Thursday 15 October 2020

Finally, online experiments with E-Prime Go: Ready to go? PART 2

E-Prime Go is now a few weeks older and PST made great steps in getting it COVID-ready, so to speak. In the latest version, there's a new E-Prime Go website that hosts your experiment (which looks way more trustworthy than a dodgy GoogleDrive link), and allows participants to download and run the experiment. Finally, your participant's data can now also automatically be uploaded to the EPrime Go Website, which makes everything a whole lot smoother. I should spend a few hours making a nice walkthrough, but I haven't got round to it. Somehow, I did get round to answering an email with a couple of friendly questions. I suppose it's a bit like a bystander effect or diffusion of responsibility or something that I will often enough answer individuals with lengthy rants, yet fail to write a comprehensive introduction that can be used by anyone. How's your book coming along, Michiel? 

That said, having answered the email, I figured I might as well post the answers as they seemed a very reasonable bunch of questions that could almost be a FAQ if you take 'frequently' not too literally. The questions were from someone asking them in the context of the usual 'should I use E-Prime, or retrain in some cool new software'. We from the E-Primer recommend E-Prime, of course, but I honestly don't really care (PST doesn't pay me!). It's about the right tools for the right job. 

So, without further ado, these are some questions you might have about E-Prime Go.

  • What are the minimum requirements to run E-Prime Go?

PST has posted the minimum system requirements for E-Prime Go, but it seems these are the same as the ones for E-Prime Studio. They also have a ‘quick verification app’, but this seems slightly less quick and easy than I would like, given that practically, you just want the participant as quick as possible to run the experiment, or they will forget all about it. Your average willing friend or curious family member you managed to convince to do your experiment (a luxury resource that tends to evaporate as you stick in this line of business) will not necessarily know or care about RAM requirements, and if their PC is still running in 2020, it'll likely just do fine. If not, they'll find out soon enough.

Practically then, I tell my participants that they likely will be able to run the experiment if they have a Windows PC (Windows 7, 8 or 10, but not XP) and have admin rights. Note: technically, admin rights are a sufficient but not necessary precondition: a PC might be able to run the experiment even without admin rights, but they need to be able to download an executable file from the internet and run it, which some places strictly forbid. For example, a PC that is borrowed from a large organisation (or, e.g. is a library pc, or a student's PC borrowed from the university with the IT department enforcing strict policies) may not work, but if it is their own, it will probably be OK. I also find some virus scanners go haywire when they notice you want to run an executable from the internet – in which case you can let them send the file to the virus scanning company (I had that once, turned out fine), but it’s generally easier if you ask them to run it on a friend’s/parent’s computer instead.

PST also told me they know of some antivirus software that can affect E-Prime

  • Is there any parallel software on a system that can cause a crash

E-Prime is generally pretty solid and hard to crash once it runs, and E-Prime Go is the same. However, I would be careful about any software running in the background, particularly ones that can cause popups to be displayed, like Skype or Outlook. You should also try to ask subjects to close any unused browser tabs before running the experiment, because of course they take system resources and can also pop up notifications. This is, of course, true for any other time-critical software out there as well and even if literally everything would run from the server, you'd still have to cope with the fact that a pop-up attracts attention and if participant starts doing something entirely else during the experiment, they'll... well, not do your experiment.

  • Does the new update give subject numbers by default?

No it does not, and as is, it will use the subject/session number stored in runtime. To be honest, though, it’s less of a problem than I thought, as long as you tell each subject ‘If you agree to run this experiment, go to [link]. At the beginning of the experiment, you will be asked for a subject number. Please use the following number as both subject and session: [number]. That way I will know when you have finished the experiment.’  And change the number for every subject. Somewhere at the beginning of the experiment, add an inline:

c.setattrib “Session”, random(1,32767)

…that way, even if the participant forgets the number, you will be able to figure out from sorting the data both by subject and session. In a way, giving a subject number also means a type of informed consent. 

  • Is it possible for multiple people to download the task simultaneously or will there be problems with duplicate subject numbers.

As far as I noticed, it seems to work for multiple people to download and run tasks in parallel – of course, it’s not being run on the server, but I doubt that PST would have a problem with the small amount of data transfer before and after the experiment. See above for how to deal with duplicate subject numbers. As such, they are stored separately, but it can be a hassle to figure out which one is which after the fact!

  • How do I know if my participants will be able to read the letters on the screen

E-Prime is run by default at an old-school resolution of 1024 x 768, which was pretty standard back in the 90s and early 2000s. I think by now we can expect higher resolutions, and although you can make an educated guess that participants will have a widescreen (e.g. 1366 x 768, 1440 x 900, 1600 x 900, 1680 x 1050), there are still exceptions. I would definitely have E-Prime use ‘match desktop resolution at runtime’, because not all resolutions are supported by all screens, even though PST recommends using a prefixed resolution (which should facilitate timing commensurability). That means, however, you will have to deal with some people with extremes, so try to run your experiment at your PC’s lowest resolution to see if everything looks okay. There’s also the slight possibility that someone has a HiDPI screen (e.g. 4K), and everything will look illegibly small, but so far I haven’t seen this occurring (if it does, you might ask them to lower the Windows resolution before running the experiment). Keeping that in mind, I would make sure your text is at a fontsize that is comfortable to read across the board. E-Prime’s default is pretty high already though! 

  • How do I get more people to actually do my experiment?
Well, I told PST that they should do something like Mechanical Turk, but for E-Prime, and with reasonable minimum wages for 'volunteers', but so far that hasn't happened. I recommend the usual tricks of the trade: flyers, student mailing lists, social media. 

1. I post an initial, 6 line ad with apologies for spamming, the central topic of the experiment, the requirements in terms of time and money for the subject, what they get in return, and the requirements for participation (e.g. age). Importantly, I ask them they need to have 'a Windows (7/8/10) PC with power-user &/or admin rights (basically you or your friend/family owns the PC)'. I ask them to directly email me back if they're interested. 

2. Once they reply, I send them a slightly tweaked email (so it doesn't feel like you're sending the same stuff all the time), with pretty much all the normal information a thorough informed consent has, and a bunch of precautions they have to take for the experiment to run well. For example:
Before you begin make sure: 1) You have no other software (e.g. browsers, Skype) running in the background. 2) You have 35-40 minutes available for doing the experiment. There will be a short break in the middle. 3) Your environment is quiet and free from distractions. Please turn your phone in flight mode during the experiment. 4) You are using only a single display. [explanation how to turn off secondary display]. 

3. I tell them if they agree to all this, they can click the E-Prime Go link. If they run the experiment, you should see it in the E-Prime Go website, and also when they're done. However...

4. ...Usually (>50?) this is so much information and they fail to actually click. Why not do later what you can do now, right? I never procrastinate myself, especially not by writing lengthy blog posts instead of that book, but I have a cool trick. A week later, I ask them with polite curiosity if they haven't managed running the experiment and if they could kindly say if they want to not do it after all. Don't make it sound like you're being sarcastic or affect that overconcerned tone of internet companies when they want you to come back to using their services. A short, personal question is enough. At least, I found that once they don't feel like a robot is talking to them, they most likely immediately do the experiment. Magic!


No comments:

Post a Comment