Generating random interviews (AskiaTools)

This AskiaTools feature creates random interview data for testing purposes and then, in AskiaAnalyse, prepares your analysis portfolio based on this randomly generated data. This process is known as Random Data Generation or RDG. It is useful when you want to test or preview your questionnaire before it goes live. It can also be used to check routing efficiency.

To generate random interviews:

1. In askiatools, open a *.QES file or SQL database.
2. In the tools menu select generate random interviews.
3. Specify the options, as appropriate.
4. Click OK.

In this topic:

• Options
• Diagnostic information on routing efficiency

Options

The following options are available when generating random interviews.

• Number of interviews: Determines the number of interviews that will be randomly generated.
• Number of attempts per question: If your survey contains blocking routing based on the values entered for a question (e.g. a numeric question expects a number within a specific range of values), sometimes the randomly generated data will not satisfy the routing, and the interview will not continue past this point. However, the random number generator can make multiple attempts to generate a random value that will get past the blocking routing, as determined by this option. For example, if the number of attempts is set to 10, then the generator will make up to 10 attempts to generate numbers; if it is blocked on the first attempt, it tries again, and if still blocked it tries again, and so on, until it either gets past the routing, or it has made 10 attempts).
• Version used: If your questionnaire has more than one version (as set in AskiaDesign), you can specify here which version the random generator will complete. This option can be very useful when your questionnaire has many opportunities for the respondent to screen out before the main body of the survey. Without it, you may find that you need to run thousands of randomly generated interviews to get only a handful of respondents entering the main body of the questionnaire. However, if, in AskiaDesign you create test and release versions of the questionnaire, then you can set your routing instructions only to run on the test version; you will then be able to generate random data operation on the release version, and thus more easily generate random data for the main section of your survey. For more detail on creating versions in your questionnaire, see versions management in the AskiaDesign Assistant.
• Save interview after unsuccessful attempts: If this option is selected, randomly generated interviews will be saved even if they fail to get past a blocking routing. Otherwise, blocked interviews are not saved.
• Save interview if screened out: If this option is selected, randomly generated interviews will be saved even if they are screened out (e.g. if they trigger the routing go without saving). Otherwise, screened-out interviews are not saved.
• Random seed: This is the number that is used as the basis of the random data generation. If you anticipate that you will want to repeat the same randomly generated data gain, you should make a note of this number (it can be copied and pasted into an email or other document). You can then enter the random seed when you later generate random data, ensuring that askiatools will create the same random value (i.e. with the same distribution across all codes and variables). This is useful for testing the effect of a questionnaire change whilst keeping everything else the same.
• Quota: if you have a task set up in askiasupervisor that includes quotas, and your QES has quotatodo routings, you can use this set of options to use these quotas while generating random data (this will work with either target counts or target percentages set up in your quota definition). For information on setting quotas on a task, see defining quotas in the AskiaField Assistant. The randomised data will take into account these quotas, ensuring that you have a balanced set of randomised interviews reflecting the quota targets.
  • Connection to AskiaField database: Enter an appropriate SQL connection string. For assistance, you may need to contact your IT department or server administrator. The format is as follows:
    Provider=sqloledb; Data Source=exampleServerAddress; Initial Catalog=exampleDataBase;Trusted_Connection=no;User Id=exampleUserId;Password=examplePassword
  • Surveys: Once you have entered a valid SQL connection string in connection to askiafield database, you can select the survey containing the quota information from this drop-down list.
  • Retrieve and use quota: Ensure this option is selected in order to use the quota information in the specified Supervisor task.

Diagnostic information on routing efficiency

This feature generates a log file which helps you determine whether the survey is running slowly, and if so, where the problem may be. The operation counts how many times each routing instruction was run and how long the operation took. A text file is generated in the directory where the QES is located, which looks like this:

RoutingID After Called Duration(ms)
241 DummyETB80aOuter 2000 475205124
240 DummyETB80aOuter 2000 474042064
242 DummyETB80aOuter 2000 469246579
243 DummyETB80aOuter 2000 242863099
239 DummyETB80aOuter 2000 242255059
555 EngBrandQuota7 2000 215363659
332 EngBrandQuota5 2000 214226001
323 EngBrandQuota3 2000 213706963
298 EngBrandQuota1 2000 212451961
355 DummyETD105aOuter 380 69458324
357 DummyETD105aOuter 380 69140587
356 DummyETD105aOuter 380 68849900
181 QPRTouch 500 60328405
177 QB120Statements 560 60171919

This shows the ID of the routing instruction, the question after which it appears, the number of times it was called (overall) and the number of cycles it took to run it (overall).