ABSTRACT
In the previous topics in this part of the book, various methods of drawing samples were described,1 with an emphasis on the importance of considering bias when evaluating the adequacy of a sample. Sample size is an important but secondary consideration. To understand why sample size is secondary to bias, consider this example: Suppose a student is conducting a survey on whether the main cafeteria on campus should remain open during evening hours. Being a commuter with only day classes, the student goes to the cafeteria at lunchtime and asks every 10th student who enters to participate in the survey. Of the 100 participants in the survey, 80% have no opinion and 20% want evening hours. After considering the results, the student decides that he or she should have used a larger sample, so the student obtains another 100 participants in the same way (asking every 10th student entering the cafeteria at lunchtime). This time, the student gets 85% with no opinion and 15% who want evening hours. Being very cautious, the student samples again, and this time obtains a 75%– 25% split. Combining results from the three surveys, the student obtains the total results shown in the bottom row of this table:
No Opinion Want Evening
Hours Sample 1 80% 20% Sample 2 85% 15% Sample 3 75% 25% Total 80% 20% Notice that for all practical purposes, the three
results are the same. That is, only a small minority want evening hours. With a total sample size of 300, the student might feel rather comfortable that he or she has pinned down an answer close to the truth. However, there is a serious problem in the method of sampling, which results from the fact that each time the student sampled, he or she sampled only from those entering the cafeteria at lunchtime. Thus, the sample is biased against those who are not on campus during lunch hours. Specifically, it is biased against evening students, who are the members of the population most likely to want eve1 The topics on sampling in this part of the book emphasize sampling in quantitative research. For sampling in qualitative research, see Topics 68 and 69.
