Abstract
OBJECTIVES: To compare the accuracy of averaged scores from the original Norbeck Social Support Questionnaire (NSSQ)and averaged scores from each of three new NSSQ versions (NSSQ-R.aid, NSSQ-R.n/a, and NSSQ-R.format). These three new versions of the widely used NSSQ were developed to address three previously identified concerns regarding score accuracy: the Aid subscale's examples of aid, lack of an n/a response option, and the network nomination/rating procedure. Missing data rates were also assessed. METHODS: A convenience sample (N = 223) completed one of the four NSSQ versions. Score accuracy (restriction) was assessed by size of correlation between averaged scores (averaged score/network size) and network size, with low correlations indicating less score restriction and higher score accuracy. Fisher's r-to-z transformations assessed the significance of the difference between all correlations from the three versions. Missing data rates were assessed using chi-square tests of independence. RESULTS: The cumulative effects of removing the aid examples and use of the n/a response option improved score accuracy; averaged Aid scores from the NSSQ-R.n/a were statistically significantly less restricted than corresponding scores on the original NSSQ. The final version (NSSQ-R.format) actually resulted in statistically significant decreased score accuracy for averaged Affect scores. There were no statistically significant differences in missing data rates among versions. CONCLUSION: Averaged scores from the NSSQ-R.n/a should be used. Future research should focus on the use of situation-specific Aid items.