Extending the Theory of Normative Social Behavior to Predict Hand-Washing among Koreans

ABSTRACT

The current study tests the predictions of the theory of normative social behavior (TNSB) in a hand-washing context in a Korean sample and extends the theory to examine the role of perceived publicness, a variable believed to activate face concerns, as a moderator of the norm–behavior relationship. The findings show substantial main effects for all of the study variables on behavior. In addition, the descriptive norm–behavior relationship is moderated by perceived publicness and outcome expectations, but the nature of the interactions is not consistent with that evidenced in previous literature on US samples. Implications for normative theory and communication campaigns are discussed.

Notes

¹ According to the Korea census data, 17% of Koreans have white-collar occupations (Statistics Korea, 2015), 20% live in Seoul (Ministry of the Interior, 2015), and 41% have a college/university degree (Korea National Statistical Office, 2016).

² After the translations were completed, a focus group was conducted with five Korean students attending a university in the United States. The primary purpose of the focus group was to provide evidence for the face and construct validity of the instruments. Minor modifications in wording were made in the items based on the focus group participants’ suggestions and critiques. An original version of the survey questionnaire was pilot tested with a sample of 32 Koreans recruited by the survey company. The participants were gathered from the same pool as those in the final survey. They completed the initial version of the online survey and provided informal feedback via an open-ended question that solicited comments and suggestions at the end of the survey. The pilot test results were used to identify items that were difficult to answer or worded inappropriately, survey design, and to examine initial inter-item correlations for the scaled items. The wording of several items was modified as a result of the pilot; the data from the pilot were not included in the final analysis.

³ The correlation coefficient between descriptive norms and injunctive norms is relatively high compared to others, r = .70, which could inflate the standard errors in interpretation of the study results. This study tried to address this possible multicollinearity issue by 1) adopting established measurement items; 2) conducting a series of confirmatory factor analyses using the internal consistency and parallelism theorems (Hunter & Gerbing, 1982); and 3) mean-centering the predictors, following the recommendation from Aiken and West (1991). In fact, despite its high correlation coefficient, when these variables and the interaction between the two norms are entered in the regression equation, the variance inflation factors (VIF) are 1.114 (descriptive norms), 1.106 (injunctive norms), and 1.001 (descriptive norms $\times$ injunctive norms). Given that a VIF of 10 or higher is usually considered as serious multicollinearity (O’Brien, 2007), thus, we would argue that multicollinearity is not a significant issue for this study.

⁴ A series of one-way analysis of variance (ANOVA) tests were conducted for detecting the main effect for the other demographic variables on the reported hand-washing behavior. The analyses showed that there were no significant effects for residential areas and education level on the dependent variable. However, the data indicated that participant occupation type yielded significant differences in reported hand-washing, F(7, 780) = 8.49, p < .001, partial η² = .07. Workers in the agriculture, forestry, or fisheries industries reported the highest frequencies of hand-washing, followed by housewives, students, and white-collar workers. The occupation type then was dummy coded based on the median value of the dependent variable (i.e., the occupations that had mean values above the median were categorized into a high group, and those that had mean values below the median were categorized into a low group). Regression models were run with this dummy-coded variable included as a covariate and it did not influence the nature of the results so it is not reported further (the data are available from the first author). In addition, a significant association between age and the reported hand-washing (r = –.12, p < .01) and a main effect for income on the dependent variable were also found, F(5, 782) = 2.95, p < .05, partial η² = .02. Nonetheless, since these demographic variables had minimal influence on hand-washing behaviors, age and income were not included as covariates in the models.

Additional information

Funding

This research was supported with fellowship funding from the College of Communication Arts & Sciences and the Graduate School of Michigan State University. Partial support for Lapinski’s time was provided by the USDA National Institute of Food and Agriculture (Hatch project number MICL02244) and by a National Science Foundation (Award No. SMA-1328503).