Today was a fun day with my statistics students. Today was the kind of day where I didn’t view my students as just students but rather as colleagues who can learn with me and from me, and I can learn from them in return.
It all started because I’m currently working on my dissertation and in desperate need for some ratings of a hot sauce I’m using in one of my studies. Previous research demonstrates that the hot sauce is rated, on average, a 7.2 out of 9 (1 = not hot at all, 9 = extremely hot). As I was mixing what is to me tasty hot sauce, my first thought was, “This isn’t that spicy.” Then I realized, maybe people in Texas don’t view this hot sauce (which is a very exact, previously published mixture of two brands of hot sauce) as hot as those in the original research (individuals from Rochester, NY). After all, we live pretty near the Mexican border, and spicy is the name of the game here. It was then that I realized, I’d better pilot test this stuff to see if people rate it as hot enough.
But where to get those ratings? I, of course, asked my lab RAs because the carryover for learning about research seems really obvious there. You’re my RAs, rate this sauce, let’s see if it’s different than past ratings. However, I suddenly realized the other day that rating the hot sauce would be the PERFECT way to illustrate a one-sample t-test to my statistics class.
What is a one-sample t-test you ask? Basically, it’s a type of statistical test you run to see if a mean you got from a sample of people is statistically different from a mean that has been previously seen across other studies, etc. So, what better way to illustrate this point than to have my students rate the hot sauce? Then, we can compare their mean (which happened to be 6.58) to a previous mean of the hot sauce (7.2) to see if those are STATISTICALLY significantly different. It’s a fun, informative way to have them actively involved in the research process. And hey, I get useful data in the meantime.
To begin this exercise, I gave them the hot sauce and had them be blind to anything about it when rating it (to get the most accurate ratings). Next class, I plan on telling them what the previously published version was (mean = 7.2), where it was collected (Rochester, NY), and asking them to create their own hypotheses. For instance, should we hypothesize that our ratings would be lower, higher, or the same as previous ratings? Why or why not? They may discuss regional differences here. For example, hypotheses could include the following: I hypothesize that our ratings will be lower because we are in Texas, where people have a higher tolerance for spicy foods. From their, we will learn how to formally test it using a one-sample t-test given the information I have (mean, number of people in my sample, etc.). I’m curious to see how it turns out and what hypotheses they come up with. Either way, hopefully it’ll get them to stop staring at me blankly when I mention a one-sample t-test. Now, they’ll get to “live” it.
I’ll put up a blog post after I finish the exercise, so stay tuned.