I want to construct a panel data set from the childrens’ perspective which includes all available information on their math and German marks. For children aged 0-10, this information is provided in the data set bioagel. However, I am unable to merge this information onto ppathl, my base file when merging. It seems that you need to use at least three identifiers for merging bioagel: pid (childrens’ id), syear and pide (id of adult respondent). But pide is missing in ppathl. Do you know a solution?
I forgot to mention that I am also interested in the bioagel variable „biochild“ which actually differs depending on the responding adult, so in that case pide seems to really matter.
in bioagel you find all parents over child questionnaires from the age of 1 year until 11 years. Information of the children themselve for higher ages you find in biopupil and jugendl. Sometimes both parents (pide) respond thes questionnaire in one year. Thus the dataset needs three identifiers: pid syear and pide. You have to solve the problem which information of the parents you want to keep and to get only one line for one children in one year or you have to merge 1:m with ppathl. I prefer the first way.
In bioagel you find information concerning marks only for children between 7 and 11 years old (bioage_h. Thus you can only keep these samples (bioage_h== 8, 101, 102) (see also bysort bioage_h: sum lamark matmark nomark lamarkno matmarkno). The problem is still the answers of both parents in one survey year. You have to decide which you want to keep (only valid information and those from the mothers or fathers) or you build 2 variables (matmark1 e.g. mother and matmark2 for the father or other way round).
If you solved this problem, reduce your sample to one line for children and syear (isid pid syear must be run without mention a mistake) and then you can merge with pid and syear as keys with ppathl.
To add further information you can use biopupil (lamark and matmark) and jugendl (jl0152 and jl0156). In these two datasets you have only one answer per child and year (isid pid syear is true). Here you don’t have to decide which information you want to keep. You can merge directly with pid and syear as keys with ppathl.
thank you so much for your detailed response! It has helped a lot.
I have decided to construct two variables for the two adult respondents. Here’s my code, maybe it’s uself to someone else:
foreach var of varlist biochild lamark {
clonevar mum = `var'
replace mum = . if sexresp == 1
bys pid syear: egen `var'_fresp = max( mum )
clonevar dad = `var'
replace dad = . if sexresp == 2
bys pid syear: egen `var'_mresp = max( dad )
label values `var'_fresp `var'_mresp `var'
drop `var' mum dad
}
Inspecting the information on childrens’ marks has however raised another question: Does the corresponding survey item always refer to the last annual report card of the child? Or can the item also refer to a mid-term report card, depending on the survey date and the age of the child? It seems to me that the latter is the case.
Dear Adrian,
The question is about the „last report card“ as you also find out. Thus I would also say it depends on the interview month if they refer to the „half-year report“ or the final year grade from the last school year.
