Fairfax County Regional Science and Engineering Fair
Recently I volunteered as a professional statistician representing the Washington Statistical Society to judge the projects submitted for the Fairfax County Regional Science and Engineering Fair.
This science fair was held at the Robinson Secondary School in Fairfax, Virginia. I was really impressed by the high level of organization involved and also the quality projects the students worked on.
One key difference from the last time I judged (for a poster contest) is that this time the students were present. We could ask the students questions to see what they were thinking, give them advice for future studies, give examples from our own work (our successes and mistakes!), and so on.
There were many judges, probably several hundred in total, representing different organizations. Five of us were representing the Washington Statistical Society. We were mainly looking for examples of exceptional statistical thinking in the students' projects.
There were about 400 projects to review, and we split up the assigments based on the topics were were most interested in. Each of us had roughly 80 projects to review. We did two passes. The first pass was each of us marking down which projects in our assigned sections had high quality statistical content and deserve a closer second pass. For the second pass, we'd pair up and spend more time talking to the students about their statistical thinking.
After we were done with the passes, we'd meet and discuss our 1st, 2nd, and honorable mentions. They received various awards such as money, a year subscription to Significance magazine, a statistics book, certificates, and letters. Because of the high quality work the students did, sometimes it was very difficult to choose! The entire review process, from start to finish, took ~3.5 hrs.
In the projects I judged, and in the projects I saw but did not judge, I noticed statistical issues that could be improved for future studies. In no particular order:
- Put titles, labels, and units on your graphs
- Looking at the means is not enough. We also need to look at variation
- Have justification for using a certain alpha value
- Interpret p-values correctly
- Don't apologize for not rejecting H0! Sometimes not rejecting a null hypothesis is important.
- If you detect a difference in means, the next step is to estimate that difference
- Make sure you can explain the practical significance of your work, not just the statistical significance.
- Know the assumptions of your methods and check that they hold
- Check your spelling
- When handling biological specimens, realize handling them and stressing them out may cause them to die. Are you ignoring this "missing data" in your data analysis?
- Make sure to state what you would improve if you were to study this subject again
- Interpret r2, the coefficient of determination, correctly
- If you have data over time, consider using time series methods instead of just simple linear regression
If you enjoyed any of my content, please consider supporting it in a variety of ways: