Methodological Considerations in Couples' Fertility Intentions: Missing Men and the Viability of Women's Proxy Reports.

Introduction Recent efforts show potential to advance research on unintended childbearing by taking a couple-level approach. However, this work has neither adequately addressed methodological concerns stemming from the challenges associated with male fertility data nor considered the viability of women's proxy reports of fathers intentions. Methods Data from the Early Childhood Longitudinal Study, Birth Cohort (ECLS-B) were used to assess the implications of low response rates among men on couples' unintended childbearing. Then, the accuracy of women's proxy reports of fathers intentions was assessed. Weighted logistic regression analyses were conducted to determine how women's characteristics were associated with men's survey participation whereas weighted multinomial logistic regression analyses were applied to determine how women's characteristics were linked with the accuracy of her proxy report. Results Almost half (46%) of women cannot be matched with data from the child's father, and this discrepancy is most problematic for women who are black, foreign-born, less educated, and unmarried at birth. Women's proxy reports appear viable as 75% of women's reports are consistent with men's responses. Yet, proxy reports underestimate disagreement in couples' intentions as mothers who intended the birth are at an increased risk of "inaccurately" reporting that fathers share their intentions. Discussion Direct approaches to couples' intentions yield privileged samples and systematically omit women at the greatest risk of an unintended birth. However, proxies underestimate disagreement in couples' intentions-a key contribution to the couple approach. Accordingly each approach has its own merits which must be considered in light of specified research questions.