Influence of positive and negative worded question on patients' answers to questions pertaining to quality of life

Background: Previous studies suggest that respondents may be more likely to disagree with a negative question than to agree with the positive one. The significant effect has not been observed consistently. We proposed to analyze differences in answers between positively and corresponding negatively worded questions and to identify socioeconomic and clinical factors associated with the differences. Methods: Data (N=532) were from the Louisiana baseline survey of the NIH funded MY-Health study. We analyzed answers of 5-point Likert Scale for 3 pairs of positively and negatively worded questions with age, sex, race, education, household income, surgery, chemotherapy, and radiation. Paired t-test, Chi-square, and multivariate logistic regressions were used. Results: Patients were more likely (P<0.0001) to disagree with negatively worded questions than to agree with the positively worded questions. Sex, race, education, and income were associated with differences in answers. Men were more likely than women to disagree with negatively worded questions (OR=2.84; 95% CI: 1.35-6.02) for the peaceful feeling question adjusting for education. For the purpose of life question, white and high income patients were more likely to disagree with negatively worded questions compared with their counterparts (whites/blacks: OR=3.15, 95% CI: 1.33-7.45; high/low income OR=3.39, 95% CI: 1.31-8.77). Conclusions: The patients' answers to the positively and corresponding negatively worded questions were significantly different, and some socioeconomic factors had relationship with the differences.