Below is a step-by-step how-to for downloading data from Qualtrics. I’ve had to tell people how to do this many times. It doesn’t take long (it’s easy), so I am only now getting around to writing a blog on the topic. (Mainly because I will soon be leaving my lab. This will provide an easy-access resource!) Please let me know if I have missed anything or if you have any additional questions.
Take me directly to:
Understanding the Data & Analysis screen
Selecting the download file format and variables
Output for Randomization
Downloading the data
OU users: your institution’s Qualtrics link is https://ousurvey.qualtrics.com/.
Step 1
Go to Data Analysis.
Step 2
On the Data & Analysis screen, you can see the most recent completed cases with the data they provided in selected columns. To hide the default columns, click on the header and select Hide Column. To add columns you want to see, hover over the header and click on the + sign that will appear. Select the one you want to add. I always add Duration (under Survey Metadata), because I want to know if people are taking more or less time than I expected.
Before downloading data from Qualtrics, make sure all your responses have been recorded. Below, the survey on the left has only recorded responses. If you have already paused data collection, Qualtrics will have given you the option to record all open responses or leave them open.
The survey on the right above still has three responses in progress. Click on the link to see them.
On the Responses in Progress you can see each participant’s progress. If you click on the Survey session ID, you can see all the responses given thus far. You will have to decide whether to delete or record partial responses. 96% complete responses, for example, should clearly be recorded; 5% complete should be deleted. Everywhere in between just depends on your survey and research design.
Step 3
Once you have recorded all the partial responses you want, go back to the Recorded Responses page. Click on Export & Import, then Export Data.
You can export data in various formats. I will demonstrate the basics using the Excel and SPSS format.
Other download options
Click on “more options” to see all the downloading options. They will depend on download format.
Most formats allow you to use commas for decimals (as you would in Europe), for example. See the image on the left below. I don’t do that.
You can also recode viewed but unanswered variables as -99, which i useful if you are going to use statistical data that reads missing data as -99 (such as MPlus, if you choose that option.) The center image shows that option. I usually do not do this, because I clean my data myself and code missing values in specific ways, but it is useful in some cases.
Convert decimals to commas. Recode missing (but viewed) as -99. Export randomized viewing order.
I always select “Export viewing order data from randomized surveys” (right image above), unless there is no randomization in my survey design. (See my post on Random Assignment in Qualtrics.)
Randomized order variables in Qualtrics output
If you download viewing order data, you will get a separate variable for each randomization, both at the block level and the questionnaire level. This last isn’t useful–I randomize the order of items on questionnaires, but I don’t need to know the presentation for each case. Unless of course you are using a question to randomize to conditions. You really need that information, and if you don’t download it, you will have to use a lot of syntax.
I randomize at the block level, in Survey Flow. Qualtrics will output a string variable that tells you which condition each case was assigned to; you can match the number of the block in your downloaded dataset to the block number in Survey flow (see below).
As you can see in the above images, the downloaded file will have a Display Order (string) variable that corresponds to the block ID in Survey flow. (In my example, it’s 34.) The values in the variable (in SPSS you will have a string variable; in excel, you must select either text or numeric field values) correspond to the block the participant was randomly assigned to view.
Above, you can see the values of the SPSS Display Order variable to the left, and the corresponding blocks in Survey Flow to the right. You can recode the string variable into a numeric variable with values corresponding to your conditions.
Step 4
Download the data by clicking on download!
What about the “Legacy” option?
Legacy refers to the old download format for Qualtrics. I do not know what the differences are for other formats, or even if it matters. For SPSS it can be very useful to choose legacy format, and I almost always end up using it.
Mainly, this is because I have old syntax files that I use to recode and clean data. In the Legacy format, unselected options (where participants can select multiple responses) are left empty. In the new format, they are coded as zero. Either I have to rewrite the code, or I have to download the legacy format. The legacy option also allows you to download cases from a given date range, which could be useful.
Other stuff
Embedded data variables (see my post on Embedded Data) are automatically downloaded. Note that to obtain the total survey duration, you must add it as embedded data (see image to right).
Qualtrics also offers the option to download SPSS syntax. I am not quite sure what that entails, as I’ve never used it. However, I assume that if you set the “correct” answers, or choice values (Under Tools–>scoring on the Survey screen), then Qualtrics will output syntax to score the test. For example, if you set correct options as 1 and incorrect as zero, then appropriate syntax should add up the values.
But I don’t really know if Qualtrics does this, because I’ve never tried. I prefer to write my own syntax.
And that’s all I can think of! Let me know if you have any questions, and if you’ve benefited from this, please….
Pingback: Random assignment in Qualtrics – Wild World of Research