network profile
| bio | website | thinkingaboutthinking.org |
|---|---|---|
| location | Halifax, Canada | |
| age | 31 | |
| visits | member for | 1 year, 4 months |
| seen | Apr 5 at 10:17 | |
| stats | profile views | 23 |
|
Apr 4 |
revised |
Intelligence and marriage satisfaction added 167 characters in body |
|
Apr 4 |
answered | Intelligence and marriage satisfaction |
|
Apr 2 |
awarded | Nice Answer |
|
Jan 18 |
awarded | Yearling |
|
May 11 |
answered | The current recommended text for statistics in behavioural sciences |
|
Apr 27 |
awarded | Enlightened |
|
Apr 27 |
awarded | Nice Answer |
|
Mar 30 |
comment |
How to analyze reaction times and accuracy together? Looking at the pros/cons to using a binomial mixed effects model of accuracy as a function of RT and other predictors is one of the projects I had planned before this discussion for our summer stats student, though I'm thinking of looking at both linear and generalized additive models of the effect of RT. |
|
Mar 30 |
comment |
How to analyze reaction times and accuracy together? IE has become lamentably standard in some fields partly, I suspect, because it feels somewhat intuitive and is certainly an easy "solution" to the vexing problem of combining speed and accuracy. Admittedly, I don't have data explicitly demonstrating the invalidity of the speed-accuracy scaling assumed by IE (though I now think I'll task a student to generate this data through simulations this summer), but it would seem rather remarkable if the simple IE scaling, which at least rather questionably models accuracy as a proportion (see Dixon, 2008, "Models of accuracy..."), comes out as valid. |
|
Mar 30 |
comment |
How to analyze reaction times and accuracy together? Unfortunately inverse efficiency assumes a particular scaling of RT and error rate that is completely unsupported. Even Dr. John Christie, a colleague of mine that gave IE its name, has since come to completely oppose its use. |
|
Mar 30 |
comment |
How to analyze reaction times and accuracy together? These days I prefer LBA because it's computationally simple/faster to fit and because I haven't seen any demonstrations of cases where diffusion can fit data that LBA can't |
|
Mar 29 |
answered | How to analyze reaction times and accuracy together? |
|
Mar 16 |
awarded | Enlightened |
|
Mar 16 |
awarded | Nice Answer |
|
Mar 11 |
comment |
How can I create computer based psychology experiments using OS X? @ArtemKaznatcheev I haven't empirically tested things myself, but I've been told by sources I trust that web-based RT collection isn't reliable. However, it's easy enough to take a python script and create a stand-alone app (or exe) using py2app (or py2exe) that your participants can download. |
|
Mar 3 |
answered | Can response time be incorporated into signal detection theory? |
|
Feb 28 |
awarded | Commentator |
|
Feb 28 |
comment |
How can I create computer based psychology experiments using OS X? Yet another addendum: the code I linked is very much written in a procedural style rather than an OOP style. This is because I find OOP unnecessary for the simple stuff I tend to do and find it takes a little more effort (both coding and planning) to do fully OOP experiments. Possibly just a personal quirk though. |
|
Feb 28 |
comment |
How can I create computer based psychology experiments using OS X? Oh, and just a heads-up: the timing in that experiment is slightly non-optimal insofar as it uses the time at the beginning of the trial loop to label found responses during that iteration of the loop. Now that I think of it, there should probably be another "now = time.time()" line right above the "pygame.events.pump()" line, just in case there is a delay during stimulus presentation. In the end, though, this will all be obviated when I finish my side-project to update SDL1.2 & pygame to implement SDL1.3/2.0's high precision timer and event timestamping... |
|
Feb 28 |
comment |
How can I create computer based psychology experiments using OS X? Here's a zip of one of my moderately well-commented experiments: filedropper.com/castforweb |
