Third, to test the function of the tested variables, I use linear regression analyses. In particular, I compare the association between variables across cultures and genders. A significant association between two variables, for a particular country x gender group, indicates the dependent variable may be causing a change (or have a specific function) to the independent variable. Although a linear regression cannot confirm causality, it nevertheless provides evidence for a possible association between variables.
4.3 INSTRUMENTS
Attitudinal measures. Attitudinal measures have been shown to predict behaviors better than beliefs (e.g., Fishbein & Ajzen, 1975; Kassin, Fein, & Markus, 2011) and will be used in this study. An attitude is defined as a “psychic and nervous state or preparation, organized by experience, with a dynamic and directional (positive or negative) influence on individuals, with regard to any objects or situations he is related to” (Allport, 1935). Attitudes and beliefs alike include a cognitive aspect (list of criteria), but contrary to beliefs, attitudes also involve a positive or negative assessment of the criteria (Osgood & Tannenbaum, 1955). For example, someone may hold many criteria (or beliefs) about relationships, but not every criteria is important. The attitude about relationships is the sum of the beliefs weighted by their importance. By weighing criteria positively and negatively with a rank-value, the produced attitudinal measure improves the ability to predict the behaviors associated with finding a partner, because it is not only descriptive of the criteria involved, but also incorporates an estimation of the importance of each criteria (whereas beliefs do not).
In the current study, the criteria about dating will be inspired from the pragma love instrument (Hendrick & Hendrick, 1986) on the one hand. Then, interviews will provide the criteria about social networks and perceived difficulties. Three attitudinal measures will be extracted, (a) from the pragma love instrument (i.e., the “pragma love attitude toward dating,” simply labelled pragmatism), (b) from social networks (i.e., the “attitude toward one’s social networks regarding one’s dating,” simply labelled social norms), and (c) from perceived difficulties (i.e., the “attitude about barriers to dating,” simply labelled dating barriers), by weighing each criteria with the importance given to it by respondents. Other measures which do not directly relate to the behavior of dating (e.g., family roles and control variables) are not conceptualized as attitudes.
44
Social norms (Table A.5 in Appendix). Based on interviews with Northeast Asian young adults (Chapter Three), I selected five items to measure participants’ perceived influence from their mother, father, friends, siblings and society in general, regarding their dating (α = .83)22. For example: “My father thinks I should have a partner.” Then, I measured if participants value these opinions to obtain social norms (Fishbein & Ajzen, 1975). Five items measured the importance given by participants to the opinion of their mother, father, friends, siblings, and society in general. For example: “My father’s opinion is important to me.” The answers were rated as -2 = strongly disagree, -1 = disagree, 1 = agree, 2 = strongly agree, and I don’t know, which was treated as missing value23. The perceived network influence was then multiplied by the importance given to that opinion and combined across targets, e.g., mother, father, siblings, friends, and society (α = .80). The measure is referred to as “social norms.”
Pragmatism (Table A.5 in Appendix). The attitudinal measure for pragma love was derived from the pragma love component of the Love Attitude Scale24 (Hendrick & Hendrick, 1986) which includes six items (α = .77). A sample item is “When I choose a partner, I consider if he fits in my family.” Participants were instructed to think of their current partners when responding. Participants not currently dating were asked to recall a previous relationship or to imagine an ideal partner when responding to the question (Hendrick & Hendrick, 1986). Then I queried the importance participants gave to each pragma love item (rated on the -2 +2 Likert response format). For example “I find it important to choose a partner who fits in my family.”
Finally, the pragma love items were multiplied by the importance given to each item and combined (α = .77). This measure is referred to as “pragmatism.”
Gender-unequal family roles (Table A.5 in Appendix). The family role scale was adopted from the International Social Survey Programme (ISSP, 2016) and included twelve items inquiring about participants’ attitudes about family roles in family and marriage (α = .82). An
22 Cronbach’s alpha indicates the internal reliability of a scale (α > .7, for a reliable scale). Cronbach’s alpha by country and gender are provided in Table A.5 (Appendix).
