you're reading...


Silent running


My blog has been quiet for almost a year now, so I wanted to share why I haven’t been blogging. Psychology is living in interesting times. Many of the statistical methods that are established in the field are being challenged and this has made the day-to-day research life of psychology researchers extra-challenging, but I feel that it’s an opportunity for the field to become stronger.

My wonderful colleagues at Massey (Matt Williams, Michael Philipp, and Stephen Hill) engaged with this challenge way before me and have been chatting about Bayesian stats and the problems associated with p-values and null hypothesis significance testing (NHST) for a few years. I knew that I was lagging behind, but it wasn’t until I was part of a conference symposium with them that I realised how far I was behind. Over Christmas this year I decided that I needed to change. My main barrier was that I was embarrassed because my grasp of the problems that they were discussing required a much greater understanding of basic statistics than I had at the time (I, like many psychologists, only used a limited number of statistical methods–except when a reviewer demanded that I used some exotic method to address a concern that they raised). This time last year I worked with ANOVA, t-tests, and correlations, and discussed these using p-values, correlation coefficients, and the occasional confidence interval).

This Christmas I switched from SPSS to R. In the past I ran all my psychophysiology analyses using thousands of lines of SPSS syntax (my laptop would spend an entire night churning through my raw datafiles to pre-process my facial muscle data). But after attending a Software Carpentry course I decided to convert to either Python or R. For years I have wanted to make my emg analyses and datasets accessible for researchers who want to replicate my findings or use these rich datasets as a starting point for their own research. I played with Matlab for a while, but this isn’t a viable option as Matlab is prohibitively expensive. I settled with R for a few reasons: 1. R is free and open source, 2. R seems to have attracted many psychologists and statisticians, 3. There are some psychophysiology and signal processing packages in R that I wanted to use, 4. Did I mention that R is free? and supported by wonderful people who provide amazing tools just for the love of it. Now that I’ve got past the basics of opening and manipulating data in R, I’ve started to learn to use it to model my psychophysiology data using Bayesian mixed models (that deserves its own post).

So I’m going to be quiet for a little while, but I’m excited that I am close to publishing quite a few exciting findings from datasets that I have been collecting for a few years. I have committed to publishing the datasets and syntax for these publications in a data repository.

I’m cautiously optimistic that psychology is cleaning up its act. Either that or I’ll be tarred, feathered, and run out of town 😉


No comments yet.

Post a Comment

About Pakiri Lab

Most human behaviour relates to emotion: either being influenced by emotions, resulting in emotions, or both. At Pakiri Lab we research these experiences using a range of psychophysiological lab based techniques, such as facial muscle activity and heart activity.

View Pete's research profile here

Pakiri Lab A Massey Blog