American Statistical Association K-12 Poster Judging
Recently I volunteered as a professional statistician to judge the posters submitted to the American Statistical Association's K-12 Poster competition. Here are some of my thoughts on that process, and thoughts on the posters.
ProcessOverall, the process was really enjoyable, and there were coffee and muffins and fruit waiting for us. The ASA headquarters building is really nice and light on the inside, something which is hard to tell from the outside driving by. The ASA staff is also very friendly and knowledgeable.
About 15 other judges were there, and our backgrounds included government, private industry, biostatistics, survey research, and teaching. We were given a scoring rubric, and clear instructions on the scoring process. The process made it so at least 2 people were reviewing every poster. After we judged the posters individually, we would then vote among our top choices. Most of the time, at least in the group I helped review, there wasn't much debate judging the top posters. The entire process took a little over 2 hours. I reviewed around 30 posters (95% CI: [27, 33]).
My suggestions for future review sessions are, in no order:
- give graphing calculators with statistical functions to the judges and/or advise judges to bring their calculators. They would be good to use to check more involved calculations like test statistics, p-values, and a variety of other calculations, for correctness, when possible
- have posters on the tables for review when possible. I found it easier to review them this way than when taped to the wall (although, at the start I was thinking it would be the opposite)
- have little snacks and drinks upstairs like there were downstairs
- give all students feedback on what they did well, and what they need to improve upon
PostersI was impressed with vast majority of the posters. Many of them asked interesting questions, had decent graphical elements, and were statistically sound. What follows is my advice to enhance a student's statistical poster for the future. In no order:
- don't add extra dimensions to your graphs. For example, if you have a 2D histogram, don't make it 3D
- try to print out your poster on one seamless piece of paper, rather than (literally) cut and paste pieces of paper to the posterboard. Make it look professional
- try to make your poster stand out (in a good way) from your peers' posters
- check your spelling
- make the fonts and graphics large enough to be seen from several feet away
- clearly state the source of your data
- designing an experiment and then analyzing that data can be more impressive than just finding data to analyze
- clearly state the study limitations
- clearly state the study assumptions
- make sure to include labels on the axes of graphs
- make sure to state what you would improve if you were to study this subject again
- If you have two separate plots and linear regressions, say, X1 vs. Y and X2 vs. Y, consider doing a multiple linear regression of X1 and X2 vs. Y
- never state that "p-value = 0", even if that is what the computer output says
- if you're doing linear regression, show why the assumptions for linear regression are plausible
- be sure to distinguish between descriptive statistics and inferential statistics
- have a research question that is interesting, and has major impact on the world
- don't confuse H0 and Ha
- don't handwrite anything on the poster
- make the conclusion really clear to the reader
- if you have a point estimate, also show a confidence interval
- interpret r and r^2 correctly
- try not to use distracting bright colors like neon pink, neon orange, etc.
- be careful when using the computer program's default fonts and graphics
- discuss your motivation for doing the project
Good job to all volunteers and organizers - it was nice working with you. Great job to the students who worked to make the posters. I found the posters interesting and learned much from them. I hope to be able to judge more in the future!
If you enjoyed any of my content, please consider supporting it in a variety of ways: