Galaxy Zoo Messaging Experiment – Results
Some of you may remember a while back we posted a blog announcing that we would be testing a new messaging system on Galaxy Zoo. Some of you may even have seen these messages while classifying on the site! This test was also part of a study of how we could use messaging to increase engagement on the project. Working with researchers from Ben Gurion University and Microsoft Research we delivered messages to volunteers at key times during their participation on Galaxy Zoo and observed how these messages affected their engagement. This research was based on previous work we had done that demonstrated that sending similar messages in emails could increase the likelihood of volunteers returning and engaging more with the project.

This screenshot shows an example of how the messages were shown in the Galaxy Zoo classification interface.
The volunteers were split into three main cohorts; One group who were delivered the messages at random intervals, one group who were delivered the messages at what were predicted to be optimal times, and a final control group who received no messages. This study has led to two peer-reviewed publications and the results show that optimal timing of an intervention message can significantly increase the engagement of volunteers on Galaxy Zoo.

This plot shows how the messages sent at the predicted optimal times were significantly more effective at increasing engagement than those sent at random.
These early results are intriguing, and we’d like to do more tests to see if it’s something we can use more broadly across Zooniverse projects. The same machinery might also be used by Zooniverse teams to send messages to volunteers – either in a group or individually – as they participate in their projects. We’ll keep you informed on the blog.
To read about the study and its finding in more detail please see the following papers:
Click to access engagement_intervention.pdf
Click to access optimizing-interventions-offline.pdf
For a discussion regarding the ethics of this study, please read this Zooniverse Talk thread https://www.zooniverse.org/talk/14/675633.
10 responses to “Galaxy Zoo Messaging Experiment – Results”
Trackbacks / Pingbacks
- July 12, 2018 -
- July 25, 2018 -
Cool! 🙂
Neither paper’s PDF has a date, but I think the first was published some time in 2016, and the second early in 2018.
May I ask, why has it taken so long to let us – zooites, citizen scientists – know about the results of this research? I’m particularly curious as we are the “lab rats” here (not to mention the recent, intense scrutiny given to the use of data on online behavior by giants like Google, Facebook, etc).
There is following discussion in Zooniverse Talk about the study.:
https://www.zooniverse.org/talk/14/675633
It was started by me 10 days ago!
The discussion was locked by Zookeeper Chris! So it is no longer possible to do posts overthere!
Yeah, it got locked with several other threads because you kept spamming them (like you do here) despite being told to stop.
In which journal/s were the papers published?
The links in the blog post aren’t the journal links!
Awaiting a reply!
Are you going to send an e-mail to the GZ volunteers, where you inform them about this blog post?
I think, that this is necessary in this case!
Awaiting a reply!
I guess, that ithis is the “stalker app”
, which had been presented at a Zooniverse workshop in 2013!
According to the notes by a volunteer, who had participated in the workshop,
Arfon Smith had done following statement.:
“How likely is someone to leave in the next 5 minutes? (We can tell!) Show them the “bacon” and stop them.”
PS: It is called “stalker app” in the workshop notes!
Besides altering/modifying galaxy images (fake AGN), you apparently also alter/modify behaviour of GZ volunteers (intervention)!
I am astonished, that you think out loud about your plans for the future! We haven’t clarified the past yet!