23 The total number of “I don’t know” answers represented 0.0% to 26.5% per variable, and were missing at random (Little’s MCAR test chi-square non-significant, p = .267).
24 The scale includes six items from the Love Attitude Scale (Hendrick & Hendrick, 1986), that describe a pragmatic love style: “Before the relationship becomes serious, I must consider my future,” “When I choose a partner, I confirm that he/she fits my life plan,” “When I choose a partner, I confirm that he/she fits into my family,” “When I choose a partner, I confirm that he/she would be a good parent,” “ My partner positively influences my career,” “When I choose a partner, I check first if our life conditions match.”
45
example item is “A man's job is to earn money; a woman's job is to look after the home and family.” Since the measure does not directly relate to dating, it is not an attitudinal construct.
Dating status and number of past partners (Table A.4 in Appendix). Participants reported their current relationship status (i.e. single or dating) as well as the number of past dating partners.
Singles’ dating process (Table A.4 in Appendix). The dating process was modeled as four steps: Singles were queried in terms of their dating intentions on a 1 - 4 Likert response format (1 = strongly disagree, 2 = disagree, 3 = agree, and 4 = strongly agree, with I don’t know treated as missing value25), as well as about their monthly frequency of meeting new people, and meeting prospects. The last step concerns single women only and inquires if they are pursued or not.
Single women not being pursued and not finding “Mister Right” (Table A.4 in Appendix).
In order to understand the difficulty particular to single women, they were queried if they are single because they are not pursued, and if they are single because they have not found “Mister Right.” Both questions were rated on a 1 - 5 Likert response format.
Singles’ dating barriers because of the opposite sex (Table A.5 in Appendix). The scale was built from interview results and was composed of seven items describing how the behavior of the opposite sex may limit participant’s ability to find a partner. The scale used four items common to both genders such as “Men are immature and irresponsible,” and “Women are immature and irresponsible.” Then, the scale also included three gender-specific items such as, for male participants “Women care if a future partner will be rich.” Answers were rated on a 1 - 5 Likert response format. Reliability was calculated with the seven items’ measure, for each gender respectively: αwomen = .72; αmen = .62. Then I asked participants to report the importance they give to each barrier item (rated on the -2 +2 Likert response format), and multiplied the two measures to obtain the attitude about this dating barrier because of the opposite sex (αwomen
= .69; αmen = .60)26. The measure is referred to as “opposite sex barrier.”
25 In total “I don’t know” represented 22% of the total responses.
26 Although reliability for these measures were relatively low (α < .7), the two constructs will be tested for construct equivalence and modified accordingly.
46
Singles’ Time barrier (Table A.5 in Appendix). Two items inquired if having time to meet new people, and to find a partner may be an issue in dating. Answers were rated on a 1 - 5 Likert response format. An example item is “I don’t have the time to find a partner,” (α = .81).
Then I asked participants to report the importance they give to these two items (rated on the -2 +-2 Likert response format), and multiplied the two measures to obtain the attitude about this dating barrier because of the lack of time (α = .78). The measure is referred to as “time barrier.”
Education, attractiveness and socioeconomic status (Table A.3 in Appendix). The experience with intimate relationships may quickly evolve with age for young adults. Education, looks and wealth may promote, or impair the ability to date. Thus, besides providing their age, participants indicated their level of education (1 = high school or lower, 2 = bachelor’s, 3 = master’s, and 4 = doctorate or higher), their self-rated physical attractiveness (1 = not attractive, 2 = not very attractive, 3 = average, 4 = attractive, 5 = very attractive), and their socioeconomic status (SES; 1 = poor, 2 = rather poor, 3 = middle class, 4 = rather rich, 5 = rich). These are used as control (or covariate) variables.
Attachment styles (Table A.3 in Appendix). Relationship formation is strongly associated to attachment styles (Shaver & Hazan, 1987), therefore the levels of avoidant and anxious attachment27 were measured with the anxious-ambivalent vignettes extracted from the Adult Attachment Types (Shaver & Hazan, 1987). Answers were provided in the 1 - 5 Likert response format. Attachment styles are used as control (or covariate) variables.