Week 6: Discussion Questions
Ruggles (1997)
1. Ruggles examines the relationship between local labor market characteristics and marital status. What does he find? What do his results tell us about the link between marital instability and women's labor market participation? What about the link between marital
instability and men's economic position?
2. Ruggles analysis only includes whites, because of insufficient sample sizes for blacks. He then applies the models that he estimated for whites to the national mean values for blacks on the independent variables. What are his results?
3. What are Oppenheimer and Preston's criticisms of this study? Do you find them convincing? How did Ruggles address these criticisms? Do you agree with his responses? Did he address their concerns?
4. Ruggles finds that higher labor force participation for women is associated with a higher prevalence of divorce/separation. Can we make causal inferences about his results? Can we ever make causal linkages when using cross-sectional data? Why or why not?
5. Ruggles argues that one problem with Oppenheimer's hypothesis about the relationship between men's economic position and marital
instability is that for most of the past century, opportunity for men was increasing, though divorce rates were increasing. However, Oppenheimer has found a strong relationship between men's economic position and union formation/dissolution in her work. How can these be reconciled?
6. Ruggles divides the independence hypothesis into two separate interpretations - the interdependence theory and the economic-opportunity interpretation. What is the difference between these two interpretations? Can we test each of these separately? How? What are some ways that we may do so?
7. Does Ruggles analysis imply that labor market variables have had the same effects on marital stability over the course of the last century?
8. "These results should not be interpreted as applying to changes in marraige patterns more generally. When I applied a similar model to predict the probability of being ever-married among young adults, the results were considerably weaker and less consistent than the models for divorce or sepration. This reinforces the conclusions of Oppenheimer (1994) and Mare and Winship (1991) that changing labor-market opportunities cannot fully explain recent declines in marriage rates. Marriage formation simply operates differently than
marital dissolution." (Ruggles 1997:464) Also, consider Oppenheimer's argument laid out in the first paragraph of p. 471. Let's speculate the theoretical implications of this discrepancy between results of marital dissolution and those of marriage formation.
Sayer and Bianchi (2000)
1. Why do we use measures of marriage dissolution that include both divorce and separation rather than simply using divorce?
2. Sayer and Bianchi are testing the independence hypothesis. Which interpretation of the independence hypothesis (using Ruggles distinction - see Q6 under Ruggles) are they testing?
3. Sayner and Bianchi include variables that measure gender ideology, marriage quality, and marital commitment in their equations that predict divorce/separation. What are these measures? What are their results?
4. How are their measures of marital quality/commitment related to other variables that have been found to affect union dissolution? For example, Waite and Lillard note that previous research has found lower ratings of marriage quality for couples who have children than for childless couples. Similarly, what characteristics may lead to reports of lower marriage quality/commitment?
5. Would we expect these variables to have the same effects on cohabitation dissolution?
Axinn and Thornton (1992)
1. To demonstrate the relationship between cohabitation on subsequent divorce, this article does not deal with divorce rates, but relies on the assumption that subjective/attitudinal measures are closely linked to the probability to divorce. According to our knowledge, to what extent is this strategy justified? Or, how much do we really know about the predictive power of various attitudinal measures on divorce rates?
2. The attitudinal variables have no effect on first union formation rates. What does this finding, in combination with the models treating
cohabitation and marriage as competing risks, tell us anything about the nature of cohabitation as an institution as opposed to marriage? That is, how would you explain that, regardless attitudes of their own or of their parents, people form unions around similar ages? The only difference is what kind of union people form.
3. Related to the 5th question on Sayer and Bianchi, let's brainstorm how we are going to study the dissolution of cohabitation unions. What are the important issues to keep in mind, and what are the best candidate independent variables? Let's generate some hypotheses, if possible.
4. It is a neat idea to use panel surveys to examine reciprocal effect. Can we think of other topics where this approach may also apply? For example, our fellow student Ann Meier has examined the relationship between religiosity and first sexual intercourse. Perhaps we can discuss analytic issues as well--i.e., how to study score change between surveys.
5. Are you surprised to see the negative association between marital stability and premarital cohabitation? Why and why not?
6. Axinn and Thornton find a relationship between attitudes about divorce and cohabitation (marriage) for both men an women, but only when attitudes about divorce are measured by a certain question, which is different for men than for women. It appears that these two questions are measuring different things. What are these questions actually measuring? Relatedly, it is a common finding that slight changes in question wording can dramatically affect the responses that a researcher receives. What implications does this have for research that includes measures of respondents' attitudes?
Waite and Lillard (1991)
1. This paper has empirically demonstrated a succinct picture of the relationships between children and marital disruption (read the paragraphs above the heading IMPLICATIONS on p. 949 and the paragraph following this same heading, for a quick grasp of this paper). How can we go about to test behavioral and/or motivational mechanisms that underlie the relationships between children and marital disruption? In other words, why people behave in the way that children are shown to be associated with marital disruption. Reflect on the "theoretical perspectives" and your own experiences and see if we can find sensible proxy variables and empirically examine these explanations more directly.
2. "South and Spitze (1986) find that the effects of these and similar variables do not change over the marital life course." (p. 938) - What is this sentence all about in this context? Let's talk about proportionality assumption if you wish.
3. "The risk of dissolution may change systematically over time as the marriage continues. The couple learns over time about the quality of the match, and the couple may acquire 'marriage-specific capital,' which makes the union more desirable so that the risk of divorce declines as the marriage endures. This change accounts for declining divorce probabilities with marital duration (Becker et al. 1977). These systematic changes in the risk of dissolution are reflected in the baseline hazard." (p. 941) In other words, we need to correctly specify the baseline hazard function. What if we make very bad assumptions (or approximations) of the baseline hazard function? This may relate back to the issue of how to use cross-sectional data to infer trends of "events," as in the exchange on the Ruggles's paper. We can talk about some details in Sayer and Bianchi's analytic strategy under this context too.
3a. An application to a substantive issue, if you have time to read the Wolfinger's (1999) paper in the optional readings, which is a very, very short paper! Allen can share some thoughts from his research with Larry Wu replicating Wolfinger's analysis -- in fact, their conclusion has general implications for studying trends without event history data. (A plea for help from Allen should you read Wolfinger's paper: if you can think of interesting theoretical arguments why there might *not* be a decline in trends of divorce transmission, please let him know. He would really appreciate your help on this matter.)
4. We will probably want to go back to the left-censoring/truncation issue. This is especially salient in panel studies. In addition, Tara may give us a brief introduction of the PSID data. A minor point: are all the numbers reported in this paper weighted or unweighted? Weighting seems to become complicated in analyzing multi-wave longitudinal data with substantial sample attrition. Maybe we can talk about this issue--"to weight or not to weight," and "how to weight if we decide to weight."