Another dot in the blogosphere?

Posts Tagged ‘correlation

This refrain might seem old, but it should be seen as timeless instead — correlation is not causation.

Case in point, the article embedded in the tweet highlighted how you can correlate almost anything with enough data and participants. Just because eating potatoes is correlated with “negative” technology use does not mean that one causes the other.

The use and integration of everyday and educational technologies are not monolithic. They are complex phenomena that cannot be reduced to soundbites or clickbait.

As the author pointed out in his article, if one is to explore the possibilities and problems in this wide field, one has to first be a student of cognitive development, epistemology, sociology, moral philosophy, etc. And yet these are so easily circumnavigated by a combination of misplaced correlation and fear of change.

Video source

This video is as much about misconceptions surrounding screen time as it is about:

  • Reading beyond headlines
  • Understanding how newspapers are not journals
  • Distinguishing engagement and accuracy; statistical significance and effect size; correlation and causation

It also illustrated how large sample sizes can make tiny effects statistically significant even though they have no practical significance.

For example, the video cited a study in Nature Human Behaviour that had a sample size of 355,358 adolescents. The video (also this article in Vox) highlighted how the study found that “wearing eyeglasses and eating potatoes also had significant yet small negative effects on teens’ wellbeing”. And yet we do not vilify either.

Add to that the fact that researchers have to decide where cut-offs are that distinguish statistically significant effects from non-significant ones (e.g., P value 0.01 vs 0.05). The same researchers or the agencies they work for might also make cut-offs like recommended screen times of no more than one hour before age five, even if the evidence does not support strict limits for any age groups.

TLDR? Newspapers oversimplify complex phenomena by providing easy answers. Real learning is not in taking these answers at face value. It happens when you explore nuance and depth instead.

One of my pet peeves is how some people confuse correlation with causation. Sometimes I cannot blame them because they were taught to think that way.

Video source

The SciShow video above highlights one common example. As a former biology student (and teacher), I was taught (and taught others) wrongly that aching muscles are due to lactic acid buildup.

Not only is the buildup due to lactate — a base that accepts protons — aches are only correlated to the buildup. The lactate might build up, but it does not seem to cause the aches; the actual cause is not yet known for sure.

This video is not just useful for highlighting how scientific facts change, but also how scientific thinking takes place. It is the latter that creates content and changes it. It is the thinking that needs to be modelled and taught, not just the content.

Anyone who needs to process scientific, medical, or social science research that involve correlations needs to watch the video below.

Video source

As the video highlights, the number of drownings can be correlated to the release of Nicholas Cage movies, but this does not mean one causes the other.

Journalists who like reporting whether certain foods are good or bad for you need this video.

People who read what these uninformed journalists write need to watch this video.

Anyone who might have heard someone else declare, “Correlation is not causation!” needs to watch this video.

Watch this video!

Say it, repeat it: Correlation is not causation. This is a tenet in critical thinking and research literacy.

Case in point:

As long as news agencies and papers continue to publish drivel without explaining the tenet, they are peddling misinformation. There are many other factors that contribute to susceptibility or resistance to disease.

There is nothing wrong with promoting healthy eating by feeding your body with fruit. There is everything wrong with perpetuating uncritical thinking by feeding your mind with misinformation.

As long as news agencies and papers practice such lazy and unethical publishing, teachers and parents have a duty to use these examples to model critical thought.

Rick and Morty is an animated series that is waiting for its third season.

It is not for the faint-hearted because it makes you laugh from openings you might not realise you have. It can be rude and crude, but -oh-so intelligent.

Video source

So it should come as little surprise that it is possible to use Rick and Morty to illustrate human cognition, confirmation bias, and how correlation is not the same as causation.

This just goes to show how just about anything can be used to teach anything else. The key is an educator who can think both creatively and critically.

Our daily rags sometimes do us a disservice by publishing articles like this.

A headline that reads “Eating too much fish while pregnant raises child obesity risk” is not only inaccurate, it is also irresponsible. The researchers highlighted that there was no direct link and said that making such a hypothesis was “speculative”. The study did not prove causation; it only suggested correlation.

The headline is what grabs eyeballs. It is clickbait based on fear or worry.

If not scientifically or research literate, the layperson typically does not distinguish between correlation with causation. Perhaps we need a SkillsFuture course on this because it is a valuable lesson in lifelong learning.

If not, then we might ponder the observation of one of the readers: The Japanese consume a lot of fish, and presumably that includes pregnant women, but they have a relatively low obesity rate. So what gives?

Rising above irresponsible reporting, I wonder if literacy in schools includes the sort of critical thinking that 1) distinguishes between correlation and causation, and 2) encourages questions with counter examples and data.

Is such literacy relegated to “cyberwellness” programmes or is it integrated in the context of actual content?

I spotted this on Twitter recently.

The layperson who sees this might conclude that you are more likely to produce Nobel laureates if you eat more chocolate.

A better informed person would have learnt that 1) you cannot assume the data to be accurate, 2) if the data was accurate, correlation is not the same as causation, and 3) description is not prescription.


Does anyone raise an eyebrow when they read headlines like “Happier youth ‘spend less time online'”? [article archive online] [article archive PDF]

I am not surprised not because I agree with the statement but because I have come to expect the Straits Times to publish such misleading information as truth.

I have said it before and I will say it again: Correlation is not causation.

There are so many things that can cause unhappiness. Taking a poll and reacting to poorly drawn conclusions can cause unhappiness.

You can sample a segment of a population and collect data on, say, the brand of toilet paper they prefer, and the incidence of crimes like peeping toms.

The data might reveal a strong correlation between those who like Itchy Bum brand and the likelihood of being watched by a peeping tom. But you cannot conclude that one causes the other. You can only say there is some relationship  between the two variables.

I have a second problem with the way the data was presented.

The article states that respondents who rated themselves “happy” spent 5.4 hours online each day. Those who rated themselves “unhappy” spent 5.8 hours online. Ignoring how happiness or unhappiness was determined, is 0.4 hours (24 minutes) statistically significant? Is is pragmatically significant? Just because there is a numerical difference does not make it significant.

Worse still, a layperson who reacts extremely to this might take the results as prescriptive. They might take the article as advice on what to do instead of analyzing it more critically.

Might a parent or teacher then insist on 5.4 hour cap of online activity? This is not as ridiculous as it sounds when you consider how people react to articles about whether to drink wine or eat a particular type of fish for health reasons. These folks switch diets, brands, and behaviours at the drop of a hat.

And at the drop of a physical newspaper at their doorstep. This newspaper is the same one that would prefer to bash a competitive medium (online bad, paper good) instead of reinvent itself to fully take advantage of positive change.

The same newspaper that had an opportunity in this case to educate but chose to misinform instead. Present the same information in an online forum or Facebook and watch what happens. The misinformation gets discussed and eventually might get corrected.


In this TED talk, Richard Wilkinson presents some quick data and analysis on the possible relationships between economic inequality and other concepts like health, lifespan and trust.

Singapore prides itself at being first or among the top few at many things. Among developed countries, we happen to top the charts in disparity between the very rich and the very poor.

Screen capture from TED video. Click to see larger version.

While it is important to note that Wilkinson’s statistics are correlational and not causal, there is no pride in being the country with the greatest economic disparity.

[There was more to this entry but I thought my rant started to wander off track. The visual, if verifiable, is enough to provoke thought.]

Click to see all the nominees!

QR code

Get a mobile QR code app to figure out what this means!

My tweets


Usage policy

%d bloggers like this: