The seven or 22 countries study - what is the issue?
At 17:16 minute of the video Dr Lustig criticized Ancel Keys that he cherry-picked 7 countries instead of all 22 countries included in the study when examining the link between the mortality due to the coronary heart disease and the consumption of either sugar or fat :
While I do not know exactly why this was done (I am open to suggestions) I have no problem in seeing the trend on the right diagram, too. The correlation simply IS there. And it is not a MINOR correlation as Dr Lustig announced. The correlation follows the same curved line, only the points are more widely distributed, some above and others (albeit slightly more in numbers) below the main curve. So what is the fuss about this? To me, the line looks like almost the best fitted line across the 22 country points.
Straight after, Dr Lustig points at outliers that were not included at all:
I again do not know the reason why these were not included. The reasons could be various: either a poor validity of the dietary assessment, high energy expenditure in the tribe, distinct physiology as a result of isolated evolution of the tribe, on which our general physiology does not quite apply... anything could have contributed and become a reason that these distinct societies were not included at the end. Ancel Keys was an experienced and endorsed scientist and researcher. Although I cannot comment on the allegations about a possible background lobby by the industry, his professional expertise is undeniable.
In fact, Dr Lustig said that these few indigenous tribes would blow a hole into Ancel's hypothesis. He further said that these tribes only consume fat, have "no carbohydrate at all" in their diet (really?) and suffer the lowest incidence of heart diseases. Maasais also consumed honey (sugar); the diet composition of Rendille tribe is hard to find, but they are nomadic, which means they also eat what they find, which is not always meat or fat; the Tokelau tribe obtains most of their energy from the coconut oil whereas Inuits get a lot of omega-3 fats from their cold-sea animals based diet.
Indeed, Ancel Keys believed that "different kinds of dietary fat have different effects on health", according to Wikipedia. This may well apply on the intake of saturated fat as on the poly-unsaturated fat, or different length of the saturated fatty acids, since coconut oil, albeit mainly saturated fat, does not seem to have the same effect as the saturated fat of animal origin. Even butter (milk fat) has a different fatty acids composition than lard, especially when the cattle is freely grazing instead of being fed with grains. Ancel may not have known all this in the past, but he was certainly on the right track.
Going back to the diagrams, having outliers in the data is nothing unusual and statisticians can handle that. Outliers are commonly excluded when they disturb otherwise obvious trend in the data and often examined separately. Therefore there is nothing like blowing a hole into the Ancel Keys hypothesis as Dr Lustig tried to imply.
Overall, it is apparent that the positive trend was there even when including all 22 countries and I believe it would have a high statistical significance and a correlation coefficient. I have not read the study myself and I welcome any relevant comments and explanations from others.