managing an LCHF lifestyle in Israel
Media is headlining some new research that reveals that eating such and such will cause to that or that (good or bad) outcome. You’ve decided to not trust the headlines and want to check the latest findings by yourself. You look up the published paper and start to read. And stop, wondering, what language could this possible be? It’s OK. You don’t need a PhD in research language, just some practice.
Here’s a short glossary for you who better want to understand research papers:
(This is just a start and I will add on more terminology with time.)
Blind tests – Single blinded is when the participants don’t know if they are in the testing group or the control group. Double blinded is when the researchers don’t know either. The purpose is to minimize bias’ and expectations’ influence on the results.
Control – The group or variable that goes on as usual, not tempered with, and serves as reference for comparisons.
Correlation and Causation – Have you ever heard “correlation doesn’t mean causation”? Correlation is when you can see a trend of things happening together but it doesn’t mean that one causes the other. E.g. you can see that eating more ice cream in the summer correlates to more drowning accidents. But the cause could be the warm weather, having kiosks close to pools and beaches, or any other variable influencing both ice cream eating as well as water dipping.
Risk – Absolute risk talks about an actual number, e.g. 3 in 5 or two in a thousand, whereas relative risk is about likelihood. Here’s an example quoted from “Death by Food Pyramid” by Denise Minger (p.72). She talks about some research findings that stated “a doubled risk” which sounds pretty frightening, until you look at the absolute risk numbers: “…the doubled risk only meant that four out of one thousand people – rather than two out of one thousand people – got …”. So four certainly is twice as much than two, but you got 99.96% chance of not being affected of whatever it was they checked.
Significant – In our daily speech we usually associate the word with ‘meaningful’ or ‘important’ but in research it means something else: Statistically significant means that the numbers themselves probably are correct and that 1+1 actually equals 2. Clinically significant is more interesting as it takes into account the number of instances needed to make something happen. If you need to give a certain pill to 10,000 people to save the life of one, this might not be the best overall treatment.
Here are some good questions to have in mind when reading a research paper:
1) How was the study conducted: is this a clinical or observational study? Blinded? Over what time period? How many participated? etc.
2) What was the hypothesis checked and what were the conclusions compared to the results? Reporting other findings is not bad science per se, on the contrary it can promote new hypothesis, but drawing conclusions when there only are correlation is definitely a no-no.
3) Who were the researchers? How were they founded? Are there any vested interests in general, or benefits if receiving certain results in particular?
Hope this helps some and happy research paper reading!