmatory factor models that were tested using the married sample, provides evidence for the multidimensional nature of the PN-RQ scale. The bifactor model has been operated with particularly with growing interest in the analysis of the dimensionality problem in psychological research. Bifactor models may be practical when there is a general factor ensuring commonality among domains that are item-related, when there are multiple domain or group factors which go beyond the general factor and influence the particular domain, or when both domain factors and the common factor are relevant for the research [
26].
In this study, the bifactor model was investigated with one positive quality factor using positive adjectives as indicators, one negative quality factor using negative adjectives as indicators, and one global factor that was indicated by both positive and negative adjectives. This global factor allowed the model to take into account participants with high scores in positive and negative factors simultaneously, named as the ambivalent group, as well as participants with low scores, the indifferent group.
As mentioned above, the relationship quality is conceptualized as one-dimensional in the literature, and a total score is obtained by means of self-report instruments. According to Fincham and Rogge [
3], relationship quality is two-dimensional and the positive qualities of the relationship are distinct from their negative qualities. One-dimensional evaluation of the relationship quality blurs this important phenomenon and over-simplifies theories. Based on these criticisms, the PN-RQ scale developed by Rogge and colleagues [
21] measures relationship quality considering its both positive and negative aspects. It also makes it possible to categorize people into four different groups: satisfied (who rate positive adjectives with high scores and negative adjectives with low scores), dissatisfied (who rate negative adjectives with high scores and positive adjectives with low scores), indifferent (who rate both positive and negative adjectives with low scores), and ambivalent (who rate both positive and negative adjectives with high scores). Comparisons that can be made after such a categorization will be much more distinct than the comparisons that are made based on a single total score.
Both the long and the short version of the PN-RQ scale are highly functional for researchers. The short version is suitable for those who will work with small sample groups, and the long version for those who will use a large number of measurement tools. The long version of the scale is already conveniently short.
The above-mentioned strengths have some limitations. In this study, the test–retest reliability of the scale was not investigated. It has been studied with university students who already had a relationship or married people. In the first and second study, the sample consisted mostly of female participants (76% and 70%, respectively). Moreover, in the first study, only heterosexual students participated, and a significant difference can be detected in the length of relationship in both of the studies. It would be insightful to have considered the difference within this variable both qualitatively and quantitatively. Future studies that will take longitudinal measurements, balance the gender of the participants, include people who are not married but do live together or people who live together even though they are divorced, and gather data from couples will help to eliminate these deficiencies.
In this study, the short version of the PN-RQ scale included different adjectives from the original study. This can be explained by cultural differences. Although characteristics in the evaluation of relationships that pertained more to Turkish participants were similar to US participants’ characteristics, they did not seem to coincide. Within the scope of this study, the short version of the PN-RQ scale is based on a sample of married couples. When considering Turkey’s cultural characteristics, the short version that can be used for romantic partners who are not married can be determined by means of future studies.