Question: How do I create a code book from a vertical data set and what do I have to consider for DataLion?
Sometimes there is no raw data in a project, but values that have already been calculated, which are then listed one below the other in a column. An example could look like this:
This means we have a column with the calculated values ("value") and then further columns with filters, metrics, target groups etc.
Here, a row does not correspond to a respondent or a case as in survey data, but to a combination of filters or metrics. So, there is often a single column with numbers, but behind these there are very different metrics (in the example percentage values and case numbers).
But evaluating such data sets is no problem with DataLion. A suitable code plan could look like this (snippet):
|2||Filter1||Full Sample||Full Sample||Sample||Auspr||Filter|
|3||Filter1||Boost Sample||Boost Sample||Sample||Auspr||Filter|
The charts are then also easy to build, but you have to remember to choose the right metric. This can be done with counter variables, for example:
For example, if you want to compare men and women: click on the question 4 (target group). Then comes a (rather pointless) chart showing the number, i.e. 33% of the rows are total, 33% are men, 33% are women. But if you then set the filter to Metric = % and select "Value" as the calculation variable, then it makes sense.
Calculation variables can be activated in the backend as follows:
Then this option appears in the charts:
Alternatively, you could also select question 11 (Metrics -> Value) and then set the necessary filters as chart or global (Metric = % as chart filter and men vs. women as breaks in the chart).
Warning: With such data sets, it is important that all filters are always set so that you get a single value. If you don't set Filter2 here, for example, you would get either 139.6% (total) or 46.5% (mean value) as a percentage, both of which are wrong.
Bitte melden Sie sich an, um einen Kommentar zu hinterlassen.