The video above from Dr. Andrew Kaufman is hot off the press. It is eighteen minutes, and you can be sure I downloaded it. Kaufman discusses a study that was finally published in the Annals of Internal Medicine after being rejected by JAMA, Lancet and the New England Journal of Medicine (most likely political decisions from those quarters – I learned during my study of Climate Change that the prestigious journals are all on some level … bought).
The important conclusions, as noted by Dr. K, were that all analyses of results, statistically speaking, “crossed zero.” Imagine that you are studying the effects of using squirt guns to chase squirrels off bird feeders*. You have two feeders and so, when two squirrels are present, you can use one as a control group while hitting the others with the water. In the end, you find that squirrels left alone were 95% likely to return to the feeder, while squirrels squirted were 105% likely to return. The variance in the numbers, some less, some more, but crossing 100%, means that squirt guns are not an effective control for squirrels. That is my interpretation of the meaning of “crossing zero” – -2% to +5% with zero in between.
A real example from the study follows beneath the fold.
Here is that part of the study discussing results:
Three post hoc (not preplanned) analyses were done. In the first, which included only participants reporting wearing face masks “exactly as instructed,” infection (the primary outcome) occurred in 22 participants (2.0%) in the face mask group and 53 (2.1%) in the control group (between-group difference, −0.2 percentage point [CI, −1.3 to 0.9 percentage point]; P = 0.82) (OR, 0.93 [CI, 0.56 to 1.54]; P = 0.78). The second post hoc analysis excluded participants who did not provide antibody test results at baseline; infection occurred in 33 participants (1.7%) in the face mask group and 44 (2.1%) in the control group (between-group difference, −0.4 percentage point [CI, −1.4 to 0.4 percentage point]; P = 0.33) (OR, 0.80 [CI, 0.51 to 1.27]; P = 0.35). In the third post hoc analysis, which investigated constellations of patient characteristics, we did not find a subgroup where face masks were effective at conventional levels of statistical significance (data not shown).
I did the bold and underline. You can see from this that when a confidence interval ranges, as above, from -1.3 to 0.9, that one of the possibilities within that narrow range is zero. “Crossing zero” can be taken, then, to mean that there is essentially no difference between the face mask group and the control group. That is the essential finding of this study.
Give Dr. Kaufman a view, and if you are so inclined, read the paper. My only problem with it were some spooky numbers, hopefully just coincidental: 3030 participants, 42 recommended masks, 53 control participants – can all be viewed as spooky. I sincerely hope not.
*I routinely chase squirrels off our feeders with a squirt gun. I do this because it annoys them but does no harm. Once, a few years back we had a daytime bear on the back deck (they usually come around at night) that was in the process of destroying a metal feeder. I stood behind our slightly open back screen door as the bear, fifteen feet away, was doing his business. First I slammed to door to scare him, and it did startle him, but he kept on. I didn’t want to harm him, so did not want to use a cast iron pan or anything heavy like that, and so threw our broom at him. No effect. So, and this is not made up, I took the squirt gun I used on squirrels, and shot the bear in the face, hitting him square with water. The bear was looking at me the entire time I was dousing him, his look was kind of like … WTF? Finally, I took a second broom, shorter, maybe three feet long, and threw that at him. He looked at me and backed down. I had annoyed him enough. I was able to retrieve the feeder. We had to give up bird feeding that year, but it was summer anyway, and food was abundant.