Monday, August 25, 2014

Regina Nuzzo's "Scientific method: Statistical errors"

This article from Nature is an excellent primer on the concerns surrounding the use of p-values as the great gate keeper of statistical significance. The article includes historical perspective on how p-values came to be so widely used as well as some discussion on solutions and alternative measures of significance.

This article also provides good examples failed attempts at replication (good examples of Type I errors) and a shout out to Open Science Framework folks.

Personally, I have revised my class for the fall to include more discussion of and use of effect sizes. I think this article may be a bit above an undergraduate, introduction to statistics class but it could be useful for us as instructors as well as a good reading for advanced undergraduates and graduate students.

Wednesday, August 20, 2014

Patti Neighmond's "What is making us fat: Is it too much food or moving to little?"

This NPR story by Patti Neighmond is about determining the underlying cause of U.S. obesity epidemic. As the name of the segment states, it seems to come down to food consumption and exercise, but which is the culprit? This is a good example for research methods because it describes methodology for examining both sides of this question. The methodology used also provides good examples of archival data usage.

Monday, August 18, 2014

If you are looking for an alternative to using good ol' Excel and SPSS to create graphs and charts, perhaps you students would like to create infographics via a free, online resource.

One such tool is Piktochart. It requires registration (via email, Facebook, or Google). It has many free templates as well as a "pro" pay to play package. Below are a few screen grabs of what it is like to personalize one of their templates with your own data. Below, I input a bit of user data from this blog into a pre-existing template.

Piktochart template

User interface for entering your own data (if you can use Excel, you can use this)

End result, with data from this blog
It is pretty easy to use, they have multiple different kinds of figures (from good old pie charts and bar graphs to visualizations that stray far from the APA style manual but still do a good job of conveying data to an audience).

This coming semester, I am adding a service learning component to my statistics lab class. We are collecting mental health awareness data from students for our university's counseling center. I am thinking that my students might use piktochart in order to create visually appealing ways to share their findings around campus.

Monday, August 4, 2014

Five Lab's Big Five Personality Predictor's prediction via status update

It might be fun to have students use this app to measure their Big Five and then compare those findings to the app (which I previously discussed on this blog), which predicts your scores on the Big Five based on what you "Like" on FB.'s prediction via "Likes"

As you can see, my "Likes" indicate that I am calm and relaxed but, I am a neurotic status updater (crap...I'm that guy!). By contrasting the two, you could discuss reliability, validity, how such results are affected by social desirability, etc. Further more, you could also have your students take the original scale and see how it stacks up to the two FB measures.

Note: If you ask your students to do this, they will have to give these apps access to a bunch of their personal information.