In this article
Statistics are available for numeric questions (open text or multi questions with the ordered property), ordered multi questions, single questions with weights defined in the answer list and grid questions with weights defined in the scale.
The following statistics are available:
- Count – the total number of responses. Note that for single questions and grids this will be the number of responses for which a weight is defined in the answer list /scale. Therefore, if there is for example a “Don’t know” item in the answer list, which will not normally be assigned in the scale, the count will not include these responses.
- Avg – the average of the responses.
- Stdev – the statistical standard deviation for the responses, treating them as members of a population.
- Stdevp – the statistical standard deviation for the responses, treating them as members of a complete population.
- Var – the statistical variance for the responses, treating them as members of a population.
- Varp – the statistical variance for the responses, treating them as members of a complete population.
- Max – the maximum of the responses.
- Min – the minimum of the responses.
- Sum – the sum of the responses.
- Mode - the most frequently occurring number amongst the responses.
- Median - the number separating the higher half of a sample, a population, or a probability distribution, from the lower half.
- Standard Error - evaluation of standard error statistic. Both weighted and un-weighted values are supported. Note: the results will be different when using table engine 1 and 2 due to the different standard deviation formulas used.
Note: The average will be selected by default when you drag in single or grid questions with weights defined in the answer list/scale, as they then will be interpreted as measures. This means that if these questions are crossed with other questions that are also used as measures, you may receive error messages such as “Statistics are defined by intersecting headers”. This is because you are trying to cross an average with an average. To avoid this problem, deselect “Average” on the question in the header variable properties.
Note: If you have dragged a Statistics object into the table, then in the table either double-click on the object or right-click on it and select Properties to open the property sheet and select the statistics to be used (go to Table Toolbar for more information).
Note: Base (N) and Statistics (Count) may well return different results, and this difference will be the number of Don’t Know (DK) answers. You can make N and Count the same by setting a weight for the DK answers. If the two values are the same (a weight has been set), then you can use either value in a report. If however the values are different, then you must ensure you always use only Count (to include the DK data) or only N (to omit the DK data).
Statistics Base
Some questions in the survey may not be answered by some respondents. This may be because the question has been set up as “not required” and the respondent has moved to the next page without having provided an answer, perhaps the respondent did not answer the entire questionnaire (incomplete or screened), or the questionnaire could have been set up with skip patterns so that certain respondents did not see all the questions. By default the percentages are calculated using the Exclude not answered setting, thereby basing the results only on those respondents who have answered the question. However, if you wish to include in the base the respondents who have not answered the question, set this to Include not answered.
This setting is usually used in combination with including “Not answered” on the header variable(s).
Include Not Answered
Use this in combination with the “Include not answered” setting for “Statistics base” in table properties to show the respondents who did not answer the question .
Figure 1 - Including "Not answered" on Favorite and Gender
If “Include not answered” is not selected in the table properties, the Not Answered fields will show zeros.