Filter Logic
Whenever you filter, remember that the filter eliminates ALL respondents who do not match the filter criteria from your tab. That means the respondents will be eliminated from the rows, columns, layer, and groups. To test a filter such as sex=male, place the "sex" question in the filter and subset for male, then include sex in your columns and/or rows and male respondents should equal 100%.
Whenever you add a question to the filter, it will automatically be added to the filter logic. By default it will first read: A, then when you add a second question it will read A AND B and after adding a third question it will read A AND B AND C. This is the filter logic typically used. Subset each filter question to allow only the desired respondents to show up in your tab.
Sometimes you will want to filter for one criteria OR another. This is as simple as clicking in the filter logic area and editing the text to read: A OR B
You may want two criteria to be looked at simultaneously. This can be done with parenthesis. Just type in the filter logic area for example: (A AND B) OR C
The only other logic you can use is NOT such as: (A OR B) AND NOT C
Group Filter Logic
Group filter logic works a bit differently than normal filter logic. To begin with, as soon as you include a group question in your filter, the filter description will read:
Group Questions Treated As Multikey
Filter Applied To Respondent
If you include more than one group question in your filter, the filter description will read:
Group Filter, Matching Responses
Filter Applied To Respondent
If you subset a normal group question in the filter window for one response and run the tab, you will notice that only information for that one response will appear in your spreadsheet. To remedy this problem, click on preferences at the bottom of the filter window and select "Group Respondent Filter" rather than the default "Group Response Filter". This means that rather than see the responses to the one response you filtered for, you will see the responses for all the questions, for people who chose the response you filtered for. A very important distinction to make.
If you are filtering for multiple characteristics of a group question, you may also run into problems getting numbers that make sense. That is because there is a special filter logic for group questions due to the way they work. Instead of using the parenthesis to group questions together, you must use the { }.
Let us use the following filter questions in our next few examples.
A - Considered Vehicle (G1)
B - Model (G1)
C - Age
Also assume our columns have "Model (G1)" and our rows have "Considered Vehicle (G1)".
So for example if you are filtering for people who would definitely consider the Viper and are under 40 years old, and your filter logic looked like the following example, your numbers would not be right: A AND B AND C
The numbers may even look correct, depending on whether you have "group response filter" or "group respondent filter" selected, but the numbers are not right. This would be better illustrated if you had subset two different models in the filter. Both models would have had a 100% if the filter had been working correctly.
However, if our filter logic was to say: {A AND B} AND C
The numbers would actually tell you what people under 40 years old, who would definitely consider a Viper, would consider in addition to the Viper. So the cell where "Viper" and "Definitely Consider" come together will still be 100% telling you that all the respondents who make it through the filter fall into these boundaries, but the other cells will also contain information about what else these respondents would consider.
