Marital coital frequency: unnoticed outliers and unspecified interactions lead to erroneous conclusions.

The authors challenge Jassos conclusions that in 1970-75 net of cohort and age effects average marital coital frequency declined by 50% and net of period and cohort effects coital frequency increased with wifes age. Both these findings are inconsistent with previous research on sexual behavior. After reanalyzing the same data used by Jasso the authors came to the following conclusions: 1) when outliers are removed neither husbands nor wifes age is significantly related to frequency of intercourse although period and marital duration effects remain; and 2) when the model is respecified separately by duration of marriage the age duration and period effects disappear for couples married more than 2 years in 1970 (88% of the sample). The only Jasso finding that persisted in this reassessment of the data was the period effect and even this effect is limited to couples married for 2 or fewer years in 1970. In response Jasso argues that her estimates of the effects of spouses ages marital duration and contemporaneous period influences on marital coital frequency are superior to the outlier-deleting estimates. Jasso asserts that if the outlying observations in question are indeed erroneous than the estimates of boths ]enrich researchers are unbiased estimates of the underlying parameters. In contrast if the outlying observations are correct then only the Jasso estimates are unbiased; the Kahn and Udry estimates now reflect biases associated with truncation of the dependent variable.