Copy value for a group

[Bene Eil]

Hey everybody,

I have a dataset with funds and fund families. 2-7 funds belong to 1 fund families,

First, I have to compute deciles (dens_perc) for the fund families at each time point.

In the second step, I want to copy the deciles to every fund of the fund family at each time point.

mgmtquarter is a identifier for each fund family and time point (egen mgmtquarter = group (fund time)).

I found that command:
Syntax: bysort mgmtquarter(dens_perc): replace dens_perc = dens_perc[1]

But I got the error message:
factor variables and time-series operators not allowed
r(101);

Thank you for your advices

Best wishes

[Clyde Schechter]

You need a blank space between mgmtquarter and (dens_perc). The error message is, admittedly, not very informative. It's Stata's syntax parser's best guess as to what you were doing wrong--but not a very good guess in this case.

[Bene Eil]

Is there also a possiblity to adjust the code: bysort mgmtquarter (dens_perc): replace dens_perc = dens_perc[1].

Every observation in my group has a value, no missing values.

But I always want to replace all observations of the group with the seventh oberservation.
So, I need to replace all values of the group with the value of exact one observation.


Any suggestions?

Right now, I set all values - except the one I need - equal to zero and then create a new variable like: bysort id_group: egen value = max (old_group)
But this is not very intuitive...

[Nick Cox]

You can use 7 as a subscript just as you used 1. Clearly, observations must be in the desired sort order.

[Bene Eil]

I tried it. But the problem is id_group is not sufficient to sort the data correctly.

I need to sort the data with id_group_time and then replace the observation within id_group with my seventh observation...

So, I have to sort the data at first and the add the "by" function within one command.

I tried something like: sort id_group_time: egen value = max (old_group), by id_group. But that is wrong!

[Nick Cox]

Your example code, defining a maximum, bears no apparent relation to your question about using the 7th value.

You may want something like

bysort idgroup (time) : gen whatever2 = whatever[7]

Note that I am using generate here. Using replace when you are not clear what code you should be using is likely just to mess up your dataset.

Please do read http://www.statalist.org/forums/help for advising on showing data examples, the code you actually used, etc., etc.