is this data significant?

26 Jul

If the word kurtosis brings to mind Grandma and a desperate need for Listerene, then we’re probably in the same boat when it comes to statistics. I love numbers, data, infographics, etc, but statistical tests have always challenged my patience intellect. For the longest time, I felt p values were like those magic eye paintings where you stare at them for awhile in frustration, and then when you finally glimpse a bit of what might be the image (is that a triceratops? sailboat?), you blink and the image vanishes.

magic eye image

magically annoying

Needless to say, this summer I bit the bullet and took a class in experimental design, methods, and stats as part of my masters degree. Totally changed my understanding of how statistics fit into science. I’ve still got plenty to learn, but I’ve resolved to take my newfound appreciation for statistical analysis and revamp my lab curriculum.

I mentioned in a previous post my recent change to teaching evolution first. One of the first lab activities I usually do is something to demonstrate the variation that exists in populations (e.g. measuring height/weight data in class, etc.).

I’ve taken this idea and created a scenario-based lab where students have to figure out a way to compare two populations. Here’s a snippet:

“Just the other day your boss, Pedro, was thanking you for saving the store money after you found a new local mealworm distributor offering cheaper prices. A few dead reptiles later, and your boss is threatening to fire you. He claims the new mealworms you ordered are different, and he’s blaming the new worms for his dead reptiles. And you thought all mealworms were created equal.

You need to design an experiment or experiments to determine if there is a significant difference in the two populations.

How will you ensure standardized data collection procedures? What outside factors might influence your data collection? How will you use the data you collect to make a comparison? How can you quantify the differences between the two populations (or lack thereof)?”

This of course leads into all sorts of interesting discussions related to statistics. This could be a little statistical appetizer or a whole can of worms, depending on the direction you want to go. By my count, at least three different statistical tests could be used to compare the samples, one of which barely requires any calculations (tukey’s quick test).

Like what you see? Download the whole shebang here: MealWormStats

(CORRECTION: Mann-Whitney U test actually compares the median values of two populations, not the mean and has been corrected in the file posted above.)

Feedback welcome.

Advertisements

3 Responses to “is this data significant?”

  1. michaeleriksson July 26, 2010 at 2:13 pm #

    As an aside, the experiment could give a false negative even with the best of procedures: It would be quite possible that just that one batch was, in some way, dangerous. (Respectively, if you still have a non-trivial amount of the batch available, that just a part was dangerous.) Further, if not all reptiles died, it could be that just the ones that died were over-sensitive to some aspect of the worms, which implies that it may be necessary to buy new reptiles too.

    The world of scientific investigation, with or without statistics, is a fascinating place…

    • benpaulson July 26, 2010 at 3:18 pm #

      Michael,

      Great point. That’s something I was thinking too but forgot to include in the teacher’s notes, thanks for pointing it out! I think it presents a nice opportunity to talk about how generalizable the results from any single study are. Of course, even if the batches of worms did differ significantly, that only demonstrates correlation, NOT causation. The few dead reptiles may have been caused by a total unrelated infection or some other unknown cause. I like the idea of presenting students with results that aren’t black and white because the real world definitely isn’t. You hit the nail on the head, the world sure is a fascinating place!

      Ben

  2. Britt Gow July 27, 2010 at 3:49 am #

    Hi Ben,
    Followed you here from Twitter – great to find other science teachers in the blogosphere. Love this mealworm prac. for statistics! I recently worked with a Tasmanian teacher and his class on “Potato Olympics” – a maths and science activity that investigated correlations between potato attributes (height, circumference, mass and volume) and performance in student-designed events. It was a great hit with year 6/7 students. Led to discussions about forces, objective vs subjective measurements, significant differences and more. You can check it out at http://technomaths.edublogs.org.
    I look forward to more of your posts!

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: