Indexed filters

Indexed filters are a type of data cut. They are designed to allow the comparison of like-for-like responses for different verticals with a single data cut. Rather than create a logical expression for each question row or question column in the Data Cuts applet and manually apply each one to its own Topline tile, an indexed filter can automate the process. It only requires a single expression to be written, and automatically tailors the expression to each column or row within the tile to which it is applied. Indexed filters allow respondents of a similar usage/awareness categorization within a funnel to be used as a basis for comparison between columns or rows. This backs out biases introduced by market share, or other external or  irrelevant factors which may skew ratings. These filters are built with logical expressions using IntelliSurvey syntax. For more information about writing syntax, refer to the NEEDS NEW DOC TO LINK TO Survey Programming section of this manual.

To use an indexed filter, you must have a compatible set of questions:

• The filter needs a question or variable which establishes the criteria that the grid tile data will be filtered by. There are no restrictions on what this question can be, but it should at a minimum include all grid tile columns as options for the filter to be valid.
• The question or questions which generate the tile upon which the Indexed Filter is to be applied. This can be any reporting tile with two or three axes of data: a Summary tile, or a Grid tile. See Topline report for more information about origin questions for this tile type.

In the following example, respondents are first asked about the frequency with which they eat bananas from various brands, and are later asked to judge each brand on a set of criteria.

setlist: BANANA_BRANDS
1. Dole
2. Chiquita
3. Del Monte
4. United Fruit Company
 
starttable: 1
intro: How often do you eat bananas from the following brands?
type: radio
rowsfrom: BANANA_BRANDS
1. Rarely
2. Sometimes
3. Often
4. Non-stop
endtable
 
newpage
 
setlist: CRITERIA
1. Price
2. Freshness
3. Flavor
 
starttable: 2
intro: Please rate each banana brand on the following criteria, where 1 is poor and 5 is excellent.
type: pulldown
colsfrom: BANANA_BRANDS
rowsfrom: CRITERIA
optsfrom: system.1to5
endtable

The tile for table 2 in the Topline Report displays as a grid tile:

The average values are the default view for the grid tile, and all data collected is used to generate those values. Indexed filters permit the data to be refined, so that only the data for brands which meet a given frequency requirement in Q1 will be used to generate the values in the Q2 tile.

To create an indexed filter, select the Gear icon > Analyze > Indexed Filter > Add.

Now enter in the filter syntax in the Expression text field. Substitute '[id]' for the row or column elements for which the filter will apply. The '[id]' acts as a placeholder for the column ID or row ID for which the system will create individual filters. Include a filter name in the Description field to help identify the filter.

For the present example, suppose analysis demands we only evaluate responses for a brand if the respondent indicated 'Often' (option 3) or 'Non-stop' (option 4) at Q1. The filter would be defined as:

anyChecked($Q1R[id],3,4)

This is equivalent to defining four separate data cuts, with each cut applied to the respective column which it references:

anyChecked($Q1R1,3,4)
anyChecked($Q1R2,3,4)
anyChecked($Q1R3,3,4)
anyChecked($Q1R4,3,4)

For more information on using anyChecked, see the anyChecked function and other similar functions.

Once created, indexed filters can be turned on and off, edited, or deleted. These options can be accessed by the Indexed Filter settings menu, in the same location where they were created:

As long as it is active, the name of the filter as defined in the Description field appears in the bracketed text in the upper right portion of the tile as shown below.