The Structure of Son Bias in Armenia: From Implicit Associations to Explicit Behavior

Recent demographic analyses have shown that Armenia has a sex ratio at birth significantly favoring boys, a pattern indicating that Armenian families preferentially abort females. The preferences supporting this behavior have not been systematically studied, and in particular little is known about the structure of son-biased preferences within households, across families, and between different regions of the country. We collected data in three regions of Armenia to study son bias in terms of both fertility preferences and outcomes. To overcome measurement biases we developed a number of new methods to elicit preferences and fertility histories. We also used our methods to collect data from three key individuals within households, namely the wife, the husband, and the husband's mother. We find that the largest differences in son bias occur within households, with husbands tending to have markedly stronger son biases than their wives and, by some measures, their mothers. In contrast to our expectations, we find only small differences between regions. In addition, stated fertility preferences do not simply favor sons over daughters. Armenian families also express a strong preference for families with a mixed gender composition. These two types of preference, however, affect family planning in very different ways. While parents who have only daughters actively seek to manipulate the probability of having a son, we find no evidence of analogous efforts among parents who have only sons. Our data also allow us to address the possibility that recent awareness campaigns have increased the risk of families under-reporting their son bias. Our findings suggest that social desirability biases should indeed be a concern among organizations trying to evaluate efforts to attenuate son bias and improve the status of girls.