Have an idea?

Visit Sawtooth Software Feedback to share your ideas on how we can improve our products.

Use one link for two different surveys with respondents having equal chance of accesing each of them.

I have created two surveys for my studies which differ in terms of context and framing. I want to distribute these using only one link. When a respondent uses the link I want it to be a 50% chance that they will be directed to survey 1 and 50% chance that they are directed to survey 2. Is this possible and if yes, how can i do this?
asked Apr 5 by Hampus

2 Answers

0 votes
Someone may have a more elegant way to accomplish, but I would use a third survey and use that link to distribute.  You can set up a survey with a quota set with two items, one for each of your surveys.  Set that using the least fill quota option and then skip them to one of two terminate screens where they are redirected to one of your two live studies.
answered Apr 5 by Jay Rutherford Gold (42,930 points)
0 votes
If you plan to do this in one survey, all respondents begin at the start of this survey, then get allocated to survey 1 or survey 2 (this can be done via a least fill quota approach that allocates the respondent to the more needy survey at the time), and then skips to the relevant block of questions (survey 1 block or survey 2 block based on the quota allocated).

So this approach has all data for survey 1 and survey 2 in the same data file.

Maybe you don't want the one survey approach? So you go with the two survey approach.

You can do this with some survey linking within termination pages.

First create your 2 surveys.

Then create a landing page survey. This is a very short survey that does the following ...

1/ Has ALL respondents directed to it with the same survey link. For this to work, you need every respondent to be allocated a unique ID or password.
2/ Include a quota variable using a least fill approach that allocates the respondent to either survey 1 or survey 2. The survey with the lower completes count at the time will be the survey the respondent is directed to.
3/ Once the quota (or the survey to complete) has been established, direct the respondent to the appropriate termination page (you can set one up for survey 1 and survey 2). When the respondent departs from the landing page survey, have them depart as an incompleted survey. Do not tick the "Terminate respondent" check box under the Settings tab of your termination question/s.
4/ The respondent now leaps out of the landing page survey to either survey 1 or survey 2.
5/ When the respondent completes survey 1 or survey 2, have them leap back in to the landing page survey. This could be a "thank you for your participation" page that appears after the termination pages that sent the respondent to survey 1 or survey 2. There is a neat feature in Sawtooth Software (sys_skipto) that helps you leap back to a desired question.
6/ Then finish off with a standard termination page that submits your survey as a completed survey.

So the map looks something like this ...


Come back to this post if you have any further questions. I'm hoping this has you heading in the right direction.
answered Apr 5 by Paul Moon Platinum (91,280 points)
Thanks for chiming in with a much more detailed response, that should provide everything they need.
So we meet again Jay. All good mate.

Your summary certainly points us in the right direction.

I just went a little deeper as I wasn't sure of the knowledge of the author for this post / topic.

I've seen this issue bob up now and then, and when you send the respondent to survey 1 or 2, you don't really know how many complete the survey just by allocating. By sending them back to the landing page and having the landing page submitted as a completed survey tells us the exact number of fully completed surveys, and we then get the even distribution of completes rather than those who were simply allocated a survey.
Great point!  That can definitely have a big impact on an even distribution for sure.
True! I have witnessed this slip up from time to time.

It's okay when you use this approach on language surveys because you have a landing page for the language desired, direct the respondent to the appropriate survey and then they complete that survey. Done! No jumping back to the landing page required.
Thanks for your answer, I just have some additional questions.

I have now createad the 2 surveys and uploaded them each to a seperate hosting server/project (correct me if this is wrong). I am now trying to create the third "survey" ( the direction page) but I do not understand how the quotas work. I am new to sawtooth/lighthouse and have never used quotas before, would realy appriciate some explanation on how the quotas work and how to set it up. Should I get the number of copletes for each survey from the admin page or should they be stored in the direction page survey in some way?
Hi there.

Your landing page can leap to any website / url. So wherever you place your 2 surveys is entirely up to you.

So let's get you sorted out.

We can take this in a step by step approach.

Create a landing page survey.

Include the following for now ...

1/ A text page which simply says something like  "Thanks for your time in completing this survey. Please press the NEXT button to continue.".

2/ Create a single response select question. The script will say "Please select the survey you would like to complete" with response options 1=Survey 1 and 2=Survey 2. Call it QSurveyChoice. We want to test this initially without a quota question and do it via a select question instead.

3/ Create 2 termination pages. One will leap to survey 1 and one will leap to survey 2. If QSurveyChoice=1 skip to the survey 1 termination page. If QSurveyChoice=2 skip to the survey 2 termination page. The 2 termination pages should leap out of the landing survey page to survey 1 or survey 2 as incompleted surveys. That means you should not tick the "Terminate Respondent" check box in the termination pages. Ensure you insert the survey 1 and survey 2 links in to the "Link to a Questionnaire or Website" section.

