Ok thanks! Sorry, I thought you were one of the contributors.

Best,

Olympia

Thanks for letting me know. I am not involved in developing the AnovaRM method. Maybe you could open up an issue on their GitHub page (https://github.com/statsmodels/statsmodels). This way it may be fixed.

Erik

]]>I just wanted to let you know that I did find a discrepancy in the AnovaRM function in the beta version of statsmodels when I ran a 1-way RM ANOVA with 2 levels (so akin to a paired t-test). I noticed the degrees of freedom of the error was off, and indeed the function gave different values than in SPSS. So in sum, the two-way RM ANOVAs I tested give identical output as SPSS, but the one-way RM ANOVAs do not. Let me know if you would like me to send the data I used?

Thanks in advance for all your help. The AnovaRM function implementation is great, so I hope this can be sorted out relatively easily.

Best,

Oly

I like Spyder (as you may understand from the post ;-)). Then again, I did not like MATLAB that much at all. Maybe I did not really get to know the interface of MATLAB well enough to think that Spyder is much worse. I have tried to use Eclipse but the learning curve was to steep for me (at least at that time). It is supposed to be really powerful once you get to know it, I’ve heard.

Thanks for your comment. I hope you find a better IDE, comparable to MATLAB, because Python is a great programming language!

]]>Looking forward to following your posts. Thanks again.

Best,

Oly

Thanks for your comment. That is great! I put together that video in a rush, glad it helped. I did not have time to figure out to get/calculate effect sizes. If, or when, I figure it out, I may put it together in a blog post and/or a YouTube video.

Best,

Erik

]]>To follow up, I implemented the AnovaRM function in the beta version of statsmodels. It worked great! Results were identical to SPSS output (I’m running two-way, 2×2 levels, repeated measures ANOVAs).

Found your YouTube video very helpful as well: https://www.youtube.com/watch?v=xzET1rpvJ_A

Thank you so much for taking the time to do this.

Best,

Oly

Thanks for the information. I will have a look at Anaconda and WinPython, and update the post with more correct information.

]]>The best Python Distributions for Windows (Which are free as well) are Anaconda and WinPython (Which is also almost portable which is great).

They both use Spyder.

]]>What about testing framework support (nosetests, pytest, etc)? How easy is it to change environments? and similar…

]]>You are right. After a comment by Damien we know what package is the culprit. It is Numpy. Pyvttbl is not maintained any more and seems only be compatible with Numpy version 1.1.x. I have created a step-by-step guide on how to run Pyvttbl within a virtual environment (http://www.marsja.se/solving-pyvttbl-error-float-nonetype-error/). I may see if I can come up with another solution if I have time. You also have the option to install the beta version of statsmodels. You can run repeated measures ANOVA with this version of statsmodels. However, you will only get an ANOVA table containing the degrees of freedeom, F-values, and p-values.

]]>-> 1240 return list(array(list(zeros((p-len(b))))+b)+1.)

TypeError: ‘float’ object cannot be interpreted as an index

I don’t know how to tell which package is causing the crash. It was previously fixed again *magically* after upgrading some packages. Now I have the same problem again.

If anyone knows which package might be the culprit, or a work around for the error it would be very helpful. This is a great package for ANOVAs and I hope I can keep using it.

]]>