Assessing quality of the questionnaire’s results


New Member
Dear all,

Consider the case that I issue a questionnaire to 100 subjects (e.g., some personality traits). I suspect that 10 of my subjects did not fill the questionnaire seriously, so I would like to exclude them. How can I statistically show that the quality of my results is better if I analyze 90 subjects instead of 100? Internal consistency seems to be about something else. Will intraclass correlation (i.e., agreement between raters) will work in my case?

Thank you,