Thread by @jsmoliga

🖐️Have you ever seen those studies claiming finger length is related to health and behavior?

There's a whole field of research exploring the ratio between index and ring finger (known as 2D:4D digit ratio).

I think it's mostly #pseudoscience!

Take a few minutes and read on!👇

Before the debunking details, a disclaimer.

I care about scientific quality+integrity, & helping people understand the nuance behind abstracts & media headlines.

I'm not intending to disparage anybody personally. Perhaps some of this research is legit, but many issues w/ 2D:4D.

Supposedly #digitratio indicates prenatal testosterone exposure.

Fetuses exposed to more male sex hormones in utero are supposed to have a smaller digit ratio (shorter index finger or longer ring finger).

But, there's very little evidence of this in humans.
#epitwitter

Some experimental studies in rodents have demonstrated that altering testosterone exposure in the womb:

- Influences 2nd, but not 4th digit
- Influences 4th, but not 2nd digit
- Influences multiple digits in complex sex-dependent manners.

Example:
journals.plos.org/plosone/articl…

We can't experimentally change hormones in human fetuses (unethical) or continuously monitor in utero hormone exposure.

So, best case is cross-sectional study. Take a sample of amniotic fluid or umbilical cord blood, measure #hormones, and then later measure digit ratio.

Main evidence for this in humans is a 2004 paper (Lutchmaya et al), which reports a negative correlation between amniotic fluid testosterone-to-estrogen ratio & 2D:4D in those children at 2y of age. Sample size = 33 (combined ♂️+♀️)

Cited >1000 times!

sciencedirect.com/science/articl…

However the correlation is:

- only in RIGHT hand (not left)
- only #testosterone to #estrogen ratio, not either alone

Bonferroni correction, but for only 2 comparisons.

Was this replicable? Let's look at other studies...

A 2012 paper reports correlations between amniotic fluid testosterone and 2D:4D, but only in females (n=54), but not in males (n=52).

Let's look at this in further detail...

sciencedirect.com/science/articl…

The paper claims that the digit ratio of both, left and right hands of newborn females (but not males) was correlated to amniotic fluid testosterone.

Impressive, right?

Not so fast.

Let's actually look at the data.

1. Right hand has p=0.09. Are BOTH hands really significant?

2. Look at large 2D:4D distribution in 0.0-0.4 range. Seems like the outliers (>0.6 w/ lower sample size) drive the correlation.

#statistics
#research
#DataScience

Let's also consider multiple comparisons (which weren't corrected for).

1. Left & right hands
2.Males & females
3. Amniotic fluid & plasma
[4. Anything not reported???]

There's more nuance to this study, but you get the point - lots of comparisons w/ a few weak correlations.

A similar study came out in 2015.

Pretty impressive long-term analysis - they had umbilical cord blood samples from birth, and measured testosterone and estrogen, and then 19-22 years later measured 2D:4D digit ratio. WOW!

Any guesses what they found?

sciencedirect.com/science/articl…

- "Digit ratio in early adulthood was not related to testosterone or #estradiol levels, or the testosterone to estradiol ratio"
- "Measures of total #androgen and estrogen bioactivity were not significantly associated with digit ratio"

While other above have done similar studies, the first true attempt to replicate Lutchmaya was done in 2021 (15+ years after!).

Details in the tweet below, but they couldn't replicate the effects.

#medtwitter
#Science
#biology

cambridge.org/core/journals/…

"...amniotic T&E that did not find evidence for these effects at 4½-year follow-up... direction of correlations observed was generally erratic, & the overall findings question the premise that second-trimester sex hormones affect the development of digit length ratios in humans"

I can't possibly do a tweet involving the 1400+ digit ratio studies in humans, but hopefully, what I have shown so far is that the evidence that adult 2D:4D is related to fetal hormone exposure is unconvincing.

Lutchmaya's highly cited study is 33 subjects, & not replicable!

Yet, the #digitratio hypothesis persists.

