As I interpret this question, the researcher has combined all the data between 2003 and 2011. There was no survey in 2003. The continuous survey started in the last 2 weeks of 2003 but the data are identified as 2004 as is the report.

Secondly, If everyone (say a survey with 20,000 women) gives a number of ideal number of children of either sex then 0 boys and 0 girls would apply to everyone but 20,000 women could have a number other than 0. If all wanted 2 children of either sex, then 20,000 would have a value of 2 for v629 and v613 but 0 for v627 and v628. Therefore, no inconsistency.