Market researchers commonly use banner tables to show results that are broken down by several key pieces of information at the same time. For example, a banner table may display brand awareness across several demographics, like Age, Gender, and Region, so that all of those key breakdowns may be viewed at a glance.
Technical details
A key difference vs other programs
In most programs, banners are created as a part of a table. However, in the software, a banner is a special type of Variable Set that is used to bring together categories from many distinct Variable Sets. (If needing to show similar variables on a table, you should consider using Combine to group them into a new Variable Set).
The banner is not fixed to a particular table - you can use and reuse the banner on as many tables as you like (as you would with any variable set). Like other variable sets, if you modify a banner on one table, all other tables using that banner will update to reflect the change. You do not need to transform your variable set into a Binary-Multi format to add it to the banner. Do note that numeric variables cannot be used in banners as you may typically expect (their values are converted into categories).
Buttons, Options, and Fields in Drag and Drop Dialogue
- Label Name for the banner to be created. In Q, this field is Name.
- BANNER Drag variable sets from the Data Sources tree (or list) on the left onto this field to create a banner question. In Q, this field is Questions to include.
- Add sub-NETs When checked, the NET of each variable set is also included in the banner.
- Add overall NET When checked, a single NET is added to the end of the banner, representing the combination of all of the variable sets' categories. An alternative is to add a Total (see Adding a total).
-
Add column spans of variable set names This causes the name of the variable set, or, where there are duplicates (see the comments above for Replace question name with span where there are duplicates), the name of the original question, to appear as a span, as shown in the example below. In Q, this field is named Add column spans of question names. This setting added the Gender, Age, and Income headers below:
-
Add Total category When checked, a column for the Total Sample is added to the banner. Note this is different than the overall NET when there is missing data in one of the variable sets in the banner. The total will always show the full sample in each category in the rows, whereas the overall NET will show the number of cases with data for all variable sets in the banner. See the difference in the snapshot of a banner with a Total and overall NET below:
- Duplicate questions dragged onto banner Only available in Q. When a question or variable is dragged into Questions to include a new question is automatically created in the Variables and Questions tab. This new question stores any modifications you make within the banner (e.g., merging of categories or modifications to the Value Attributes. If Duplicate questions dragged onto banner is not ticked, a new question is not created, and any changes made in the banner will also be applied to other uses of the question in the project.
- Replace question name with span where there are duplicates Only available in Q. Q does not permit multiple questions to have the same name. Consequently, when a question is dragged onto the banner and duplicated, the new question is automatically created with a unique name. For example, the duplicate of Age will be called Age 2. However, usually, people will prefer to have the original name appear. Consequently, when Replace question name with span where there are duplicates is selected, Q automatically hides the name of the question and instead creates a new span using the original question name.
Significance testing
When using Column Comparisons, significance testing in a banner is determined by the spans, with tests being conducted within a span. In Q, you can also right-click on a category within a span and select Comparisons to get more fine-tuned control within the span. To access this setting in Displayr, see How to Specify Columns Compared on a Table.
Each variable set added to the Banner will be tested against the other columns within that variable set. In order to test across variable sets, you will need to use column comparisons or convert the sets into one large Binary-Multi variable set (do know that in doing this, you should review your Statistical Assumptions for your overlap settings, as it's likely there will be a lot of overlap between columns).
With a nested banner, the columns compared across for the exception testing depend on the structure of the variable sets nested and their Value Attributes. For example, nesting two nominal variable sets will have each Column % compared to the average of all other columns in the nesting. Below you will see Gender nested under Age; this is why the two Gender columns within the Under 30 span can both be significantly higher.
To test within a span in the example above, you must restructure the span variable Age set so that there are no respondents in the "NOT" group for each span. This requires you to transform the Age variable into a Binary-Multi variable set and adjust the Value Attributes so only 1/Selected values are included in analyses. This essentially turns the spans into filters on the data beneath. Now you will see the Gender columns are tested within each Age span:
With Date/Time questions, the Date/Time setting in Statistical Assumptions controls whether the tests are compared with the rest of the data (the default in Q), or, Compare to previous period (the default in Displayr). Where appropriate, this setting will apply across spans. In the example below, the arrows for Coke Zero - When 'out and about' for Jul-Sep 19 and 45+ means that this age group had a significantly higher score than in the previous time period of Apr-Jun 19.
Method - Creating banners
In Displayr: How to Create and Customize a Banner
In Q: How to Create a Banner
Method - Editing banners
Editing a banner in the Edit Banner dialog box
This dialog box is most useful for reorganizing banners, such as adding questions or changing any nesting. In Displayr, you select a table using the banner or the banner itself in the Data Sources tree, and the object inspector will show the drag-and-drop interface. In Q, you can access the drag-and-drop dialog by either right-clicking on a banner as it appears on a table or on the Variables and Questions tab, and selecting Edit Banner.
Editing a banner by modifying a table
Modifying questions
Once a banner has been created, each of the questions within the banner are modified in the usual ways. For example:
- Drag-and-drop to merge categories.
- Right-click on categories and select Values to modify their Value Attributes.
Adding a total
in Q, right-click on the banner and select Add Total Category to create a total on the banner. In Displayr, check the Add Total category checkbox in the object inspector.
Adding categories to a banner
Where there is a desire to add an extra question to a banner, the most straightforward way is by editing the banner. However, in Q, it is also possible to right-click on any of the categories in the banner and select Add Category - Logic, which allows you to create a Logic Variable with a simple expression, either by selecting an individual category from a variable, or, using simple expressions such as as Q2 = [65 or more] AND Q3 = [Male], or, Q5(1) OR Q6(1-4).
By right-clicking and selecting Add variable you can simultaneously create other types of variables and have them included in a banner. This feature is not yet available in Displayr. If you want to add a total to a banner in Displayr, please contact support.
Spans
A span is the term for groups of categories (e.g., Male, Female, and Income in the example below). You can modify and create Spans on banners in the usual ways (see Spans for more information). In Q, you can also automatically place spans above questions by right-clicking and selecting BANNER > Question > Display question name as span.
Dynamic updating
When a data set is updated, the banner itself will automatically update to reflect changes in the underlying data and data structures. In the example shown above, when a new data file is used which contains an additional quarter of data, the new column is automatically added to the banner in the appropriate place, and all the significance tests are automatically updated.