Rebasing is a filter that can be applied to your data
when you need to calculate the shares or indices only from a part of the respondents (e.g., the Intention to buy a brand can be calculated only from those who are aware of this brand or who have purchased it before).
when you need to filter out irrelevant or unrelated responses and adjust to your screening criteria. This is done by altering the base (sample) used in the calculation.
Info! Rebasing of questions with missing values can instead be automatically handled using sysmis.
An example of a rebasing case
Consider a survey with 10,000 respondents with a gaming question, e.g. ‘Have you played any video games in the last 12 months?’
If half of the respondents (5,000) answered 'No' to this question, the following frequency of gaming question - ‘How often do you do each of the following?’ - will only be served to those who answered 'Yes', as the question is irrelevant to the respondents who do not play video games.
When served this question, any further calculations should be based on only those who answered 'Yes' to the original question.
Here are the calculations:
We have a total sample of 10,000 in a survey. 5,000 respondents answered 'Yes' to whether or not they had played video games in the past 12 months, and 500 said they played video games on social networking sites (at least once a week).
After the rebasing, the percentage calculation formula will be 500/5000 x 100 = 10%. Here we get % of people who play games on social network sites calculated from those who played video games.
Without rebasing the formula is 500/10000 x 100 = 5%, and we would conclude that only 5% of respondents play games on social network sites, and it might not be relevant because we include all respondents, regardless of whether they play video games or not.
Therefore, rebasing is important as it gives you accurate results. It excludes irrelevant data (for example, people who aren't asked a question, or blank data), and only uses those who have answered a question to calculate shares.
Rebasing does not delete your data or information, it simply ignores it and treats the rest of the data as a subgroup when calculating percentages.
Rebasing in Crosstab / Rebasing in Grid-report
Manual rebasing
Manual rebasing is applied individually for each category or a group of categories.
Hover over the LEB icon
next to any column/row label and click it.In the pop-up window switch to the BASE tab.
Drag&drop the variables for rebasing into the base tab. You can also build any logical expressions.
Click SAVE.
To rebase several options at once:
Select the options within the rows/columns that you want to rebase. Note, that you can’t rebase splits.
Drag the variable for rebasing onto the table, and drop it into the field marked with an asterisk *.
Select the ‘Rebase’ option in the pop-up.
The rebased options that are now rebased will be underlined.
You can edit/delete rebasing for each option using the LEB icon again. Note, that you can’t delete rebasing expressions for multiple series at once.
Automatic rebasing
You can also automatically rebase your data using DataTile’s label-matching algorithm in the following steps:
Select the variable for rebasing and drag it onto your table, dropping it into the field marked with an asterisk *.
Select the ‘Rebase: match categories with labels’ option in the pop-up.
DataTile automatically compares options' labels in the crosstab and attaches the rebase to matched rows/columns.
Thus, it is easy to get a calculation when we, for example, want to know how many of the people who know a brand (the awareness variable used for rebasing) have bought it. With automatic rebasing, the base will be different for each brand. However, ensuring that the brand labels are the same in all variables is important.
Likewise in crosstabs, rebasing is available when working on dashboards.