Often it's self-contradicting. Check out the referenced paragraph I wrote regarding the supposed link between 2D:4D and #schizophrenia.

Digit ratio may be increased and/or decreased in the left and/or right hand. Anything's possible!

Often, researchers attempt to relate 2D:4D to multiple different outcomes - and one of them happens to be statistically significant.

Example - it relates to facial, but not voice or body odor attractiveness.

Real biological effect or Type I error?

royalsocietypublishing.org/doi/abs/10.109…

One of my personal faves is the @BBC internet study, in which 200K+ people answers 200+ questions, including SELF-MEASURED digit ratio.

This has resulted in LOTS of publications. With enough outcomes, you can find an association.

One example:
cambridge.org/core/journals/…

There's a bunch of papers about how critical it is to measure 2D:4D properly, the importance of calipers and multiple measurements, etc.

So, we should also trust diffs in self-measured digit ratio at 3 DECIMAL PLACE level?

Even the authors didn't account for this on y-axis!

From that same BBC internet study dataset, one of the leading digit ratio researchers claimed that 2D:4D was related to #COVID risk, and that it should be explored to see who should social distance vs. who should not.

Yes, that happened. #Covid_19

sciencedirect.com/science/articl…

Thankfully, somebody from WITHIN THE SAME UNIVERSITY wrote a letter to the editor stating that the link between digit ratio and #COVID19 was junk science!

#fakenews
#misinformation

sciencedirect.com/science/articl…

Don't even get me started on the studies claiming that digit ratio is linked to athletic performance... let alone that coaches should be using it to select athletes for the team!

But, here's one of my favorites related to #sports performance. #running

journals.plos.org/plosone/articl…

In that study, they measured digit ratio and half-marathon completion time. Didn't consider #Training history (or even mention it).

And somehow from just that shotgun-blast figure, they connected a bunch of dots with evolution, persistence hunting, and reproductive fitness.

How the outliers in the scatterplot influenced the relationship (including potentially smaller sizes for people that fast, which influences 2D4D) is another story.

Google "Longman marathon sex" to see the absurd media headlines came from it.

Example 👇
freepressjournal.in/health/choose-…

There's so many digit ratio studies, too many to report on. Plenty on cancer, cardiovascular disease, and so much more.

But, did you know digit ratio is also associated with which @CocaCola product you choose from a vending machine?

sciencedirect.com/science/articl…

And, of course, a higher digit ratio predisposes females (but not males) to worship celebrities.

Is there anything that digit ratio measurement's can't predict?

sciencedirect.com/science/articl…

Gotta love this quote:

"...under some circumstances yet to be identified, left-hand 2D:4D systematically out-predicts right-hand 2D:4D whereas the opposite is true under other circumstances."

My explanation? Most of the findings are random chance.

onlinelibrary.wiley.com/doi/10.1002/aj…

In my opinion, the weak biological justification for digit ratio, lack of preregistered research + statistical analysis protocols, flexible analyses (multiple comparisons, various covariates), and selective reporting means that a lot of the digit ratio lit is chance findings.

This is a phenomenon known as the file drawer effect.

Here's the link to the original article on the topic, from 1979.
psycnet.apa.org/record/1979-27…

And here's my simplistic diagram illustrating it. The lit seems strong at first glance, but further examination reveals differently.

This thread just scratches the surface, and I am sure some in the field will disagree w/ my opinions. That's okay.

However, I am not the only skeptic of digit ratio research.

Check out this great article in @ScienceMagazine on the topic.

science.org/content/articl…

I hope you enjoyed this thread on digit ratio.

I think it is the epitome of the reproducibility crisis in science.

Please retweet and like it you found it interesting! Thanks!

And, as a bonus, please check out and share my latest research, published in the 2021 @bmj_latest Christmas Issue.

Digit ratio is significantly correlated to a randomly selected poker hand! Prenatal testosterone predicts good luck! 😜

bmj.com/content/375/bm…