Hi everyone
Could anyone help me please?
I am a new user of LimeSurvey. I am going to run a 2x2x2 between-subject design experiment. Therefore, I need the respondents to be randomly assigned into 8 groups EVENLY. Does anybody know can LimeSurvey do this or not please? If yes, how?
Any response will be appreciated.

You can asign them randomly, but this is totally by chance, so it will most probably never be evenly. There are quite a few explanations here in the forum on how to do this.

Now, as you need the distribution evenly, I only see one chance: you need to use tokens and randomly give them the respective designs beforehand. You then include the information about the designs into custom variables in the token table which you can read later and apply the respective designs.

Hi thank you very much for your kind reply.

It seems to be out of my knowledge league. I will have to get familiar with the site first and see if I can manage to do what you said.

Thanks again

Tokens is quite a simple concept.

Basically you have a specific code which can be connected to personal data of the respondents (but doesn't really have to).

To help you with a little bit more detail, it would be important to know how you get your sample (how you recruit your respondents) in the first place.

Because with tokens you need to give your respondents a code or a link with a personalized code or send them an email.

Thanks for explaining that.

I now see what you meant. But it seems respective design is hard to avoid the resubmission. I will see what I can do based on your suggestions.

Thanks again.

I am not sure what you mean with "re-submission". The tokens can only be used once, if you don't increase the "uses left" number. Once they reach the end of the survey the token is invalid and they can't submit a survey with this token again.

Let's make a simple design: Let's say you want a total of n=800 participants, then you create n=1000 tokens. You can create tokens in a spreadsheet and then export it to CSV to import the token table with all relevant variables into Limesurvey.

Then you create n=100 tokens for each group of designs randomly (adding a custom variable that will contain information about which design should be used by this specific token).

As mentioned earlier, evenly distributing participants is by default not possible. This requires checking previous answers within a current survey session which can't be done for data security reasons.
The only work around would be to code a custom script which is called prior to the survey and checks the current distribution and based on that assigns the current user to a certain user groups and directs him to the survey.

If there is a sufficient number of participants, randomly assigning users should do the trick pretty well. That approach is described here:
www.limesurvey-consulting.com/how-to-ran...a-limesurvey-survey/

The only work around would be to code a custom script which is called prior to the survey and checks the current distribution and based on that assigns the current user to a certain user groups and directs him to the survey.

Even that will not account for respondents who are mid-survey and decide to quit later.