Judging effect sizes from simple plots

By Patrick Shrout posted 05-15-2017 11:07

Recommend

My student Qi Xu and I have constructed a short survey where we ask respondents
to make guesses about how how large some effects are based on simple
figures that are often used in talks and publications. I am hoping to get both
researchers and skilled statisticians to give their impressions of effect sizes from
these plots.

I am writing to ask two favors:

1) Would you please do the survey and ask your students and colleagues to do it also?

2) If you are teaching undergrad or grad courses, could you ask them to do
it also? For this second request I have attached some text that might
be helpful to send to the students (below the ****** line below)

The survey takes five to seven minutes. We are not collecting names
or other identifiers. You will be randomized to either a t test
effect size of a 2 by 2 interaction, but I hope you will
only do the survey once.

Click on

https://nyu.qualtrics.com/jfe/form/SV_77lgD7z6QlXoBeJ

Thanks so much!
Patrick Shrout

****************************************************************** Script for sharing with classes below

Researchers often show results using graphs and figures. We
are interested in whether different versions of graphs provide
more accurate information about the size of study effects.

We have constructed a short survey that asks respondents to
say how large certain effects are in a standardized metric.

The survey takes five to seven minutes to complete. We are not
collecting names or other identifiers.

Completing this short survey will not give you credit to any
research requirements, and your instructor will not know if
you participated or not.

If you can help, would you please take the survey ONE time?

Click on

https://nyu.qualtrics.com/jfe/form/SV_77lgD7z6QlXoBeJ

No repeats, please!

Thanks for your help!

Patrick E. Shrout
New York University

0 comments

123 views

Blogs