6

I recently wrote a blog post (original here) about a scientific study. At the end, I included some "fine print" which states, among other things:

Correlation does not mean causation. It’s theoretically possible that trans fat isn’t the culprit here, but that some other, as-of-yet undiscovered factor is the true cause of the poor memory test performance.

After reading the post, my wife asked me "Why did you write a blog post where you explain the dangers of trans fat, then at the end you say the study is wrong?"

sigh

How can I find a balance between unfounded, sensationalist reporting ("Drinking coffee makes your breasts smaller"), and confusing everyone who doesn't know how to read scientific literature?

Some of my thoughts:

  • Eliminate the "fine print" section entirely, and address accusations of false reporting if/when they arise.
  • Make the fine print truly fine, so that it doesn't appear as part of the article, but is still present for sticklers.
  • Fully explain how science works, so that people like my wife won't think I'm saying the science was bogus.

What are your suggestions?

  • 2
    Change to: Correlation does not prove causation. In other words, finding a linkage between A and B does not prove that A caused B. Maybe B caused A. Or maybe C caused both A and B. – dmm Jun 25 '15 at 15:56
  • Thanks to everyone for their input. I can only accept one answer, but I've upvoted each answer that I found beneficial. – Flimzy Jun 30 '15 at 22:16
7

Consider incorporating your fine print into the body of the article. Nothing in your Fine Print section requires any special understanding of the science involved, or even a good understanding of how science proceeds.

As a bonus: If you do that, you can drop the opening part of that section, which is about you, and not about the topic.

If the authors' report describes any of the same limitations and caveats that you do in your Fine Print section, make sure to attribute the caveats to them. If not, well, hang 'em.

A bigger problem, I think, is your own speculation about how this study might apply in situations not covered by the authors' research. If you reel in your own speculation, you will have far less need to qualify your article and your own viewpoint.

  • 1
    Thanks for the advice. I'm working now to move most of the fine print into the body of the text. I'm still concerned, though, that the "Correlation does not mean causation" phrase itself is over the heads of many scientifically uninitiated people. Fully explaining that concept to someone who's uninitiated can be daunting, in my experience. – Flimzy Jun 24 '15 at 21:52
  • True. Perhaps you could rephrase it in simpler terms. "There may be other factors involved ..." Or "There may be other explanations ..." – Dale Hartley Emery Jun 24 '15 at 21:53
  • 1
    I have moved the "fine print" into the body. Specifically, the 2nd to last paragraph under "How the study was done" and the last paragraph under "Half a letter grade." I've reduced the "correlation does not mean causation" principle to an implication, rather than stating it explicitly. I hope that's enough to ward off any nit-pickers. Revised version – Flimzy Jun 24 '15 at 22:02
5

Don't Minimize the uncertainty, celebrate it! Sensational science reporting does not worry about details just bold headlines: 'Cell Phones Cause Breast Cancer!' Responsible reporting hedges their bets: 'Link between Cell phone usage and Breast Cancer Found.' Great reporting shows what the details mean: 'Developed nations have more cell phones and breast cancer.' Now that we have the headline, lead with the uncertainty. Why is it important? Uncertainty is knowing the limits of knowledge not ignorance. You have heard it said the more you know the more you realise you don't know. Children learn one thing and being excited think they know everything, Scientists find a limit of knowledge to be a place to start. Take the next step in true learning and embrace ignorance (as that is the only way to put it in a half-nelson).

Ok I'll get off my soap box now.

3

There's nothing confusing about the paragraph you have at the end.

Given the replication rates for correlation-based studies, there no reason for the reader to believe that the study established a new insight into the effects of trans fats.

A layperson shouldn't base his idea of the effect of trans fats on a single correlational study based on self reported food questionnaires.

On the other hand you are misleading the reading when you omit issues with people being bad at self reporting food choices. You are also misleading when you make it seem like the study demonstrated that eating fat one day has a direct effect on the scores on the next day. If such an effect would exist they could run a proper controlled study to document the effect and wouldn't need self reported food questionnaires.

  • Thank you for your comments. My intention wasn't to deceive, but to bring the numbers into perspective. "1 gram of daily tf intake = 0.76 lower score on the memory test" is pretty abstract for people to understand. My goal was to make those numbers more concrete, by showing an actual amount of junk food, and an actual test score. Perhaps there are more "honest" ways to do that, than by providing my example, with a disclaimer. But that's probably for another question. – Flimzy Jun 25 '15 at 13:42
  • The problem isn't your intention. Few people who write misleading articles do that because they have bad intentions. The problem is that the example doesn't describe the study. It would be less abstract if you say that people where given food questionnaires and on average the people who eat the equivalent of French fries every day would be X worse. Daily intake means in this context taking something every day for a longer period of time. It doesn't mean taking something one day and then having an effect the next day. – Christian Jun 25 '15 at 14:59
  • I have updated the sentence in question to If, on the other hand, you're the type of person who eats one large order of fries every day, and an icing-covered cupcake at the convenience store for a snack (for a combined 7 grams of daily trans fat), you're likely to lose more than half a letter grade, and walk away with only a C+--a score just below 80%! Does that feel more honest to you? – Flimzy Jun 25 '15 at 16:30
  • @Flimzy: Yes, it does. – Christian Jun 30 '15 at 22:08
1

You have to target your writing to the background of the reader.

"Correlation does not prove causation" is a well-known slogan in scientific and statistical circles, so if you were writing for such technical people, just quoting the phrase should be sufficient to make the point. But if your writing for the ordinary layman, the odds are that they are NOT familiar with the catch-phrase. They may or may not be familiar with the concept, and they may or may not grasp it quickly. So if you want your writing to be clear to the layman, you have to spell it out.

Sorry, I didn't read your article, but if this is an important element of your point, then I think you should spell it out in detail. If it's a side issue, if in this case you think that the correlation DOES indicate causation or whatever, then a sentence or two of fine print disclaimer might be sufficient.

I'm a software developer by profession. When I'm talking to other software geeks, I may say something with bunches of technical terms and acronyms with no explanation, because I expect them to know what I mean. But if I'm talking to non-computer people, I use simpler language, metaphors, and explanations.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.