I’ve finally finished my exams, and after a few days to decompress, have been thinking — a dangerous pastime, I know. But that thinking led me to a way to tie my two blogs together, practice running an experiment, practice writing up an experiment, and answer a little question I have! So over the next month or so, there’ll be a sort of series of posts, writing up this little experiment of mine. Aside from the style, which will remain chatty, I’ll try and be thorough in documenting the process of this experiment. If you’re not used to running experiments or reading about them, this might give you some basic insights into the process. Ready?
I’ve noticed over time that my mood seems to correlate strongly with how much I’ve actually managed to read on any given day. It’s not perfect: a terrible event such as a death in the family will obvious ruin my mood, and even minor events like failing to find the cheese I want in the store (I love cheese, sorry!) can put a damper on things. But broadly speaking, I believe — and other people have observed as well — that my mood goes up if I’ve managed to read a lot, and down if I haven’t.
When you’re running an experiment, it’s important that the first thing you decide are your hypotheses. That way, once the experiment is done, you can use a statistical test to see whether the results you got were significant, or if they could be purely the result of chance. Remember, a fair coin flipped a hundred times could land on heads every time, but the probability of that happening is low. You always have at least two hypotheses: a null hypothesis, and an experimental hypothesis. The null hypothesis is always that no correlation exists, while the experimental hypothesis is that there is some significant correlation or difference between the variables you’re recording. Using your statistical test, you can tell whether your results are significant enough to accept the experimental hypothesis (making it a theory!).
My null hypothesis, then, is that the amount I read does not correlate with my mood on any given day.
My experimental hypothesis is that the amount I read does correlate with my mood on any given day, and that correlation is positive.
Given what I know so far, I’m planning to use a Spearman’s rank correlation coefficient test on the results. This might change as I get the data in and look at it, but since I’m writing this ahead of getting results (which you wouldn’t normally do for a published report), I’m not positive about that.
There are two variables I need to measure: the amount I read on each day, and my mood on each day, and I need to make sure each pair of results stays paired, otherwise they’d be meaningless. I plan to measure my mood using Moodscope, which measures various mood factors on a scale of 0-3 based on your responses to cards which you can flip around and think about as you record your mood. I’ve used this for some time, so I’m pretty sure I use it consistently, even though results between individuals might vary. I’ll do the mood test at the end of the day, and any reading after completing the mood test counts toward the next day.
I’m going to quantify ‘amount read’ by the number of pages read, since either ‘chapters’ or ‘books’ would be very inconsistent from book to book — some are long, some are short, some have many chapters, some have none, etc.
There’s a possibility in both cases that there could be what’s called confounding variables in the data. In this case, it could be that on two different days I’m feeling equally (say) determined, but on one day I happen to click ‘3’ and on another day, ‘2’. Because mood is subjective and I’m human, inconsistencies like this are likely to creep in. The problem with the page count variable is that not all books have the same size font or the same margin settings, and some are more difficult to read than others. Non-fiction, for example, might take me longer to read. For the sake of this experiment, I’m going to ignore these potential variables.
Another issue that may be harder to unpick is the fact that if I’m not reading as much, maybe I just don’t have enough time or energy — and that could be because of low mood, instead of causing the low mood! Or it could be that when I’m busier, the tasks tend to have an impact on my mood: I’d find time to read on a normal working day, so a busier working day might lead to me reading less, and that business might be the cause of my mood. I’ll try to unpick some of this later in the discussion of whatever results I get.
You may have also noticed another inherent problem: I can’t blind this study or have a control. I’m scoring it myself, and I know all the moving parts. Normally, you’d want a control — a situation that gives you the baseline — and for the participant not to know whether they’re an experimental subject or a control. In this case, that’s not possible.
The next thing to worry about is sample size. The more data you have, the better you can distinguish outliers. For example, I might read 560 pages in a day, and then receive bad news at the end of it just before I record my mood. But I’m not likely to get bad news in that way every day, so having more results reduces the effect of that aberrant result. The bigger the sample size, the better. I’m going to start with a month of data and see what happens.
Now, given this is a really casual tiny study just on myself, but I need practice in this sort of thing, so if you notice a flaw, give me a shout. I might have already thought of it, but it won’t hurt me to think it over again!
Note: all the images are via StockSnap.io; I normally use them for all my stock photo needs.