Survey crashed for multiple participants

Question

Dear SoSci Survey Online Support Team,

Some of my participants reported that the SoSci survey page kept crashing on their side. Some of them emailed me around the same timestamp. I wonder if this is an issue on the survey platform side or the participants' side. And is it possible that's because too many people are trying to assess the survey at the same time?

Thank you in advance for your help.

Here are some of their messages regarding the connection issue:

29 Sep 2021, 01:01 (EST)
> "Hello, I was just completeing the survey where there was a 3 minute
> window of browsing articles about XXX, but when I tried to click on
> another browser in the last 70 seconds, my browser tab crashed. Not my
> entire browser, just the one tab. Then I was unable to reload the
> survey. I did not press the back button or anything like that. "

29 Sep 2021, 01:02 (EST)
> "I reached the task to view the search page of articles. After
> reviewing several I clicked to return to the list and a Resubmit Form
> error page loaded. I was unable to continue."

28 Sep 2021, 23:11 (EST)
> "the survey link is not stable. it keeps cutting and i have to start
> again every time. i have done that like 5 times now."

follow-up from participant:

it cuts at about 60 percent in, and one will have to start all over again. i did that about 5 times.. the network for the webiste link kinda resets and takes one back to square one. not the survey taker's network, it was unique only to this particular study.

28 Sep 2021, 22:31 (EST)
> "I was almost finished with the survey and had just finished answering
> questions about the article when I got the following error: "MySQL
> error No response from database server. "

Answer 1

And is it possible that's because too many people are trying to assess the survey at the same time?

Very unlikely ... the server han handle about three times the load that it is currently handling.

where there was a 3 minute
window of browsing articles about XXX, but when I tried to click on
another browser in the last 70 seconds, my browser tab crashed.

That sounds like you offered third-party websites as stimulus. And the description sounds like any content (in any other browser tab) crashed the browser itself, not just the survey.

I clicked to return to the list and a Resubmit Form
error page loaded. I was unable to continue."

A "resubmit form error" is displayed when respondents use the "back" button of the browser to navigate.

"the survey link is not stable. it keeps cutting and i have to start
again every time. i have done that like 5 times now."

Interesting. Could you ask ho "cutting" looks like?

I assume that it may again be an issue with navigation. If you open external URLs from the questionnaire and theny try to navigate around, you could run into a load-issue.

"MySQL error No response from database server. "

That actually looks like a server-side issue. Please note, that the server will restart on Sundays (4:00 UTC) and may reboot on other days on 5:00 UTC if necessary to install security updates. The downtime is usually <3 minutes.

What was the time that the MySQL error was reported for?

Comments

Thank you so much for your response!

The stimuli are actually not from outside links. I integrated a mock website using the technique developed by Dr. Unkel (https://github.com/joon-e/OSEM/wiki/What-is-OSEM), so it should be an internal page.

I also added the timestamps those errors reported in the original post.

---

> I also added the timestamps those errors reported in the original post.

22:30 EST is 03:30 UTC ... that may be a time where a server restart caused the issue. The other issues, however, sound like different sources of trouble. In two cases most likely the respondents pressing the back-button. I have talked with Dr. Unkel about possible solutions for that, but they will note be available before mid 2022, I'm afraid.

---

Thank you so much for the response! Another participant just followed up about "the survey link is not stable. it keeps cutting and i have to start
again every time. i have done that like 5 times now." The participant said: "it cuts at about 60 percent in, and one will have to start all over again. i did that about 5 times.. the network for the webiste link kinda resets and takes one back to square one. not the survey taker's network, it was unique only to this particular study."

---

Sounds like the interview ID is lost at some point of the process. Possibly a invalid backlink or some invalid HTML on any of the article pages?! Please go into the collected data, look for the record of that person (if anyhow identifyable) and try to give the same answers, so you may be able to replicate the issue. If you can replicate it, you can solve it.

Also go through the data and check if there are articles with a substantially higher drop out rate.