SOC341 -- Homework #4 Feedback

SOC 341-01/51 (Kuechler)
Fall 1997

Homework Assignment #4 - Feedback

What you were supposed to do:

1. Do chapter exercise #3 for chapter 5 of the textbook (p.185). However, you do not need to look at the findings. Rather, critique the sampling based on the sampling strategy as summarized in the textbook. No need to get hold of a copy of the actual publication.

2. [12 points]
Do the multiple choice questions #4 to #9 for chapter 2 in the Study Guide (p.29/30). Submit this page from your copy of the Study Guide. In addition to marking the correct answer, provide a brief explanation in the margins of the page.

How you should have done it:

1. You needed to address the key question (stated in the problem): "Is Hite's sample likely to represent American women in general?"

Hite distributed 100,000 questionnaires to church groups and other organizations and received back 4.5% or 4,500 questionnaires --
not all answering all 127 questions.

The size of the actual sample is quite large, there are very few studies that would have 4,500 completed questionnaires. However, the response rate is very low. In addition, this was not a random sample: not every American woman had the same chance to be selected into the sample. Without a random sample (following the lottery principle), there is no sound base to generalize the findings from the sample to the larger population. Especially, when participation is totally left to the initiative of the individual coming in contact with the questionnaire. So, the sampling strategy is seriously flawed in more than one way:

Not every American woman had the same chance of even seeing the questionnaire, learning of its existence. Apparently, women in church group had a much better chance of coming in contact with it.
Even among those who came in contact with questionnaire, the chances are participation are not equal. Due to mail distribution, women had a chance to browse through the questionnaire first and then decide whether they wanted to participate. This results in a self-selection bias, people with a specific interest in the subject or people liking the way the questions were asked are more likely to participate.
In addition, due to the nature of the questions -- requiring verbatim responses rather than selecting among given choices -- a additional education bias is very likely. Women with less formal education, with lesser language skills are less likely to participate.

One might say that the women who participated in Hite's survey are as representative of all American women as are callers to talk radio shows. With 4,500 responses, Hite was certainly able to capture a significant sentiment among American women in qualitative terms, but all quantitative statements must be taken with great caution.

2. I had told you (incorrectly as it turned out) that the Study Guide does not have the answers to the multiple choice questions. Well, as one student has pointed out to me, the answers are there (on page 223), but it appears that this page was never proof-read. Three of the listed answers to the six questions (#4 through #9) are simply wrong. And a number of students were not fooled and relied on their understanding of the material. So, in a nutshell here are the correct answers with a brief explanation:

#4: d, because the dependent variable describes the effect ("level of perceived stress") which is reduced if the task is more interesting.
#5: d, because "to be or not to be a veteran" (veteran status) is the cause (independent variable) for patriotism. "Veterans" are just a group of people not a variable.
#6: b, because, the way the hypothesis is phrased, regularity of study affects (is the cause of) the GPA. One could think of a wording that would reverse the roles of independent and dependent variable like: "The higher a students' GPA, the more regular he/she studies". The mechanism being "committed to one's college work". But that's not the hypothesis was wording in the question.
#7: d, because "Texans" is not a variable. With a change in wording this could be turned into a hypothesis: "Texans are more likely to be very assertive than people from other parts in the US".
#8: a, because "shelter" is not a characteristic or trait that can take on different values or categories. "Ethnicity" can take on values like "African American", "Irish-American", "Polish-American", etc. "Conflict level" can be high, medium, or low. "State" can be NY, NJ, CT and 47 other things. "City size" can vary from a few thousands to 7.5 Million (like NYC).
#9: a, because that this exactly how "empirical generalization" is defined in the textbook, a summary of a pattern found in the data.

If you handed in this assignment, you will find the points for each assignment as well as a total of the points you have accumulated so far. This is stated in the form T = x/65 , where x is your actual point total. 65 is the current overall total, not counting the bonus problem (#2). Make sure to check whether your bookkeeping agrees with mine. If you think you have more points, please get in touch.

HW #5 will offer a chance to get another 18 bonus points. Be aware of more typos. Do not blindly trust the answers to the multiple choice questions stated on page 223.