Explaining the Explanations of 100 Million Missing Women - Mechanism and Causality in Biology and Economics - page 222

not sustainable. As for β 3 is

very large and significant; these results together indicate that variation in the sex
ratio at birth mainly affects third and later children and thus suggests that the
cultural explanation is more plausible.
To test the other part of the further hypothesis—that the effect of HBV varies
with the sex of previous children—Lin and Luoh ran another regression as follows:

β 2 and β 3 , the value of β 2 is small and insignificant, and

Boy ¼ α þ β 1 HBsAg

þ β 2 First two children are girls ð Þ

þ β 3 HBsAg ð First two children are girls Þ

þ β 4 Mother Age Dummies ð Þ

þ β 5 Child Birth Year Dummies ð Þ

þ β 6 Birth Township Dummies ð Þ

þ ε ð 12 2 Þ

:

Again, following the same logic of identifying whether the effect of HBV varies
with birth order, Lin and Luoh checked whether the values of

β 4 were large
and significant; if they were, the result would indicate both that HBV indeed plays
an important role in determining the offspring sex ratio and that there is indeed
some interaction between HBV and the sex of the first two children. According to
Lin and Luoh's result, both

β 1 and

β 2 is very large
and significant. This result thus also suggests that the son-preference explanation is
more plausible. So this allows the authors to rule out the “complex biological
mechanisms” of HBV inflections on the sex ratio (Lin and Luoh 2008 , p. 2264).
Oster soon struck back. She teamed up with Blumberg (Blumberg and Oster
2007 ). They argue that, empirically, paternal, not maternal, hepatitis carrier status is
more strongly correlated with sex ratio at birth. Yet their 13-page manuscript was
never completed and published. Perhaps Oster found that the hypothesis was not
sustained; her recent 2010 paper, written with three Chinese medical officials and
researchers (Oster et al. 2010 ), claims that by analyzing the data of Haimen City in
Jiangsu Province, China (sample size ¼ 67,511 individuals), no relationship
between paternal HBV and the missing women is found. Consequently, she
conceded that HBV cannot explain the missing women in Asia.

β 1 and β 3 are small and insignificant, but

4 Difference in Methodology

The case of missing-women debate can be seen as an example of extrapolation in
the sense of Daniel Steel ( 2008 ; see also Steel's chapter in this volume, 2013 ).
Researchers compare the target with the sources originating from other disciplines
and geographic regions and draw similarities that could serve as a guide for further
investigation. Where the strategy of discovering the missing women is concerned,