Foreach question

[alex badalyan]

Hi, I have the following data structure (although more data points). How can I instruct Stata to replace x with 1 for wave 2, if the data has the same id? Can I do it with foreach?

[Mike Lacy]

"If the data has the same id" is not meaningful to me. Perhaps you mean "if there is another observation that has the same id value?" I see nothing to make me think that -foreach- would be relevant, which is used to loop over values, variables, or the content of locals. The solution to your problem likely involves something like:

replace x = 1 if (wave == 2) & (some other condition involving "same id")

[alex badalyan]

"If the data has the same id" is not meaningful to me. Perhaps you mean "if there is another observation that has the same id value?" I see nothing to make me think that -foreach- would be relevant, which is used to loop over values, variables, or the content of locals. The solution to your problem likely involves something like:

replace x = 1 if (wave == 2) & (some other condition involving "same id")

Apologies, yes, if the id value is the same! I want to replace x=1 if wave ==2 and id value in wave 2 is the same as id value in wave 1. Essentially copy the x from wave 1 to wave 2 for the same id.

[Daniel Bela]

Apologies, yes, if the id value is the same! I want to replace x=1 if wave ==2 and id value in wave 2 is the same as id value in wave 1. Essentially copy the x from wave 1 to wave 2 for the same id.

sort id wave
replace x=1 if (wave==2 & id==id[_n-1] & wave[_n-1]==1)

This mechanism is called explicit subscripting. See help subscripting.

Note that this does not, as mentioned in your latter post, copy values from wave 1 to wave 2; it sets x to the value 1 for all observations from wave 2 that feature a wave 1 predecessor. In order to copy the earlier value, you need to use explicit subscripting in the replace instruction as well:

sort id wave
replace x=x[_n-1] if (wave==2 & id==id[_n-1] & wave[_n-1]==1)

There are more elegant ways to do this, quite likely involving declaring the data to be panel data with tsset, and using time series notation to insert lagged values, but you don't tell us enough about your data layout nor the result you try to achieve to give you more precise hints.

Regards
Bela

[alex badalyan]

sort id wave
replace x=1 if (wave==2 & id==id[_n-1] & wave[_n-1]==1)

This mechanism is called explicit subscripting. See help subscripting.

Note that this does not, as mentioned in your latter post, copy values from wave 1 to wave 2; it sets x to the value 1 for all observations from wave 2 that feature a wave 1 predecessor. In order to copy the earlier value, you need to use explicit subscripting in the replace instruction as well:

sort id wave
replace x=x[_n-1] if (wave==2 & id==id[_n-1] & wave[_n-1]==1)

There are more elegant ways to do this, quite likely involving declaring the data to be panel data with tsset, and using time series notation to insert lagged values, but you don't tell us enough about your data layout nor the result you try to achieve to give you more precise hints.

Regards
Bela

Amazing, thank you! This was what I was looking for.

Actually this is what it is for my case:

 sort id wave

replace x=1 if (wave==2 & id==id[_n-1] & x[_n-1]==1)