I was thinking about how to keep my skills tight after finishing my degree and I decided one of the best things I could do was to consider how to explain statistical “stuff” to people who don’t know or don’t care.
I think I’m going to start with some simple statistical tests that work within the normal distribution.
There are a lot of variations on this test, but in it’s essence, the t test asks “is this different?”. It could be:
- are these two groups different from each other?
- did this thing we did cause any difference from the original?
- is this group different from what I thought it would be?
T test will tell you how extreme your difference is, and you can determine using your expertise whether it’s extreme enough to indicate a difference. Industries can vary on what might be “big enough”.
You’re a retail company who’s considering releasing a new clothing line, but you want to test whether it is a hit with consumers. You know based on your extensive experience that you need to move $1000 per store per week in order to make it worth your time, so you select 50 stores in a variety of markets and set up a test for a week of the product. After the week, you have 50 stores sales data for the week. You can run a t test to determine if this is significantly different from the $1000 per store per week you had hoped to achieve.
The t test is one of the simpler tests, though there are a lot of variations, but the root of the test is that I have a small sample, is that different from my population or not. It takes into account your sample size and how much variation is in your sample to estimate whether it is different.
- independent random samples
- normal distribution
- interval measurements
- depending on the test, there are assumptions made about the variance