Hi there, just to let you know, for those that haven’t already noticed… PsychoPy/JS 2022.1.1 is now available from all the usual locations.
Highlights of this new release:
Improved interface for searching, syncing and viewing Pavlovia.org projects from
within the PsychoPy app. No more of those timeout errors and better ability to search and sort
Can now insert new Routines from pre-packaged templates (such as Fixation, 2-alternative forced choice, feedback, etc).
QUEST handlers now available in PsychoJS including interleaved staircases
Most visual stimuli now have an “anchor” parameter to control how it’s laid out relative to its position, and you can now set the alignment of text within a Textbox (with a handy demo to illustrate the difference between anchor and alignment!)
Added support for Pupil Labs eye trackers (thanks to Pupil Labs themselves) directly from the Builder
In addition to adding the we’ve done a lot of work to make the experience of using PsychoPy just generally smoother and more work on the PsychoJS engine to bridge the gaps in functionality
Thank you for the announcement. It would be great if something can be done to protect against losing datafiles. For instance, I have often noticed that MTurk workers complete their task and enter into MTurk the appropriate surveycode but their Pavlovia datafiles are missing. This may happen ~10% of the time, or more, and each time it does happen I end up paying the MTurk worker and don’t have any datafile so it’s a financial loss in addition to a datafile loss. Would you consider doing something to protect against this problem?
I would love to find out if this is definitely due to data loss. If you use the VESPR Study Portal you can restrict payments to participants who reach the debrief. I would like to find put whether data files are ever missing for participants tagged as completed by my web app.
Yes, to expand a little on Wakefield’s response, the way the data losses typically occur, I believe*, is through participants closing the page while the data file is being sent. To know whether that explains your case we need to know a bit more about your use of the site.
Are you forwarding participants back to MTurk using the CompletionURL setting in PsychoPy, with their ID number built in? That should certainly wait until the data file is sent and then return them back to MTurk.
Alternatively you might be printing their number on a screen, or providing a clickable link, and allowing them to go to MTurk themselves, in which case they might be exiting before the data file has completed the upload. When you say, “enter into MTurk the appropriate surveycode” it makes me think you might be doing this and
Basically the solution is not to give the participants their own code to type in. Let the system do that and it should automatically only do that when the data are uploaded. Wakefield’s solution effectively pushes you to use that approach, which is why it might provide information here.
best wishes
Jon
* this is the only way l’ve been able to cause a data file loss myself
David P Crewther | Professor | Centre for Human Psychopharmacology | Swinburne University of Technology, Melbourne, Australia | +61 3 9214 5877 | 0423 032 661 | Adjunct Professor, La Trobe University| Melbourne, Australia | National Co-Chair Asia-Pacific Conference on Vision