Training

You’re working on an analysis project using IBM SPSS Reports for Surveys, and you realize that your results seem skewed based on the fact that returns are proportionally different by region compared to your original sample

IBM SPSS Reports for Surveys allows the use of any numeric variable as a source of weight factors. Within a table’s properties, you can select the appropriate weighting variable. It seems like an easy solution to “weight” results based on a region weighting variable. But what if you don’t have an appropriate weighting variable in your dataset?

Depending on your organization’s processes, you may have options to go back to an earlier "data management" stage and add the weighting variable to the dataset. Tools such as IBM SPSS Statistics Base or IBM SPSS Data Collection Professional would provide the capability to add these variables using factor, target or rim weighting techniques.

In many situations, the weighting technique is more straightforward. You can calculate the weight factors by hand and apply them directly within the IBM SPSS Reports for Surveys software. Let’s look at a situation like this.

In the table below, we see the distribution of results by region, the desired number in the sample and a calculation of the weight factor that can be used to redistribute the results.

In this case, the calculation of the weight factors is relatively easy and use of weighting capabilities in data management software is not necessary.

With the factors above, we simply need to create the variable using the following steps:

• Click on the Region variable and then hold down the <CTRL> and M key simultaneously to bring up the dialog box to create a new variable.
• Apply the following options:
  • Name: Weight_Region
  • Description: Weighted region factors
  • Type: Real
  • Based On: IIF(Region={North},0.718563,IIF(Region={South},0.868613, IIF(Region={East},1.337079,IIF(Region={West},1.416667,1))))

In the variable above, we choose Real because there will be a calculation resulting in a number with decimal places.

We also use the IIF function as the basis for the variable. IIF works like an "IF ... Then … Else" statement in programming.

We would read the nested set of IIF statements above as:

Whenever we reference a categorical response in the variable Region, we put the category name in braces { }. These names can be seen in the Properties View of a variable.

• Click on Save and Close in the Edit Variable dialog box.
• Add the Region and Weight_Region to a profile table and click on the Results tab to validate that the correct weights have been applied.

• If everything looks good, apply the weight by selecting the variable under Weight in the Table Properties.

The ability to set up a quick weight matrix can be very helpful in your analyses. This application of the IIF function represents just one of many ways to conditionally control the definition of new analysis variables in IBM SPSS Reports for Surveys.