Test-retest reliability analysis on world health survey-China pilot study

  Objective   To assess the reliability of world health survey(WHS)questionnaires being applied in a Chinese environment.

  Methods   Cohen's Kappa, concordance correlation coefficient and intraclass correlation coefficient were used to calculate the test-retest reliability of categorical, ranked and quantitative measurements.ANOVA method was used to compare the testretest reliabilities of objective items with of subjective items.

  Results   High reliability was found in objective items and in selfreliabilities of objective items, and the items with a Kappa larger than 0.9 in these two categories were more than 90% and 80% separately.The confidence intervals of reliability of objective items' selfreported questions and vignette questions were(0.9262, 0.9617), (0.7742, 0.8178)and(0.6912, 0.7333)separately, which showed a statistical significant difference.

  Conclusion   WHS questionnaire shows a high dependability in collecting reality data, but is reliability need to be improved in vignette questions.