Surveying, although the most familiar method of research, seems to have its fair share of flaws. I wonder if that even with the use of a pilot survey, would the questions' wording and order can really be free of some bias. One concept that seemed a bit confusing was that of stratified sampling. If a sample is chosen at a truly random sample, wouldn't the chances of varying groups to be represented already fall into place? For instance, if I was to sample all psychology undergraduates, wouldn't I likely end up with all minority, disciplines, and/or any other variable anyways?
When choosing the sequence of questions, would it be too inefficient and costly to possible administer the survey in a fashion close to the Latin Square method? If each question had an equal chance of order and placement, would that even out the effect of sequence?
No comments:
Post a Comment