This article describes what Qualtrics Export settings to use when you intend to import an SPSS (.sav) file into Displayr.
- A Qualtrics account that allows you to export to SPSS
- Log into your Qualtrics account.
- Select the project you wish to export.
- On the Data & Analysis tab, click Export & Import > Export Data.
- In the next dialog box, select SPSS.
- Click More Options at the bottom.
- Tick Recode seen but unanswered questions as -99 and change the blue -99 to 0 (zero).
- Tick Recode seen but unanswered multi-value questions as 0.
- Click the Download button at the bottom.
Notes for Using Qualtrics
There are some aspects of the way Qualtrics' SPSS files are generated that require extra care and consideration when the file is imported into Displayr:
- Qualtrics truncates their Variable Labels to 99 characters. Qualtrics' support recommends that question labels and/or variable labels are added for answer choices, that labels are shortened, or that they are edited after the download. Alternatively, when scripting a Qualtrics survey, ensure that long introductions and question texts are kept in Descriptive Text nodes, with shorter text in the question text proper.
- Multiple-response variables in Qualtrics SPSS exports do not include a Missing Value for respondents who skip a question or option. Ordinarily, Displayr uses the missing values to determine the base sample size for the question when it is displayed in a table. For Qualtrics data files, Displayr will automatically rebase multiple-response variable sets to the NET, which corresponds to the set of respondents who have made at least one selection among the options in the question. In situations where this is not desired, such variable sets can be rebased by using one of the methods outlined here.
- When multiple select style questions show images as the options to choose (rather than labels), the file names of the images are not stored in the data, and this results in blank Variable Labels. Manual labeling may therefore be required.
- It is further possible to duplicate question/variable names in Qualtrics. Always ensure that these are unique prior to export, as these are instrumental to Displayr when deciding which variables should be grouped together into variable sets.
- The data stored by Qualtrics for their Conjoint question type shows the results of the experiment (e.g. utility scores), but does not include the raw data used to calculate those results. As such it is not appropriate for use in Displayr's Experiment question type.
- Value Labels for Bipolar and Profile questions may require manual editing (see below for details).
- Like/Dislike Hotspot questions can sometimes have extra, empty variables included (see below for details).
- Displayr's ability to set up Qualtrics files depends on certain aspects of the structure of the data file. It is best to not edit the data file in SPSS or any other data processing software before importing it into Displayr.
Qualtrics question types mapping in Displayr
The below table clarifies how the various Qualtrics question types are set up by Displayr upon data file import.
|Qualtrics Question Type||Qualtrics sub-type||How it appears in Displayr||See also|
|Multiple Choice||Single Answer (all layouts)||Nominal|
|Multiple Choice||Multiple Answer (all layouts)||Binary - Multi|
|Multiple Choice||Dropdown List; Select Box||Nominal|
|Multiple Choice||Multiple Selection Box||Binary - Multi|
|Text Entry||Single Line; Multiline; Password; Essay; Form||Text|
|Rank Order||Drag and Drop; Drag and Drop Graphic; Radio Buttons; Horizontal; Horizontal Graphic; Horizontal Text Entry; Selection Box; Vertical; Vertical Graphic; Vertical Text Entry||Numeric - Multi||If open text is also collected, this will appear as Text in Displayr.|
|Matrix Table||Bipolar; Likert, Single Answer; Likert, Drop Down List; Profile, Drop Down List; Profile, Single Answer||Nominal - Multi||Notes below Bipolar and Profile Questions|
|Matrix Table||Likert, Multiple Answer; Profile, Multiple Answer||Binary - Grid|
|Matrix Table||Constant Sum, with Total Box; Constant Sum, without Total Box; Rank Order||Numeric - Grid|
|Matrix Table||MaxDiff||Numeric - Multi|
|Matrix Table||Text Entry, Short; Text Entry, Medium; Text Entry, Long||Text|
|Slider||Draggable Bars; Draggable Sliders; Stars||Numeric - Multi|
|Constant Sum||Draggable Bars; Horizontal Graphic; Horizontal Text; Vertical Graphic; Vertical Text; Vertical Text Entry; Vertical Graphic w/Total; Vertical Text w/Total||Numeric - Multi|
|Hot Spot||Like / Dislike||Nominal - Multi||Notes below Hot Spot - Like/Dislike|
|Hot Spot||On / Off||Binary - Multi|
|Graphic Slider||Text Above; Text Below; Text Right; Text Left||Nominal|
|Heat Map||Heat Map||Numeric - Multi|
|Net Promoter Score||NPS||Nominal|
|Side by Side||Side by Side||N/A||Qualtrics - Side by Side|
|Pick, Group, and Rank||Drag and Drop; Drag and Drop, No Columns||N/A||Qualtrics - Pick, Group, and Rank|
|Drill Down||Drill Down||N/A||Qualtrics - Drill Down|
Bipolar and Profile Questions
In Qualtrics the Bipolar and Profile question types allow the survey author to have several different scales within the same question layout. In most cases people would probably use the same scale for each item, but this is not mandatory. When the data is imported into Displayr, these different scales will get grouped together as there is no information in the data file to say that the scales are different.
For example, we may have two five-point scales within the same Bipolar question: one may be a satisfaction scale, and the other a frequency of consumption scale. In cases like this, there is no reason to show the two scales in the same table because they are not comparable. Thus, in Displayr, we split these items into separate questions.
A Bipolar question in Qualtrics appears as a Nominal – Multi question in Displayr where the meaning of the scale is shown in the row labels rather than the columns. To split these up, do the following:
- Highlight the variables for the Bipolar question in the Data Sets tree.
- Right-click and select Split.
- For each of the variables, note the text in the Label column (which describes the scale), then click Properties > DATA VALUES > Labels on the right and modify the entries in the Label column appropriately for the scale.
- Update the Variable Set Label text to reflect the scale.
A Qualtrics Profile question is similar except that the categories do not necessarily have to be a scale. You can employ a similar strategy to the above. However, when exported from Qualtrics, the value labels are stored as “Answer 1”, “Answer 2” etc., and so you will need to enter in the correct category labels manually. In both cases, if the sets of categories are the same between items, then you can simply leave the questions as Nominal - Multi variable sets.
Hot Spot - Like/Dislike
For Like/Dislike Hotspot questions, Qualtrics sometimes stores extra variables in addition to the variables for the areas of the image specified in the questionnaire programming. These variables appear to be empty, and it is likely just an error in the Qualtrics export process. To remove the extra variables from the variable set, highlight the relevant variables in the Data Sets tree, right-click and select Split.