The last time I taught this course to undergrads, I used data from the General S

Question

The last time I taught this course to undergrads, I used data from the General Social Survey. Despite the fact that all students were either political science or PPL majors, a full one fifth of them decided to write their research paper on the sociologic subject of happy marriages. They hypothesized that straying during one’s marriage causes a person to have less than a very happy marriage. A look at the evidence on the “ever strayed” and “happy marriage” questions showed (something like) two-thirds of the people who have not strayed during their marriage report having very happy marriages versus half of the people who have strayed during their marriage report having very happy marriages. As this is a statistically significant difference, the students concluded that their hypothesis is confirmed: straying during a marriage causes one’s marriage to be less than very happy. a. Identify the independent variable and the dependent variable. b. Define measurement validity and measurement reliability. c. Discuss why you might have doubts about the measurement validity of the independent variable. d. Discuss why you might have doubts about the measurement reliability of the dependent variable. e. Discuss why you might have doubts about the causal inference that says straying during a marriage causes one’s marriage to be less than very happy.

Explanation / Answer

a. The objective here is to identify whether straying during a marriage causes one’s marriage to be less than very happy or not. Therefore, the independent variable here is whether a person has strayed during their marriage or not (it seems like, in the problem they defined this as a dichotomous (Yes or No) variable), Also, the dependent variable here is whether a person is having a very happy marriage life or not (Here also, it seems like this is defined as dichotoous variable).

b. Validity tells you whether the scale, which you are using to measure something, is measuring what it is supposed to measure. For example, if you want to measure the height of a person, you cannot use a clock to measure that.

Reliability tells you whether the scale you have developed is consistent or not, i.e. whether it is giving similar result for similar respondent or not. For example, you want to measure weight, and you have a weighing machine for that, Now, in the morning your weight came to be 100 lbs, afternoon 160 lbs, and night it came to be 80 lbs. This much deviation for same respondent confirms inconsistency, therefore, the tool is not reliable.

c. Here, your independent variable is whether the respondent has strayed during his/her marriage or not. However, the answer to this question may be interpreted to be reflective of the respondent's integrety and sincerity. Therefore, the respondent may have a tendency to give a "socially desirable answer". This bias in responses is known as "Social Desirability Bias". To counter this bias one needs to be covert in their approach while asking the question. They may ask the question indirectly or make the respondent comfortable first to get a more candid response. Therefore, if those steps are not taken, this bias will be there in the answer, and the measurement will not be valid.

d. Here the dependent variable is whether the respondent is happy with his/her marriage or not. Now, the objective here is a person's mood, which is generally very volatile. A person's mood depends on many things. Assume that a person just had a fight with his/her partner. Therefore, if you ask that respondent whether his/her marriage life is happy or not, and though the person has led generally happy married life, he might respond negatively. Therefore, there is no guarantee that the results that is obtained is reliable. However, if you can capture different aspects of moods of person, and from there you can segregate the portion reflecting the married life, then probably the easurement would be reliable.

e. First, statistics is just a tool. Therefore, statistics can be used to triangulate a study. However, the relationships in a study must come either from theory or logic.

Second, statistical tools can never predict causation. Let us say there are two variables, gender and political awareness, and you want to find out the relationship between two variables. Let us assume, you identified a statistically significant relationship. Now, you may argue that political awareness is dependent on gender. However, some may argue that gender is dependent on political awareness. You refute their claim by saying, how can gender be a dependent variable, as it is an inherent characteristic of the respondents. However, in cases where, this distinction is not possible, how can you conclude whether X causes Y or Y causes X. For example how can you conclude that straying during a marriage causes one’s marriage to be less than very happy, when the reverse, i.e. straying during a marriage depends on whether one’s marriage is less than very happy, can equally be true.

There might be case of "spurious correlation" as well.

Related Questions

Navigate

• Browse (All)
• Browse T
• Subjects

Integrity-first tutoring: explanations and feedback only — we do not complete graded work. Learn more.