Issues with Visible if questions

Hi

I’ve an info sheet / consent for survey. It has 5 yes/nos on the first page. It has 2 html questions that follow on the second page. one of the html questions displayed if not all 5 were yes, the other displayed if all 5 were yes. Previously this worked as intended.

Today is the first time in a while that I’ve re-tested this survey. its broken.

I’ve re-done the visible if logic with the gui, ive changed logic gates seen if chat gpt thinks ive done something daft etc. its not working, but its not even predictable in its not working. changing things that shouldnt impact visibility logic (like adding or changing the logic on the second question) is impacting the first question being shown.

I’ve also got a debrief survey that also used to work, uses different syntax than those produced by the gui. the visibility logic there is also broken. The gui seems to be reading the old syntax as intended, but the survey is not displaying the right questions.

I’ve no idea how to fix this

If they are standalone surveys, the first step would be to try launching with 2024.1.0 in the URL instead of 2024.2.0

Alternatively, the issue might be that your logic needs to include the block name (e.g. {block_1/consent} instead of {consent}.

looks like the 2024.1.0 url is working as intended.

does this mean that if i manually set the study version to 2024.1.0 that this will be what any participants get?

If you share the edited link then that is what the participants will see.

There are currently four different survey “interpreters” on Pavlovia, all of which will run any survey in your library. However, 2024.2.0 or 2024.3.0 is needed for multi-block designs.

Thank you

You’re welcome. You may also find this thread illuminating.

Thanks

One a practical level how do you force the survey to run 2024.1.0?

I’ve just noticed that even if I change it with the drop-down in the survey overview, it defaults back to 2024.2.0 when I open the survey again.

Just give your participants the edited URL.

these surveys and others are embedded into a builder experiment using survey components.

I’m not directly giving participants the URL to the surveys.

I need to be able to direct them to the correct URL through a builder experiment.

Is this doable?

I’ve just relised I misunderstood what was meant by stand-alone in one of the previous replies.

these are not stand-alone in the sense that i can just give participants the URL to this survey alone

Then please try to use the block_1/ format. Does that work with an embedded survey?

Check a response to confirm the block names

this is the output from the responses looks like this:

“{
“1”: true,
“2”: true,
“3”: true,
“4”: true,
“5”: true,
“1_rt”: 2.196,
“2_rt”: 2.7470000000000003,
“3_rt”: 3.368,
“4_rt”: 4.232000000000001,
“5_rt”: 4.817
}”

to me theres not obviously any block name?

Is that using the 2024.2.0 link?

I believe so.

even if i copy and paste the 2024.2.0 URL i seem to get pretty minimalist data in the form shown above

I’ve just downloaded the datafile too - theres nothing thats strikes me as a block designation.

This may not be helpful additional info,

but if i preview the survey it behaves as intended. if i run it, it does not.

That is useful information, but not for me.

Have you tried using block_1 (the default block name)?

Seems to work neither in the preview nor the run. i suspect I’m not implementing it properly

the wizard wants me to use this:
{1} = true and {2} = true and {3} = true and {4} = true and {5} = true

I’ve tried

{block_1/1} = true and {block_1/2} = true and {block_1/3} = true and {block_1/4} = true and {block_1/5} = true

and

({block_1/1} = true and {block_1/2} = true and {block_1/3} = true and {block_1/4} = true and {block_1/5} = true)

Personally I would rename your questions so they aren’t just digits. That could be causing a problem for the logic

changed that, its not obviously the problem:

{q1} = true and {q2} = true and {q3} = true and {q4} = true and {q5} = true

works in preview, not in run

{block_1/q1} = true and {block_1/q2} = true and {block_1/q3} = true and {block_1/q4} = true and {block_1/q5} = true

works in neither.

incidentally I used single digits as these questions inherit details from the name of the survey in the experiment. so you end up with things like consent/consent1 which just looks ugly. if this is a possible problem generally, the default interaction between builder and surveys is going to push people toward doing things like single digits

Just looking at other surveys I’d like to use.

using the 2024.2.0 URL

a visible if using this:

{1} = true

doesnt work in run. but works in preview.

using 2024.1.0 it works in both

If i rename and re-run:

{test} = true

same behaviour. works in 2024.1.0. only works in 2024.2.0 in preview. whatever is going on seems pretty fundamental and not easily fixable by an end user amending a survey.