Sources of Errors in Survey
1. I wouldn’t anticipate that the proportion of “yes” responses in the same for the two group. Because the survey starts from the random samples of adult Amer
icans, the answer couldn’t be the same, everybody has different options, then the answer wouldn’t be quiet the same. 2. I would expect the proportion declaring agreement with the policy might differ between the two groups. Because for each group there has same number of student, and one for smoking students, another one for non-smoking students. Then for this question, the answer should be totally different. 3. I wouldn’t expect the proportion of: yes” responses to be close to the actual proportion of married people in the community who have engaged in extramarital sex. Becausefor the face- to- face interview people sometimes don’t tell the truth. Then at the end the actual proportion cannot be close to the “yes” responses. 4. I wouldn’t be surprised if the proportions of people expressing approval differed between these two groups. Because is the random samples of Americans, the answers are different shows that everyone has he/she own option. And as what I thought that the answers are not different, then that will be the problem.
1. Surveys were mailed to 19,000 California doctors, but just 2000 completed surveys were returned. The first paragraph said “more than half of CA doctors……” that’s not true. The reason why they didn’t get enough people to get the responds, that’s because they were mailing survey. This is Non- Responds Bias. 2. Yes, there are some individuals in the population will not have chance of being included in the sample. Just think about how many teenagers out there in the society are not in school. So this is the Undercoverage Bias. 3. This is Undercoverage Bias, because the survey is through the internet, but back to that time period, not everyone uses internet, just upper-income, highly educated, white and urban to produce result. But those cannot represent whole country nations. I think today, if we do internet polling is much more accurate than 19990s. 4. This is Non-Responds Bias, this will influence the result for this study, because not everyone responds to this question, and I think for some people who do gambling, they probably will lie or just not to reply the mail. So the answer well could be not accurate.