Is OSWeb suited for my experiment?
Good afternoon everyone,
I've built an experiment and I'm wondering if I could carry it out online with OSWeb (after a few adjustments of course), in other words, is OSWeb suited for my experiment? It's a psycholinguistic experiment where response time to a visually presented Spanish target word is recorded.
Participants first hear the first syllable of a Spanish word over their headphones (e.g. SÁ_), and at the word offset they see a word on a screen which is either a complete match (SÁBADO), a match but with stress on a different syllable (e.g. SALÓN), a mismatch (e.g. MOSQUITO) or a non-word (e.g. FÍTULA). They need to judge whether the word exists in Spanish or not, yes or no (SÍ/NO). Presumably their response time is faster when seeing the matching prime (the target word would have been primed in their minds).
A few caveats; I cannot control the participant's RAM speed, wifi connection, quality of their keybord, and quality of their headphones, but then of course those would presumably be constant factors for each individual participants so...
Does this sound problematic or good?
Many thanks in advance,
Daan
Comments
Hi @DPvanSoeren,
Your experiment design is entirely compatible with OSWeb. It's a straightforward design.
OSWeb differs slightly from OS in the sense that some of OS' features wont work in the browser. You can read about some of the limitations here: https://osdoc.cogsci.nl/3.3/manual/osweb/osweb/
The latest version of OS comes with OSWeb 1.4, which includes a series of new features making online experiments even easier and more versatile: https://osdoc.cogsci.nl/3.3/notes/3310/
Best way to make sure your experiment will work in OSWeb is to test it using the browser. If you end up having to program some code, you'll have to do it using the inline_javascript object (Javascript language; Python is not supported in OSWeb). For debugging, you can use the browser's console to help identify possible problems.
As you probably know, to run your experiment, you'll have to save it as a JATOS experiment (using the dedicated menu in OS) and import it into JATOS. JATOS needs to be running on a server. You may install it on your own server (it does require some technical knowledge of server management and setting), or you could apply for a free account on Mindprobe (https://mindprobe.eu/). the service is run by ESCOP and is free. Finally, while developing your task, you could also use the free tryout JATOS service here: https://www.jatos.org/JATOS-Tryout-Server.html (however, this is only useful to develop your task, not for testing, as the server wipes out all experiments every 24 hours).
For online experiments with auditory stimuli, I'd recommend including some pre-experimental check to make sure your participants have sound (for example, I present some auditory sequences of 3 digits and ask them to type them in the same order; that way I can be sure they hear the stimuli). As for the caveats you mentioned, they are inherent to all online studies. For discussions of the pros and cons of online testing, these papers may be of interest to you:
Birnbaum, M. H. (2004). Human research and data collection via the internet. Annual Review of Psychology, 55, 803-832.
Reips, U. D. (2002). Standards for Internet-based experimenting. Experimental Psychology, 49, 243-256.
Going back to your question: yes, OSWeb is suitable for your experiment. There are other platforms and systems, as you probably now, but I find that OS has the advantage of being user-friendly and free.
Good luck with your experiment!
Fabrice.
Hi @DPvanSoeren,
Completing my earlier reply... There is in fact an already made example of lexical decision task in Open Sesame, which you can access here:
This example will work in OS, not OSWeb, but it can very easily be adapted to OSWeb :
(1) The example task initially contains OS features that are not compatible with OSWeb, but few: (a) it contains a bit of Python code to counterbalance the response keys (this would need to be replaced by some Javascript code, which is easily done), and (b) the logger (the object that makes sure to collect all kinds of information in every trial would need to be set manually (by default it logs absolutely everything going on, which is actually a lot of information, and that puts a strain on the bandwidth when you run the experiment online). The latter is easily done too.
To see whether an experiment is compatible with OSWeb, a good indication is to look at the compatibility check available here:
In this case, there is a bit of Python code making sure that the two response keys are counterbalanced between odd-numbered and even-numbered subjects:
We need to delete the
counterbalancePython object (delete it and emptying theunused itemsfolder) and replace it with an inline_javascript one with code doing the same job:if (vars.subject_parity = "odd") { vars.word_response = "z" vars.nonword_response = "m" } else { vars.word_response = "m" vars.nonword_response = "z" }(2) To change what the logger will keep track of, go to the logger object, disable the "Log all variables" option, bring up the Variable inspector, and just drag whatever variables you want to see in your output into the logger's window:
This needs to be carefully done, for if you keep some crucial variable out, your output will not contain it. So worth taking the time to think this through.
That takes care of adapting the example task for OSWeb. The rest is just tailoring it to your needs (adding the sound, changing the instructions, etc.). If you're new to OS, following some of the tutorials will quickly help you to find your way around the program.
You can download the modification I described above here:
Hopefully, that should save you time and enable you to get your experiment ready fairly easily.
Good luck!
Fabrice.
Hello @Fab ,
Thank you very much for your helpful answers! I'm glad to hear I can use OSWeb for this, and happy to have your reply for future reference.
Thank you very much again!
Daan