Hello! I’ve followed the steps in creating a forced choice task on the Forum and Tutorials, but noticed that in the Forced experiment’s design, there is a warning that says ‘Design has 21 columns, but 20 are used’. Can you help me out and let me know what this means and how to resolve it?
This is a valid warning. You have uploaded a design with 21 columns. The final column ‘BLOCK’ is not used and has no data in it.
It should have no impact if it is your intention to upload a design with redundant information in it.
To remove the warning - upload a design with 20 columns.
Hello, thank you for your reply. I generated the design using the Ngene add on the platform. I did not upload the design.
Normally this is not possible - did you regenerate the ngene design - this warning should disappear.
Hi, the warning disappears when I regenerate the design in the forced experiment. But since the opt out experiment and the forced experiment should be the same design, when I run my scenarios, the forced choice sets are different from the opt out scenario.
There’s quite a bit that’s vague here, can you fill in the details.
- The first experiment, was generated with Ngene and is fine right?
- The second experiment: The problem disappears when you generate with Ngene - correct? If so - then how was it originally created? - did you upload a design into it? where did that come from? Or did you copy the experiment then delete the opt out?
- What do you mean the opt out and the forced should be the same? That’s not true - the opt out will have an constant column for the opt out alternative
- What do you mean ‘when I run my scenarios’ the forced chociue sets are different - whyt, how and isn’t that to be expected?
Hi, thanks for your reply.
- Yes, the first experiment (optout) was generated with Ngene and is fine.
- The second experiment (forced) is a duplicate of the first experiment, but I deleted the third alternative, optout.
- Yes, I know that there will be a constant column for the optout alternative in the first experiment. What I mean is that, if the respondent selected optout, in the next page, the same choices should be presented, just without the optout alternative.
- To try and fix the “design has 21 columns but only 20 are used” error that shows in the forced experiment, I clicked the Generate using Ngene button in the Experiment tab > Run. After I did this, the error disappeared and I tried the experiment through Preview in the Pages tab. However, what happens is that when I choose optout for scenario 1, the next choice set (forced) shows a different mix of attribute levels than the one in the optout scenario. Shouldn’t it be the same since we are ‘forcing’ them to choose after the optout?
Thanks again in advance for your help.
There are Two separate issues here.
Firstly - and this is extremely important and fundamental to DCEs.
An experiment design is always generated after the x-matrix (The attributes and levels). It is plan on what levels to change.
Changing the X-matrix (removing alterniatives in your case) afterwards invalidates the experiment. Furthermore There is a clear lock on the X-matrix screen that requires an ‘I know what I’m doing’ override. In your case the change is trivial but still that’s why you get the warnings.
Secondly I understand what you are trying to do and this is a common use case - however experiments are independent of each other. If you want to synchronise the row allocations - then you need to explicitly allocate them - the system cannot understand this is your intention.
There is a tutorial on ‘Opt Out with Forced’ on the tutorials page that goes most of the way here:
I suggest your workflow is the following.
-
Get your opt out experiment working properly
-
Duplicate it as ‘forced’
-
Download the design as a codeplan
-
Remove the additional alternative in ‘Forced’
-
Edit the Experimental design (in Excel) for opt out - removing the uneeded opt out alternative column.
-
Upload this design (there is a helper).
-
Put a test for the opt out choice scenario and display the forced.
-
Ensure that the rows for opt out and forced are synchronised.
-
Publish and run a simulation.
-
Check the data is as you expected (is there a forced only for opt out chosen etc).
I think you’re almost there.