select_one questions irregularly not answerable

[JürgenBiendara]

Hello,

An extensive customer survey (Survey123 version 3.11) contains numerous select_one questions "Present/Non present". Depending on the answer, further questions are displayed in each case.

It happens again and again that individual of these select_one questions cannot be answered, one cannot select one of the two options. Only the select_one questions seem to be affected.

So far, the customer cannot determine any regularity (neither entered values nor specific steps), when and for which select_one the problem occurs. If the problem appears, it can only be solved by closing the survey completely (and thus losing all entries) and then starting a new collection.

The problem occurs irregularly both on the desktop (Windows) and on the iPad.

Has anyone also observed this behavior, can anyone help us find the cause?

Thanks a lot,
Jürgen

[Anonymous User]

Hi @JürgenBiendara,

Are you using the Inbox and do you have repeats in the survey? Are the select one questions that are not answerable inside a repeat, and do they have calculations, relevant expressions, or other configurations applied that are different to questions that do work as expected?

Can you share your XLSForm (xlsx file and media folder files) so we can take a closer look?

Have you tried the latest version of Connect and the field app, currently released version is 3.12.

Regards,

Phil.

[JürgenBiendara]

Hello Philip,

Thank you very much for the quick reply. Here are some answers from the customer to your questions:

Are you using the Inbox and do you have repeats in the survey?
-> We’re not using Inbox. Some questions in the survey are related to each other.

Are the select one questions that are not answerable inside a repeat, and do they have calculations, relevant expressions, or other configurations applied?
-> Yes, every time the problem occurs it is the ‚radio button‘ question and it is related to other questions.

The customer would be willing to provide their XLS form for troubleshooting, but has made it clear that all data and information will be kept strictly confidential and must be deleted upon completion. Therefore, providing the survey as an attachment via GeoNet is unfortunately not possible. Can you send me an upload link for the submission?

Thanks again and best regards,
Jürgen

[Anonymous User · Accepted solution]

Hi @JürgenBiendara,

I have sent you a private message with my contact details.

Regards,

Phil.

[Anonymous User]

Hi @JürgenBiendara,

Just letting you know that the Early Adopter Community (EAC) has just been updated with new 3.13 October Update RC builds. These builds include a fix for large surveys with slow loading times, when trying to open the drafts or inbox folder, and when submitting surveys. If you have time, please take a look and test out your surveys with these new builds, and let us know any feedback in the EAC forums. We hope to get this update released and available in the app stores as soon as possible.

Regards,

Phil.

[JürgenBiendara]

Hello @Anonymous User,

Given this fix, I'd be interested to know if this also has an impact on the number of questions recommended for a single survey (around 500)?

Thank you very much and best regards,
Jürgen

[Anonymous User]

Hi @JürgenBiendara,

The latest 3.13 October update is now similar to the performance you saw with 3.12 or earlier. Large surveys may still take additional time to load and open. The number of questions by itself is not the exact problem, it is more related to how many of those questions are linked via expressions, such as relevance or calculations.

You can have a 2,000 question survey with no expressions and it will load fast, or you can have a 500 question survey with 500 relevant expressions or calculations that link every question together, and that survey will load very slowly. Finding a balance between the total number of questions and number of expressions that link those questions will give you the best performance on large surveys. In general up to 500 questions work well, once you are over that number, the more expressions will slow the survey down.

Regards,

Phil.

[JürgenBiendara]

Hi @Anonymous User,

Thank you for the explanation, that makes a few connections a bit clearer. My customer still has a few questions about this:

1.
Is the ArcGIS Survey123 Quick Fix Release for October 8, 2021 (https://community.esri.com/t5/arcgis-survey123-blog/arcgis-survey123-quick-fix-release-planned-for/ba-p/1104621) announced by Ismael identical to the Survey123 3.13 October Update RC you mentioned?

2.
You wrote: "The latest 3.13 October update is now similar to the performance you saw with 3.12 or earlier." What exactly does that mean - it's faster or still "as slow" as "earlier"?

3.
Given our customer's issue discussed in this chat (select_one questions irregularly not answerable), what benefit can the customer expect from the fix? Is there a direct correlation between the customer problem and the fix?

Thank you very much for your help, best regards,
Jürgen

[Anonymous User]

Hi @JürgenBiendara,

To answer your questions:

1. Yes, the 3.13 Quick Fix Release on October 8, 2021 is the exact same build as Survey123 3.13 October Update RC on the Early Adopter Community (EAC). First we share the RC builds on EAC for early testing and feedback, then on the release date the app stores are updated with those same RC builds. You will notice the build number in About screen is exactly the same.

2. As you know, the 3.13 September release was very slow with large surveys, in some cases the Drafts and Inbox could not be opened or it took several minutes. Also loading the survey for a new collect was much slower. The fix in the latest 3.13 October update resolves these issues and the survey load time and time to open Drafts and Inbox is similar to what is was in 3.12.

3. As for "select_one questions irregularly not answerable", the behaviour will likely be similar to 3.12. However, I have not been able to reproduce that issue on any of my devices with testing that survey many times. Therefore can't say for sure if any difference, but given it is likely memory related, and we have improved the memory usage with that updated 3.13 builds, was hoping this may resolve the issue. The only way to know for sure would be if customer no longer reports that issue.

Regards,

Phil.