Based on the data from 2017 national fertility survey, this paper analyzes patterns and determinants of fertility intention in China under the universal two-child policy. The survey suggests that womens average ideal number of children is 1.96 and the average number of intended children is 1.76. For women born in the 1990s and beyond, the number of children they intend to have is 1.70, while for women who have not yet had any children, their intended number of children is 1.60, both are lower than the current average, indicating that there is room for a further decline in the fertility in China. About 9.3 percent of women intend to have three or more children. Among women who have at least two children, the corresponding proportion of women whose first two are girls is 45.6 percent. The regional difference of the intended number of children is more apparent. Sex-selective abortion inhibits fertility intentions and reduces the number of birth, however, in the case of unsatisfactory sex composition of children and tendency to continue to have children, son preference plays a role in strengthening fertility intention and increasing the number of births.

This article introduces the process of the 2017 China Fertility Survey from the perspective of data quality control.The individual questionnaire includes four sections:fertility desire,fertility behavior,childbearing and its services,and the determinants of fertility behavior.The sampling method is the three-stage probabilities proportional to size (PPS).The survey is applied with Computer Assisted Personal Interviewing (CAPI),which plays an important role in the quality control of questionnaire designning,interviewers training,last-stage sampling,interviewing,and questionnaire review.After the on-site survey,the survey results are compared with the relevant registration data in the population and family planning data repository,and the inconsistent data are fed back to the local supervisors for verification.In order to correct the potential sample structure bias caused by some defects in the sampling frame,ex post facto weighting is applied based on relevant data.In the future,the sampling frame may still be subject to distortions because of population migration.Regarding these issues,this paper makes suggestions for improvement.

In preparing for fertility policy adjustment and implementing the “Two-child Policy for Couples One of Which Has No Siblings”，a national fertility intention survey was conducted in 29 provinces in China in August 2013. Survey results indicate that，currently，the ideal number of children is 1.93 in China. The reported ideal number of children in regions with 1 child，1.5 and 2 children policies are 1.84，1.98 and 2.01 respectively. For the couples both of which have no siblings，one of which has siblings and both of which have siblings，the reported ideal numbers of children are 1.79，1.83 and 1.95 respectively. For the couples one of which has no siblings with one child，the reported ideal number of children is 1.81. The overall sex ratio of reported ideal numbers of children is 104 males per 100 females in China. The sex ratio is 128 for those whose ideal number of children is 1 and 102 for those whose ideal number of children is 2. For the couples one of which has no siblings with one child，59% are“uncertain”about the birth timing of next child，20.5% plan to give a birth in one year，12.1% plan to give a birth in two years and 8.4% plan to have one in three years. Adjustment of family planning policy will help narrow the gap between fertility intention and fertility behavior，and it is highly unlikely that a nation-wide birth heaping would occur if the“Two-child Policy for Couples One of Which Has No Siblings”is well implemented.

This paper describes and discusses issues on the establishment of sampling frame，replicate selection and linking after selection，the relation between target population and frame population in migrants surveys. Experience shows that when preparing the sampling frame of migrants surveys，researchers
not only need the support of relevant database，but also should pay particular attention to the effects of accidental factors. For oversized or undersized sample units，a relatively simple approach is to replicate selection，or link sample units after selection. In addition，in a survey whose target population
is the“Registered-household Population”，the relation between target population and frame population in specific areas will appear as“one-to-many”as a result of impacts of cross-border migrant outflows.Thus，post-weighting is necessary when processing data. To ensure the effectiveness of the postweighting，the characteristics of population flows in different provinces should be taken into account，especially the size of floating population both across and within province in the survey sample. Data processing of 2013 National Fertility Desires Survey indicates that the grouping variables and number of groups of post-weighting in different provinces are not exactly the same，but subject to some flexibility.