Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down at the end of May, and its replacement, is already up and running.

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: RE: Create variable with historic means

From   Nick Cox <>
To   "''" <>
Subject   st: RE: Create variable with historic means
Date   Tue, 16 Aug 2011 18:05:26 +0100

Your bottom line implies surprise, but in general -if- automatically ignores excluded observations, both in using data and in assigning results. 

Also, at first sight -egen- is fairly hopeless for problems that somewhere involve leads or lags. But it may be that this will do what you want. 

by zone wave mode, sort: egen hckr = mean(catch/(year<2006)) 

There are two tricks here, one standard and one non-standard. 

1. -egen, mean()- will work on expressions, which can be (much) more complicated than single variable names. 

2. The expression here is 

catch / (year < 2006)

The Boolean (year < 2006) evaluates as 1 or 0, so the whole evaluates as 

catch if year < 2006
.     if year == 2006

So the effect is to ignore 2006, but nevertheless to assign the result to all years including 2006. 

Another way to do it is 

by zone wave mode, sort: egen hckr = mean(cond(year<2006), catch, .) 

but I am fond of the first way. 

For a write-up, see 

SJ-11-2 dm0055  . . . . . . . . . . . . . .  Speaking Stata: Compared with ...
        . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  N. J. Cox
        Q2/11   SJ 11(2):305--314                                (no commands)
        reviews techniques for relating values to values in other



I'm trying to create a new variable that contains the historical means of
CATCH, but only using the previous years.  My dataset looks something like

2004       2           1           1          1
2005       3           2           3          2
...          ...          ...          ...        ...

In this case I want to create a new variable that will take the value of the
mean catch for the previous years (2004 and 2005) by ZONE, WAVE and MODE,
but not use any of the 2006 observations to calculate the means.  Its OK if
the variable is created for the previous years as well, as I plan to drop
those observations once I can get the new variable.

I have been able to create a table that gives me the mean of CATCH by ZONE,
WAVE and MODE using:

table zone wave mode if (year<2006), contents (m catch) 

And I can create a variable that has the means, using 2006 data, with:

by zone wave mode, sort: egen hckr = mean(catch)

When I try to create the variable using only the previous years, I have

by zone wave mode, sort: egen hckr = mean(catch) if(year<2006)

But it just gives me missing values for all observations from year 2006.

*   For searches and help try:

© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index