Hey guys,

I've got a bit of a problem I can't get my head around with Limesurvey. A friend of mine wants to test two different direct mailing letters with some people. She wants one half of the group to read letter 1 and the other half to read letter 2. How can we keep track of who's read what letter and possibly automate the whole process? Please note that the letter needs to be a word or pdf document since it contains formatting.

Is this what token attributes are for? If so, how do I use them. Preferrably I would like the respondents to be able to download the letter from the survey screen and have limesurvey pick the right letter and register what letter the respondents read.

Any thoughts? Help would be greatly appreciated!

The easiest way is to create a survey where you ask "Do you want to be group a or b? Then use conditions (see manual) to direct the users to the questions. The following question should contains somethin like "You have chosen to be in group x. Therefore please download <link: insert a link to your document here>this document</link> because the following question refer to it.

Please read the manual, everything is explained there.

Mazi wrote:
The easiest way is to create a survey where you ask "Do you want to be group a or b? Then use conditions (see manual) to direct the users to the questions. The following question should contains somethin like "You have chosen to be in group x. Therefore please download <link: insert a link to your document here>this document</link> because the following question refer to it.

Please read the manual, everything is explained there.

Thanks for your response.

Hmm, that's sort of what I'm looking for. Isn't it possible to do this with token attributes though? (if attribute1 = x: do this, else: do that) We need two groups of equal size and with this method you're never sure beforehand.

As far as I know it is not possible to use conditions on token attributes.

If you have got some coding skills there are several other ways to do this. E.g. you could read out a timestamp and if the actual minute is an even number auto-redirect to group 1, else redirect to group 2.

Otherwise you could create two surveys (creat one completely firest, then use export/import to edit the second on). Might even be easier to evaluate the results.

You could use quotas to limit how many could choose either option?

You *might* be on to something with your idea of using the token attributes. The manual is very cryptic at this point, but in regard to the 3rd table of token field descriptions it suggests that the token fields could be used "if your survey is NOT anonymous". Try creating one token with ATTRIBUTE_1 value of eg. 'letter1' and one with 'letter2'. Then create a boilerplate question to use that value, eg. "You shall read {TOKEN:ATTRIBUTE_1} before answering the questions". If the placeholder is replaced by the respective ATTRIBUTE_1 values for the two respondents, you may be halfway there.

I'm not much good with Javascript, so I'd test the next step with a variation of the Alternate exit workaround:

In an 'entry filter' survey, I'd try redirecting respondents to letter1 or letter2 surveys depending on the attribute, something like



If that too works (or nearly so, with the value accessible to the script), then you might be able to use the following workaround section for a Custom OnLoad function that does things more elegantly. But I'm out of my depth there - more expert people might have good ideas how to proceed from there.

Regards.

Sounds good!

Walter, if you succeed please post your solution here because it might be very useful for other users.

Hey guys,

I'm sorry to disappoint you but I went with the easy way. I could have done something with ITEd's suggestion (great one), but the invitations needed to be sent quickly so I decided 'no'. I simply created two identical (different names) surveys, uploaded the two letters and put a link the invitation e-mail (letter1 for the first group, letter2 for the second). Another solution would be to put the link to the document in the survey description as that gets displayed on the first page of the survey (right?). Thanks for the help guys!

Kind regards,

Walter

Glad you found a way that works - simple is often best.

I just wanted to report that I did a quick and dirty, successful test of the suggestion to use javascript to perform certain actions according to the value of a token field: Token field data IS accessible to scripts in question text via the replaceable field names in the form of {TOKEN:ATTRIBUTE_1}.

So one can evaluate token field data in scripts. Just not in anonymous surveys because then the token is isolated to preserve anonymity.

Regards.

Thanks for your feedback!

Maybe you can add your helpful advice to the workaround section in the manual?!