Error in 2010 Indonesian classwk variable

I am looking at this variable for only children 17 and under. I find it strange that for 2000 and 2005 only 7-8% of children are working but in 2010 over 50% of them are working, and what’s more, 96% of them are working in self-employment w/ temporary workers. I find that hard to beleive.

I read the questionnaire and can’t really pinpoint any discrepancies between the three years in how the employment questions were asked. Is it a possible survey error?

Based on the questionnaire, the 2010 Indonesia CLASSWK variable has a slight change in Universe. In 2000 and 2005, only those persons working in the previous week were asked the CLASSWK question. In 2010, those on “temporary leave” in the previous week were also asked the CLASSWK question. As a result, the 2010 version includes unemployed persons and is not comparable to earlier versions. However, recoding those persons in 2010 with an EMPSTAT value not equal to “1:Employed” as CLASSWK=“0:NIU” will make CLASSWK consistent with 2000 and 2005 both in terms of the Universe and the percentage of sample children working (approx. 4%).

Hope this helps.