Dear all,

To keep the researcher blind we would like to randomize the participant to 1 of 3 interventions within LimeSurvey. Although I noticed that randomization to different interventions is possible (although not super straightforward: www.ryananddebi.com/2015/04/09/limesurve...xperimental-designs/ ), it is not clear for me how to perform block randomization (to make sure an equal number of participants ends up within each group). Also, we want to use the same ID and randomization for additional follow-up questionnaires (i.e., participant is randomized to condition A in survey 1, and then receives the same condition when we give him/her survey 2; while participant who receives condition B in survey 1, automatically also receives condition B in survey 2).

Is this possible in Limesurvey and if so, how do I do this?

Best regards, Sanne

This is possible, more or less.

There is no real way to guarantee that you will have exactly the same number in each group. At least not out of the box. Because it is random and you know that random does not guarantee an equal distribution. The higher the number of cases, the closer you should get to equal distribution, but that's about it.

There are a number of options to manipulate this, but they are all somehow pretty manual. E.g. you need to check the distribution and then close one option and allocate the "random" number to the least filled "bucket" or something.

If you have the contact details of your participants, you could randomly distribute the "groups" they should be allocated to. Joffm tells people to let their grandmother pick... hahahahaha.

After the first survey you definitely have the contacts of those respondents to be invited again. Then you know which group they were in and you can add this to the custom attributes and you can use it in your later surveys.

Thank you for your response. Makes sense, so we use questions to randomize as the link above.
But: I want the researchers to be blind, also for the second survey.. This is of course not possible if we add the randomization in a custom attribute (unless someone else does this of course). Is this the only way to do this?

Yes, it is.
And if you do not want to bother your grandmother (I always misuse her to say "a person who is not all involved in the survey").
To avoid the manual work you may have a look at this plugin "updateTokenByResponse".
gitlab.com/SondagesPro/TokenManagement/updateTokenByResponse
Add the attribute in the first survey, but empty, fill it by the plugin, and use this sample as the new sample of the second survey.

I want the researchers to be blind

This phrase you did not explain.
I can understand that there is a reason to conduct a "double-blind" survey.
But in my opinion this is only relevant if there might be a bias by the behaviour of people in personal interaction (Does the patient get a placebo or the drug to be tested?).

I mean: Who is the researcher?
If the team who conducts the survey only exports data not related to the groups (neither the random number of the first survey nor the TOKEN:ATTRIBUTE of the second), the analizing team will not know anything except that there where two groups.
You see, this is how you organize the survey.

Joffm