The 4-Hour Body_ An Uncommon Guide to Ra - Timothy Ferriss [200]
1. IS A RELATIVE CHANGE (LIKE PERCENTAGES) BEING USED TO CONVINCE?
This concept is best illustrated with two potential news headlines.
“STUDIES SHOW PEOPLE WHO AVOID SATURATED FAT LIVE LONGER”
Should you start avoiding saturated fat?
First, find out exactly what “longer” means. Based on available data, it turns out that reducing your saturated fat intake to 10% of daily calories for your entire adult life would add only 3–30 days to your lifespan. Considering this, is the trouble worth it if a rib-eye is one of your pleasures in life? Probably not.
“PEOPLE WHO DRINK COFFEE LOSE 20% MORE FAT THAN THOSE WHO DON’T”
Should you start drinking coffee?
Leaving aside the question of whether or not this is an observational study (discussed next), it’s worth looking at that mighty impressive 20%.
Relative increases or decreases, most often expressed as percentages, can be misleading.
Relative isn’t enough. It’s critical to ask what the absolute increase or decrease was—in this case, how many pounds of fat did both groups actually lose, and over what period of time? In most cases, percentages are used in media and sales brochures to mask the fact that changes were minuscule.
If it were 0.25 pounds lost for the control group and 0.30 pounds (20% more) for the coffee group over eight weeks at three cups per day, is picking up the coffee habit worth the side effects of high-dose caffeine? Nope.
Distrust percentages in isolation.
2. IS THIS AN OBSERVATIONAL STUDY CLAIMING TO SHOW CAUSE AND EFFECT?
This is the mother lode. If you learn just one concept in this chapter, learn this one. It’s the cardinal sin.
Observational studies,3 also referred to as uncontrolled experiments, look at different groups or populations outside the lab and compare the occurrence of specific phenomena, usually diseases. One example is the often misinterpreted “China study.”
Here is the most important paragraph in this chapter:
Observational studies cannot control or even document all of the variables involved. Observational studies can only show correlation: A and B both exist at the same time in one group. They cannot show cause and effect.4
In contrast, randomized and controlled experiments control variables and can therefore show cause and effect (causation): A causes B to happen.
The satirical religion Pastafarianism purposely confuses correlation and causation:
With a decrease in the number of pirates, there has been an increase in global warming over the same period.
Therefore, global warming is caused by a lack of pirates.
Even more compelling:
Somalia has the highest number of Pirates AND the lowest Carbon emissions of any country. Coincidence?
Drawing unwarranted cause-and-effect conclusions from observational studies is the bread-and-butter of media and cause- or financially-driven scientists blind to their own lack of ethics.
Don’t fall for Pastafarianism in science.
It is critical not to take advice based purely on observational studies. In 2004, a commentary published in the International Journal of Epidemiology titled “The hormone replacement–coronary heart disease conundrum: is this the death of observational epidemiology?” highlighted the dangers of doing so. Observation of one group of women using hormone replacement therapy (HRT) showed lower heart disease, and media and HRT proponents were fast to promote this sound-bite conclusion: HRT reduces heart disease! Sadly, randomized and controlled trials (RCT) later showed no protective effects, and even a slight increase of risk, for heart disease among those using HRT.
How was this possible?
It turns out that the observational studies didn’t sufficiently account for different socioeconomic statuses between groups, or for the influence of doctors selecting women for HRT who were less predisposed to heart disease to begin with. The latter is an example of how failing to randomly assign subjects to groups (randomization) leaves observational studies open to bias from experimenters.
The 2004 hindsight commentary stated, rightly:
The differing results between