Deletes districts from all variables already read in by
commands YVOTE, YVOTE2, XVARS, XVARS2, and XNEW. This command works
by actually deleting the districts (or rows of your data) rather than
just setting them to missing values, so be careful. If you decide to
read in more data following a DELIF command (and before issuing a new
YVOTE statement to wipe out all data in memory), these data may not
have the same number of rows as data already in memory. If you do
read in additional data after a DELIF command, the new variables (and
dataset) must have the same districts as those remaining in memory
after deletion-the deleted districts must not be included in the new
data.
For example, suppose you have some variables measured for all districts in the
entire state but other variables measured for only the northern districts, and
you wish to perform some analyses for all districts and others for just the
northern part of the state. You could accommodate this situation in two ways:
- Put all the data in a single dataset, with southern districts coded as
missing for some variables. When you use only variables for which you have all
the data, every district will be included in the analysis; when you include at
least one variable with missing data in the southern districts, those rows
automatically will be left out of the analysis. If you have only a few
variables with missing data in the south, then this is the preferred option.
- If you have a lot of variables with missing data on the south, then
Option 1 would be tedious because you would have to type in a large number of
observations with only missing value codes. It also would waste disk space.
The alternative is to create two datasets, one with variables for which you
have all the observations, plus an extra variable coded 0 for northern
districts and 1 for southern districts. The second dataset contains variables
for which you have data on the northern districts only. To do an analysis with
variables based on all the districts, just use the first dataset. To use these
variables and also some variables in the second dataset in an analysis of the
northern districts only, (a) read in the desired variables from the first
dataset, (b) execute a DELIF using the northern/southern variable you added to
the first dataset, and (c) read in any variables you wish to use from the
second dataset. For example: XVARS money
c:iowa; delif south c:
southIA; XVARS vote66 iowa2;.
Gary King
2006-01-07
