You can open AskiaSurf campaigns (QEW files) in AskiaAnalyse and AskiaVista, and explore this data just as with normal projects (QES files). A wave variable appears in the data, allowing you to analyse or filter your data by wave.
You can also select which wave or waves you want to work with. In AskiaVista, a list of waves appears in the general tab during chart or table definition; in AskiaAnalyse, you can select waves from the edit menu. You can choose any combination of waves, and only the selected waves will appear in your chart and tables. If you have a very large amount of data, selecting only the waves you wish to work on can improve performance.
The AskiaSurf module provides a two-level structure to define continuous research projects or trackers: the campaign (the entire project) and the wave. When starting a new campaign or tracker, you should identify the frequency with which new data will be added and at which points the questionnaire will be likely to change and use this to determine what will constitute a new wave in your campaign.
First, to take a simple example, if a new batch of data is added every calendar month, then this will determine the wave structure, and there will be one wave added per month. This also allows for the questionnaire to change each month, and for those changes to be reconciled on a monthly basis, at the time the data are added.
However, your tracker may be more complex, and the point at which questionnaire revisions may occur may not coincide with the regular cycle of additional data. For example, there may be a new batch of data every week, even though the data will only be reported on a monthly basis. In this case, you may either choose to add each week’s data as a new wave, or to join the files together, outside of AskiaSurf (using AskiaTools), to create one consolidated QES file for the month.
You would not be able to join the data together, however, if any questionnaire changes or corrections affecting the data structure had occurred. For this reason, we recommend that you aim to build your wave structure at the most granular level - in this example, at the weekly level - so that even unforeseen changes in the data structure can always be accommodated.
A similar situation can arise if a questionnaire change is introduced part way through a wave. For example, in a tracker which is conducted on a quarterly basis throughout the year, some additional questions were added, at the last minute, in June and July. Unfortunately, this falls outside the normal wave pattern as it introduced a questionnaire change in the 3rd month of quarter 2 (April-June), and another after the 1st month of quarter 3 (July-September). In this case, a new wave should be introduced in June, to allow the mid-period changes to be introduced and reconciled in askiasurf. The second change, if it is merely that the questions are removed at that point, may not require reconciliation, as the questions can be retained but not asked for the remaining months. The resulting wave structure would therefore be 5 waves, with a second wave for quarter 2 alone.
One key benefit of using AskiaSurf is that it provides an automatic wave variable, which is very convenient during analysis, as a profiling variable to apply to the columns of tables. However, if your reporting periods do not coincide with the AskiaVista wave structure (as would be the case with both examples just given), then you will need to create and maintain your own reporting wave variable, independently of the AskiaVista automatic wave variable. The automatic wave variable is still most useful as a means of checking and validating your data in AskiaAnalyse or AskiaVista.
If you have existing projects that you are considering converting to AskiaSurf format, it is likely that you will have a consolidated QES file or data file for all previous waves. You will need to weigh up the benefits of using this consolidated data file against importing each wave individually and reconciling the differences in AskiaSurf.
If you import all of the consolidated data, then this effectively becomes wave 1. The built-in wave variable created automatically by AskiaSurf will assign all prior data to wave 1 and will not be able to differentiate between the different reporting waves that this superset of data contains. You will need to create your own reporting wave variable to define the actual historical waves that the AskiaSurf wave 1 contains.
If your project has only a few previous waves, it is certainly worth importing them individually. If you have several years' data, then importing the consolidated data is likely to be more time-efficient. Alternatively, you could consider importing them as separate years, and then importing all the waves from the current year individually.
Even though the scale of the work may at first appear daunting, if you have a number of existing waves to import, there can be considerable advantages in importing each wave individually and using the askiasurf reconciliation tools to rework the changes. Mistakes when manually consolidating wave data are very common; using AskiaSurf will not only help avoid such mistakes, but may uncover errors that already exist and allow these to be corrected.
When performing analysis on wave data, you need to bear in mind the implications of questionnaire changes on the total base size from wave to wave.
If a question is missing from one or more waves, the total base size across all waves is reduced to take into account the number of waves in which the question actually appears. For example, if you have 3 waves, each with 100 respondents, and if the question appears in wave 2 onwards, then the total base would be 200 (100 each for waves 2 and 3).
If the question is present in a wave but a response within a question is missing from one or more waves, the total base size is not affected. For example, if you have 3 waves, each with 100 respondents, and no responses were provided to that question in wave 2, then the total base would be 300 (100 each for waves 1, 2 and 3).