Randomizing Sentence Order in a Paragraph | XM Community
Skip to main content

Hello,
I am trying to randomize the order of sentences in a paragraph and am having trouble thinking about how I can do so. I have 4 sentences, and I want to keep the last one fixed while varying the first three. For example, presentation order could look like:
1) Sentence 2, Sentence 3, Sentence 1, Sentence 4
2) Sentence 1, Sentence 3, Sentence 2, Sentence 4.
I know I can do this by writing out the possible combinations (six), and then entering them into a loop & merge while randomizing the loop it picks and choosing only 1 loop to display. I also believe I can do something similar by creating a hidden MC question and setting up randomization on there (in order to use advanced randomization), and then using that invisible question to complete the loop an merge. However, I do not want to use either of those options for one main reason: It is a lot of manual work, and that means a lot of possible error. If I were only showing participants one vignette/paragraph, this would not be an issue. However, each participant will be reading 28 vignettes. Does anyone know how this can be done?
For further information on my study, I have attached some similar example prompts and more information as a screenshot. If anyone has any ideas for how to do everything I ask for there as well as here in one step, I am more than happy to hear.
Screen Shot 2022-12-07 at 7.38.29 PM.pngThanks in advance!
Carlo

You can create a multipunch hidden question with all the sentences as choices. Then keeping the last one anchored randomize the three and use default choices check them. Then in the question needed just use pipe text to get the selected choices. This should solve your query!


Hi, Carlo! If I'm understanding your question correctly, here's my suggestion for a way to approach it:
In your survey flow, you could add a Randomizer element for sentences 1-4. Each Randomizer should randomly present 1 element -- the element options should be the possible sentence configurations saved as an embedded field. For example, for sentence 1, the two combination options are each saved as an embedded variable called "Sentence 1," and the Randomizer will choose one for which to assign the value for that respondent:
image.pngYou would create a Randomizer doing the same for each of the four sentences:
image.pngimage.pngimage.pngAt this point, the respondent should have 4 randomly selected sentences, each of which assigned to the respective embedded field name (e.g., Sentence 1, Sentence 2).
Now you can add another Randomizer to the survey flow to assign a random combination of sentences 1-3 and assign that selected value to a new embedded variable named "Sentence 1-3" -- there should be 6 possible combinations that you would list out using piped text to reference the previously established embedded variables (assign the value to "Sentence 1-3" by using piped text to pipe in the value for embedded variables, "Sentence 1" "Sentence 2" "Sentence 3"):
image.pngThe Randomizer with the different combinations should look like this:
image.pngimage.png*Be sure that your randomized statement is presented AFTER this process in the survey flow (you can see above that it is in its own block).
Then, within the survey editor, create your statement using piped text to reference the embedded variable "Sentence 1-3" followed by the embedded variable "Sentence 4." This should provide you with a randomized combination of sentences 1-3, with sentence 4 always appearing at the end of the statement. Here are some response combos when I tested it out:
image.pngimage.pngimage.pngIn your collected data, you'll be able to know the values of sentences 1-4 for each respondent since they'll be assigned to their own respective embedded data fields:
image.pngimage.pngAdditionally, you could add an element to the survey flow to capture the combination that was presented to each participant and save that as an embedded data field as well. To do that, just set a new embedded variable to equal the value of the question that displays the sentence combo in the survey:
image.pngimage.pngI hope this helps, but let me know!
-Erin



https://community.qualtrics.com/XMcommunity/discussion/comment/53776#Comment_53776Hi Erin,
Thank you so much for the very detailed response! I really appreciate.
Your solution is very helpful (and does give me an idea for how I may be able to address the issue), but I realize in my original question I did not specify that there cannot be repeats.
If I am understanding how your randomization and embedded data works, it would mean that participants could potentially see repeated scenarios and/or not see all the possible combinations of the independent variables (which comes down to the base 24 vignettes), correct? In my study, each participant needs to read *all* 24 scenarios, as it is primarily a within-subjects design. I want to introduce the randomization of the sentence order on top of the presentation of all 24 scenarios to prevent ordering effects with regards to variable presentation (i.e., I dont want to have the participants read the same 4 sentences in the same order presented 24 times). The most important part for participants is that they are presented all 24 unique scenarios. Does that make sense?
Would there be a way to ensure that all 24 scenarios are presented to participants, with the randomized order of the sentences within, while ensuring no repeats? I came across using groups and embedded data within (https://community.qualtrics.com/XMcommunity/discussion/19414/adding-non-repeating-piped-data-elements-into-questions-text-for-within-subjects-design), which could be useful - but I think it may have similar issues.
I have included a screenshot of the 24 possible combinations of IV levels that each partcipants must read, just for clarity.
Screenshot 2023-01-08 170851.jpg


https://community.qualtrics.com/XMcommunity/discussion/comment/53769#Comment_53769Hi Grahulp,
How would you make a hidden question? I have been able to hide them before with javascript, but almost any changes made to questions or the survey seems to break it. If there was a block wide option to hide, then that would make it much easier.
Thanks,
Carlo


You can click the next button using js or use timer question and click the next button at 0 sec. While using timer question you can just use the "wait" text or image while the next button will be clicked automatically.


Leave a Reply