What his means is that instead of using 50 data points to define the trend of the data, the original "researchers" used exactly 2 for each variable. Take a piece of paper and draw X and Y axes on it. Draw two data points anywhere you want. Now draw a circle around each data point, the circles represent the degree of inaccuracy inherent in taking the measurement, and all science includes inaccuracies. Now for the interesting part - Add two more lines. Draw the first line tangent to the upper portion of the first circle and the lower portion of the second circle. Draw the second line tangent to the lower portion of the first circle and the upper portion of the second circle. The two lines you just drew represent just how widely the results can vary with only two data points. As I've often said, "Give me any two data points and I can prove anything I want to prove."
>> “Give me any two data points and I can prove anything I want to prove.”
Ahh — So that’s how the chocolate ration got “bigger” when it went from 40gm per week to 28!