Here’s how this statistical discovery was made. Our trio ventured to Saudi Arabia and gathered 150 patients (half men) suffering from asthma, then the same number of heart disease and cancer patients. Then they found 250 folks (half men) who did not so suffer. These healthy were culled from “nursing and administrative staff departments” at the same hospitals were the sick people were. All were between 40 and 60.
They asked participants “Were you beaten as a child?” (they also looked at insults). Curiously (Table 2), 24 of the 450 sickies (5%) said they were beat “At least once a day.” Maybe that’s true, but golly that seems high. Anyway, 78% of the asthmatics were beat at least once; 93% of the cancerous were; and 75% of the cardiac patients were. Lo, only 55% of the healthy nurses and administrators were beaten.
Theses numbers were fed into a fancy statistical model and it was decided that lack of spanking makes one healthy; or to put it another way, spanking causes cancer, asthma, and heart disease. Case closed?
Perhaps there were other factors at play? Turns out (Table 1) that the healthy folks were on average younger than the sickies: 47 versus 50 (cancer), 53 (heart), and 50 (asthmatics). And then the healthy nurses and administrators had more eduction on average, too: 4.5 (higher is more) versus 3.4 (cancer), 2.6 (heart), and 3.4 (asthmatics).
Same thing turned out to be true for the education of the participants’ mothers: healthy people had more educated moms. And then the same thing turned out true yet again for the education of the participants’ fathers, etc. Any other differences? We do learn that healthy people are nurses and administrators. We never learn what the sick people do.
The authors at least recognized the discrepancy in measured variables and used the controversial technique of “propensity analysis” to “adjust” the samples between the healthy and sickies, so that age and education would be “balanced” between the groups. This balancing still gave the same results: spanking causes cancer, etc.
The difficulty with propensity analysis is that it can only “balance” that which it sees. Here it only takes notice of age and education. Not job, and not other sociodemographic characteristics, and not other circumstances that would lead to differences in health between nurses and administrators and hospital patients.
There is some question of measurement error: these are all self reports and there is some likelihood interviewees were telling interviewers what they wanted to hear, which might account for the high rate of people who claimed to have been beaten every day. Perhaps less educated people would over-estimate the number of beatings compared to more educated people?
Or maybe the number of beatings are accurate, but that nurses are just people who suffer fewer beatings. Isn’t that more likely than spanking causing some mechanism to at once cause more cancer, and then, via a different pathway, more asthma, and then by a third pathway, also cause more heart disease?
And then there are some strange results. People who claimed to have been beaten at least once a day their whole childhood long actually had a lower risk of cancer than those not beaten. Beating protects against cancer? Or, do much for a coherent dose-response effect.
Good news is that the authors claim, “Although our research was carried out in Saudi Arabia, it will almost certainly generalise to other cultures.” Why? Answer: why not?
The best news is that “further research is needed” especially with regards to “reversability.” Perhaps two hugs for every one swat? Three “atta boys” for every “you brat!”?