Gauging the Validity of Responses to Questions on Family Size Preferences in China

Abstract

The article discusses the issue of the accuracy of survey data on family size preferences in China. Concern about poor sampling design non-response errors and response errors is heightened by Chinas firm policy on low fertility and the adverse repercussions when the policy is challenged. A comparison of 2 concurrent studies conducted on the same population in and around Shanghai indicates that Shanghai women underreport their desired family size in face-to-face interviews which were used in the China In-depth Fertility Survey (IDFS). It is suggested that this trend is nationwide. The Shanghai Survey of Desired Family Size (SSDFS) used anonymous self-administered mail-in or drop-off forms. The discrepancy in desired family size is greatest in the lowest education groups (the illiterate elementary-educated and junior high-educated) but for high school-educated the difference are insignificant. The 2 surveys also show urban and rural differences with rural women desiring one-half child more in the SSDFS. 32% of the variance in family size is accounted for in a multivariate equation by residence educational level and number of living children. Validity of either survey cannot be determined. There is speculation about why the non-response was highest among the poorly educated and more fertile women. This may reflect a lack of ability to complete the form or a higher desired family size. If average family size is determined by combining the educational distribution of IDFS with the mean desired family size of the SSDFS then the result of 2.29 children. This is one-half child greater than the 1.8 average from the IDFS. The implication is that the true mean is 25-30% higher than the face-to-face interview. Both surveys show an interest in low family size although not the government recommended 1 per family even without sanctions. In the SSDFS 2-child families expressed an interest in having the same or fewer children.