I've had my site now for a few months. And between full-time work, part-time school, and full-time family, finding time to blog has been a challenge. I've started several posts, but have a bad case of "I-can't-finish"-itis.
One of my classes for this spring term of school is an Intro to Stats class. Our first group assignment addressed the turn capacity of a busy left-turn lane in the GTA. The second of two group assignments allowed us to pick a topic and perform hypothesis testing. What fun! Choosing a topic wasn't easy; figuring out what hypothesis test to use was even harder. There were so many directions we could have gone with this.
For its simplicity, we chose to test Nabisco's claim that Double Stuf Oreo cookies actually had "double-the-stuff". It stemmed from a controversial viral test performed by a high school math class in 2013 and blogged about. They took a sample of 10 original Oreo cookies and another sample of 10 Double Stuff Oreo cookies, took the average weights of the cream fillings by cookie type and determined the ratio of Double Stuff to original Oreo cookies to be 1.86...not exactly double the stuff.
In our test, we used sample data from statscrunch.com that contained a sample of 94 original Oreo cookie filling weights and 79 Double Stuff Oreo cookie filling weights. We could have easily made our own samples, but I'm sure our thighs are thanking us for using an existing data set instead.
Since we had two different samples we determined, based on info we attained from our stats class and via our textbook (Devore 2016), that we needed to perform a t-test for two independent samples. What we didn't know yet was whether we had equal or unequal variances. After performing the f-test, we determined that our variances were unequal. We proceeded with the t-test for two independent samples assuming unequal variances.
We concluded that we cannot reject Nabisco's claim that Double Stuf Oreo cookies contain double the cream filling compared to original Oreo cookies. Our conclusion differs from the original high school test in that we had a larger sample (think Central Limit Theorem) and we took variation into account.