We now want to see this landing page survey select a survey via the select question "QSurveyChoice" and then skip to the correct termination page. From there it should leap to survey 1 or survey 2.

Go set that up and get that working first. Then a little later we can replace the select question "QSurveyChoice" with a quota question which will select the most needy of the 2 surveys via a least fill method. To do this, insert a quota question and remove the select question "QSurveyChoice". Name the quota question as the same name. Create 2 cells and call them Survey1 and Survey2. Select "Least Fill (Counts)" and select "Always Qualify". Set their target as the same targets you desire.

The final step will be leaping back from survey 1 and survey 2 to the landing page. There is a neat feature in Sawtooth Software (sys_skipto) that helps you leap back to a desired question in the landing page survey. At the survey 1 and survey 2 termination pages, you insert the landing page survey link with the "&sys_skipto=QLPThankYou" added at the end.

From there, you can create a thank you page (call it QLPThankYou) in the landing page survey (place this after the 2 termination pages). This just says something like "Thank you for completing the survey ...".

Then you finish off with a termination page in the landing page survey that submits the survey as a completed survey in the landing page.

If you need demo ssi files (survey 1+2 and landing page), I can make them up and send them to you. They will be simple examples that should guide you.

But try to set this up with a select question first. Then we can test survey 1 and 2 leaping back to the landing page survey and then replacing the survey selection with a quota question rather than a select question.
Hi again,

I  tried your suggestion and got the landing page survey to direct to the right survey using the select question. I then tried to replace the select question with a quota and it seems to work (when i press "next" on the landing page I get directed to one of the surveys). What I do not understand is how the landing page knows which survey has the least completes.

My second issue is when I try to go back to the landing page after finishing either of my surveys. I can not get the links/url to work, tried using the sys_skipto but I can not manage to make it work as intended.
Hi there. You got me just before I called it a night. I'm in Australia so it is well after midnight here. But I'm a midnight owl so I'm buzzing with energy.

The quota question now captures a "1" or "2", instead of the select question. By selecting the "least fill" option within your quota question and looking at the number of completes at the time the survey is being conducted within that quota question is how it determines which survey to allocate. The "lesser number of completes" survey is allocated. If the number of completes is equal at the time, a random selection is made. Does that make sense?

Now moving on to the leaping back from survey 1 and 2 to your landing page survey ...

Your landing page survey should look something like this ...

Intro text question
Quota question to determine survey 1 or survey 2
Terminate question for survey 1 (must terminate as an incomplete survey)
Terminate question for survey 2 (must terminate as an incomplete survey)
###Thank and close text question (this is the question we skip back to from survey 1 or survey 2)
Terminate landing page survey as a completed survey

What ever survey link you use to first go into the landing page survey is the same one we need to link back from survey 1 and survey 2, but we just add "&sys_skipto=QX" at the end. Replace "QX" with whatever you have called the thank you page within the landing page survey. That's the page marked with "###" above.

And ensure the termination pages that are used to leap from the landing page to survey 1 and 2 do not submit the survey as a complete or disqualified survey, otherwise the leaping back from survey 1 and 2 will not work correctly.

So to summarise, you should see respondents enter the landing page survey, be allocated with survey 1 or 2, leap to the allocated survey and complete it, then leap back in to the landing page survey where we say thank you and close the survey as a completed survey. When you look at the landing page admin module, you should see the number of completes there should equal the number of completes in survey 1+survey 2.
And you need to be using unique passwords or IDs to leap back and forth between surveys. This enables the whole process to work.
Thank you for all your help Paul.

I now realized why you suggested in your first answer that both surveys should be included in the the same survey and decided to go with this approach instead. I managed to get this working fairly quick with help of your exlpanations on quotas and how these can be used to skip to different parts of the survey. It also alowed me to not having to deal with all the links back and forth were I had the most problems.

My only concern with the approach I am using now is how to seperate the data depending on which survey a respondent has completeted when I am doning my analysis since they are all mixed together. But I think I can manage this on my own.

And again thanks for all the help, it is extremely appreciated.
Good to see you have the one survey approach working.

Don't give up on the two survey approach either. Once you get it working the first time, you'll say "Wow, this is pretty cool". So when you get some spare time, have a go at it and test it out.

First time I was working on it, I was bumping in to all sorts of hurdles. Then when it worked, I learnt from all my early mistakes. I'm sure you can do the same.

And the one survey approach and two survey approach are both fine. There is that linking issue to deal with in the two survey approach, but both methods are handy to know.

And ensure you fully test your survey to confirm when survey 1 is allocated, the survey 1 version is completed, and then the same for survey 2.

Use the quota question to filter to survey 1+2 when conducting your analysis.