Next
A filter defines one or more conditions--for example, that the respondent's gender must be male and age must be over 45. Regardless of how you define your filter, internally the conditions are expressed as a conditional expression (for example, "gender = {male} And age > 45"). A conditional expression returns a value of true or false for each case.
When you generate a table that has a filter applied to it, IBM® SPSS® Data Collection Survey Reporter applies the expression to each case. If the result is true, the case is selected and included in the table. If the result is false, the case is not selected and is excluded from the table.
In a simple filter, such as the one described above, it is quite easy to see which cases pass the filter. In more complicated filters it is not always so obvious. For example in a filter that defines certain categories that are to be excluded, only cases that are not in those categories pass the filter.
For example, if the variable stores the respondent's gender, you might want to specify that the gender must be female. How you define the conditions depends on the variable's type:
The conditions must be based on the values that are actually stored in the variables. For example, when you filter on a categorical variable, you can base the condition on the responses to that variable, but you cannot base the conditions on any statistical elements that you have defined for the variable using the Edit Variable window. Similarly, when you filter on a numeric variable, you must base the filter condition on the raw numeric values stored in the variable and you cannot reference any bands that you have set up for the variable on the Banding dialog box.
The reason for this is that editing a variable changes how the variable will appear when you include it on the side or top of a table. It does not actually define the structure of the variable.