Originally posted by CanadianBoltFan
View Post
- his overall grade is a sub-standard level
- data trend the last 3rd of last year was up and appears different
There are actual statistical tests to quantify the probability that the observed difference is random noise from the same population, or indicates an actual change. They are called hypothesis tests, extremely common statistical tests / tool. The null hypothesis is that they are actually the same, we want to see a P value < 5% to disprove this assumption.
Fortunately YAC described his statistical expertise, let’s let him go apply a paired analysis test cuz I’m in my joggers sipping early morning coffee …
I've returned. It seems YAC may be stuck so I performed a t-test of two groups to test the hypothesis that they have the same mean (average):
The P-value for the null hypothesis is 0.0397. We reject the hypothesis and conclude his better performance is actually better
| Group | Grade |
| 1 | 38.5 |
| 1 | 11.8 |
| 1 | 27.1 |
| 1 | 63.7 |
| 1 | 18.3 |
| 1 | 59.3 |
| 1 | 48.7 |
| 2 | 60.6 |
| 2 | 86.4 |
| 2 | 53.9 |
| 2 | 69 |
| 2 | 60.4 |
| 2 | 30.9 |
| 2 | 66.8 |
Comment