Statistical thinking: What data would you need to collect to disprove the predictive power of astrological signs?
Okay. I haven't used this in class yet because it is July, and I just found it. However, I will open the Fall 2024 semester with this example. It is fun and accessible, showing how research can be used to study whether personality varies based on astrological signs.
I will start by showing them a bunch of funny astrology memes (see above). Then, I'll ask them to think of ways to design a study to prove that astrology is/is not bunk. What sort of data would they need to collect to do this?
Then, I'm going to show them this study (Joshanloo, 2024):
 |
| https://onlinelibrary.wiley.com/doi/epdf/10.1111/kykl.12395?domain=author&token=BKSRDREAX9F3BKAWGVBD |
Statsy things to share with your students:
1. Archival data: The used repurposed, vintage, federal data. The General Social Survey, specifically. Data scientists are trained to see the potential of random data sets.
The horoscope sign was simple to determine since the GSS collects birthday data. The author was able to pick this personality data out of the GSS:
2. The author then did a shit load of ANOVAs to see if scores on those scales varied by astrological sign. They did not. For an example, see the data visualization for Life Is Dull. Nothing is going on—not a thing. Like...it is meta-dull.
3. Some of the GSS data items had two or three response options, and they used Kruskal-Wallis tests on those. I think this is the first example I've ever had of that test. There are also some ANCOVAs.
4. There are non-significant/small effect size findings for DAYS. Sometimes, these are hard to find in published research. It is useful to show our junior statisticians situations when the null confirms underlying research hypotheses.
One of my mentors in grad school and I used to discuss this endlessly. She hypothesized that there were self fulfilling prophecy-type mechanisms at work, and that the strength of these might vary by generation. She felt that her generation had come of age during a time when astrological signs were considered a way to represent herself and people adopted them as core parts of their identity. I suppose this would be analogous to how many people represent themselves with their Hogwarts house--a shorthand for describing personality traits.
ReplyDeleteHer idea was that if we could collect personality traits and astrological signs and stratify the sample by generations or decade of birth, we might find differences in the relationship between personality and star sign. Along the same lines, I argued that we should also control for degree of belief in the importance and validity of star signs. If you strongly believe that you have personality trait Y because you are star sign X, then maybe you lean into it, again, that self-fulfilling prophecy type mechanism.