Skip logic issue

I’m encountering an issue with the skip logic for ego questions.
Questions still show up when the skip logic is set up to not show for a specific age or sex.
I can share the test protocol I made to double-check (but can’t share it as an attachment).

I’m probably doing something wrong. Can someone help?

I’m sending a wetransfer link with the test protocol, the 2 last skip logic questions don’t work: WeTransfer - Send Large Files & Share Photos Online - Up to 2GB Free.
And it seems when I tried the rule “includes” or “excludes”, that didn’t work either. But maybe that only works for specific types of questions (categorical vs ordinal types for instance)?

I’m looking into this today for you, Maya. Our preliminary tests show this might be another bug.

Hi Joshua, ok, thanks!
We are planning to launch our data collection soon, so hopefully this will get sorted soon, and not too many other bugs will show up in the coming days/weeks :grin:

Good luck solving the issue!

Hi Maya

You can work around the problem for now by using includes/excludes rather than exactly/not rule types. This will work in your case because you restrict the categorical variables to single option values.

This should let you launch your data collection without issues.

Hi Joshua, I was wondering whether you have managed to solve the issue?
Because what I shared was just a test protocol to check the error, but the actual one for data collection does contain categorical variables with multiple option values.

Hi Maya!

Yes, we have found and fixed the issue that you encountered. We are just testing the release to make sure that we haven’t introduced any additional issues.

I think the workaround I posted above should work for all scenarios currently. It may be slightly more complicated to implement, but it will definitely work.

We hope to have the new release out early next week. I’m afraid it will require a new minor version of Interviewer as well as Architect.

1 Like

I’m having issues downloading the latest release of Interviewer on tablets. It automatically downloads version 6.4.0 and notifies there is a new version available when opening the app, but then it sends you back to PlayStore where no new version is available.
Can someone help?

Hi Maya

We are having issues with updating the Android app to the latest version, due to some changes Google has made since we released our last version. We are “stuck” on version 6.4.0 of the Android app for now, until we have time to resolve the issue.

As I think I mentioned in another post, we recommend most users use a normal laptop for data collection for the time being, as Windows and macOS are much more heavily tested at present, and have no issues with updates.

With that said, I will try again to see if I can release the current version of Interviewer on Android. I cannot promise how long this will take as the technical problems are substantial.

Hello- I cannot get my logic to work. I have included both includes and exactly and the other options show up. Can someone help please?

Hi Juliet,

Thanks for posting your question. You’re encountering a bug that was fixed in the latest version of Architect (6.4.6). Please update Architect to the latest version, and let me know if your issue persists.

HI caden. Yes my version is 6.4.6

Hi Juliet,

Apologies, I initially misunderstood your issue. While skip logic determines if a stage should be shown, network filtering determines which nodes should be shown within a stage. You can read more about network filtering in our documentation article. Configuring your rules within the Filter section should give the expected behavior.

Hello I need further help to use skip logic.
I used filtering if alter chooses to snort/smoke but if alter doesnt choose snort/smoke it still jumps to question. When I added skip logic -pic below- it skips the 2 other options in this stage.

It only works for the first choice option of the stage but not the other 3.

Hi @jpiner2

Sorry to hear you’re still having issues.

I’m having a bit of trouble understanding the specifics.

Could you email me at with your protocol, along with a specific and detailed description of how you would like your protocol to behave? I will look into it for you as soon as I have a chance.

Just a followup for anyone who might have the same issue as @jpiner2. We were able to resolve the problem, which was conceptual rather than being a bug in the software. The below info may help others, so I am posting it here:

Skip logic is complicated to reason about, and I can see why you set your rules up the way you did in your protocol. What you had written was effectively:

SKIP this stage if the network contains any person where the attribute main_method_drug_usage does not include X”

There are a lot of negatives in that sentence! Perhaps you can see where the problem is?

Basically, your rule is effectively skipping this stage if any node in the interview network doesn’t have the attribute, rather than if all of them don’t. This means you will almost definitely get false positives.

What I did to fix your issue was to rewrite your rule to always skip the stage unless the specified criteria are met. You do this by toggling the rule option “skip this stage if…” to “show this stage if…”. This removes all the double negatives, and lets you write your rule intuitively (so as to match any node in the network). Thus, your rule becomes:

SHOW this stage if the network contains any person where the attribute main_method_drug_usage includes X

Once again, the key part here is that I am using “SHOW” rather than “SKIP”, and then flipping the criteria to “includes” rather than “does not include”. I think you can see that is quite a bit simpler to reason about and seems to work correctly in my testing.

I hope this helps!

1 Like