Filtering with several variables - Crosstabulation

An example is to look at a result filtered on the variable "Gender", what women, men or employees with a different gender identity have answered in a survey. It is also possible to add another variable to the filter, for example "Country", to see what the result looks like specifically for women in Sweden, or men in Norway, for example. 

You can choose to show the broken down results individually, or grouped.

How does it work? 

In "My results" you can filter the results, both in the "Questions" page and in "Analyze".

1. Open the setting "Selection" 
2. Select which group you wish to see the results for by marking it
3. In the "Filter" view, select the first background variable you wish to use, and mark the value(s) you wish to include
4. Then, select "+ Add filters" to add an additional background variable
5. Select the additional background variable you would like to use, and again mark the values you wish to include
6. In "Grouping", select if you wish to break down results by groups (you find more information regarding this further down in this article)
7. Save! 
   
   

How is anonymity ensured?

And why can I see some group results broken down, but not others?

As with the breakdown of results using one (1) background variable, cross-tabulation is only allowed on summaries.

The minimum number of people needed in a group to allow cross-tabulation is 100. This is based on the fact that breaking down results using one (1) background variable requires 50 people.

A maximum of two (2) variables can be used and combined at a time when filtering results, for example "Gender" and "Country".

Threshold values

For filters to be useful and provide relevant results, there are some thresholds/limits to consider:

• What is the minimum number of people to create a single filter?
  
  A filter is only created if there is enough data for it to actually produce a result.
  
  For example, if there are only 2 people in an organization who have the background variable values Woman ("Gender") and Sweden ("Country"), no filter is created at all.
  
  The threshold here is five (5), which is the same number of responses that according to our standard is required to break results on one (1) background variable.
  
  
• What is the maximum number of unique values ​​for filters on an attribute/variable?
  
  In order for an attribute to be used for simple filtering (an attribute/variable), it must have a maximum of 100 unique values ​​as default .
  
  An example would be "Location", where cities are registered, or "Role" where title or position is registered. The number of cities or titles must not be more than 100 in order to be used for simple filtering.
  
  
• What is the maximum number of unique values ​​that can be combined in a filter?
  
  When multiple attributes are combined in the same filter, a stricter limit applies. By default, a maximum of 10 unique values ​​are allowed. So, combining "Gender" + "Age" works well if they have 3 + 5 values ​​together, but if there are for example 40 countries ("Country"), it is not possible to create a filter for "Country" + "Gender".
  
  
• What is the maximum number of attributes that can be used in combinations?
  
  If there are a large number of filterable attributes, combinations are limited to the 15 with the highest priority.

Grouping (Group By) - Different ways to display combined results

This function is based on the same concept found in, for example, Excel and the pivot tables that can be created and used there. 

Example: You have selected "Country" = SE + NO + FI followed by also "Age" = 30-39 + 40-49 (years) 

Without grouping, we get all combinations of the selected variables, in other words a pure cross-tabulation;

SE + 30-39
SE + 40-49 
NO + 30-39
NO + 40-49 
FI + 30-39
FI + 40-49

If we select grouping by "Country", we aggregate all values ​​based on that variable and get;

SE (30-39 + 40-49)
NO (30-39 + 40-49)
FI (30-39 + 40-49)

Grouping here then provides the opportunity to actually combine different sub-results. However, the sub-results must meet anonymity requirements in order to be included.