I know, I know, it’s been ages and I promised to write up my experiment. Since I miss the eclipse here in Britain (and I missed it back when we got a total eclipse too, because of cloud cover!), I’m going to finally keep up that promise. You can find part one of the experiment here. The hypothesis: that my mood improves when I read more!

# Results

Day | Number of pages read | Mood (%) |
---|---|---|

1 | 408 | 75 |

2 | 691 | 81 |

3 | 196 | 66 |

4 | 220 | 77 |

5 | 225 | 72 |

6 | 230 | 57 |

7 | 300 | 72 |

8 | 113 | 52 |

9 | 355 | 77 |

10 | 224 | 64 |

11 | 110 | 66 |

12 | 159 | 81 |

13 | 240 | 65 |

14 | 348 | 79 |

15 | 150 | 59 |

16 | 680 | 84 |

17 | 170 | 77 |

18 | 429 | 91 |

19 | 247 | 84 |

20 | 188 | 84 |

21 | 299 | 90 |

22 | 311 | 82 |

23 | 256 | 84 |

24 | 200 | 77 |

25 | 400 | 91 |

26 | 326 | 85 |

27 | 300 | 91 |

28 | 305 | 84 |

29 | 450 | 91 |

30 | 495 | 85 |

Which gives us a chart that looks something like this — the top line shows you the number of pages I read on a given day, while the lower chart shows you my mood on the corresponding day.

Eyeballing that, you can see a pretty obvious correlation between the two lines, though there are clearly other factors. But we can be more precise.

I still think a Spearman’s rank correlation is the best way to analyse these data. In a Spearman’s rank correlation test, you rank each column in order of the size of the number — so for example, day one’s page number ranking is 25, while the mood ranking on day one is 10. You then calculate the difference between them — 15 — and then square it — 225. You do that for all the ranks, and then you put the numbers in the following equation…

If that looks like Greek to you, don’t worry. *r _{s}* is what we’re trying to calculate. In English, the top bit means six times the total of the difference squared, so for our purposes that’s 10719. That’s then divided by

*n*, the number of variables, multiplied by the number of variables squared minus one — which is 26970. 10719/26970 = 0.39744160178. That needs to be subtracted from one to give

*r*. Our answer is 0.60255839822 — let’s call it 0.6, since we know from the way we’ve set up this experiment that it’s not very precise. That means that there’s a fairly strong positive correlation between the numbers: 0 would be no correlation, and a negative number would mean a negative correlation.

_{s}Nobody’s surprised: there is a correlation between the days I read a lot and the days I’m in a good mood. Whew. Glad we got that out of the way!

My next post will be all about discussing these results — what they really mean, and what I might do in the future. The discussion part of an experiment is important, even if the results look pretty self-evident!

ETA: I never got round to this. Bad Nikki! Sorry.