Natural experiment of history
Case control study design
Say you want to find out if Ledermix can reduce post-operative pain in Endodontics. We’ve got to compare Ledermix to something. Standard intra-canal medication is calcium hydroxide. The study, therefore, will be a comparison of post-operative pain of Ledermix and calcium hydroxide. Say you have a whole bunch of cases that have been completed and you have records of what cases were medicated with what medication. Also you have record of which cases had post-operative pain. We divide patients into two groups one with post-op pain and the other without. Then we go through patient’s record in each group to find out how may patients had Ledermix and how many had calcium hydroxide. Based on this info we can calculate the strength of the association between medication and post-op pain. The result of the study will be stated along the line of patients with calcium hydroxide medication is ... time more likely to get post-op pain.
The advantage of this design is that it can be conducted inexpensively. The disadvantages are no control over environment, subject to selection bias, cannot establish causation only association etc.
It’s interesting to see historians apply epidemiological concepts to their study methods. Epidemiological study designs such as (observational, retrospective) case-control study was applied to the study of history in this book. History as far as I know (at high school level) is about who did what how and when. That’s pretty much it. I heard, however, that history as it is studied at higher level is about ‘why’ too. That means historians make educated-guess based on written document, letters, pictures, relics etc. Most of the historical studies are reported in descriptive, narrative manner. This book, on the other hand, proposed a different approach. Comparative studies which utilise the same principles as case-control study. The same statistics as applied in case-control study is used too. The book consists of seven studies of history with quantitative data and statistics as supporting evidence. Each study/chapter is very academic and a bit too far out of field for me to fully understand. One thing is that the limitations of case-control study also apply here e.g. causative relationship cannot be established with this study design, only association.
This book makes me think that disciplines are just all made up. Knowledge is just that, knowledge. Concepts can be borrowed and applied across disciplines. In fact, the holy grail of evidence-based concept or clinical epidemiology, a randomised-controlled study design, is borrowed from agricultural science. As it had been utilised in the field long before epidemiologist stumbled upon it.
Diamond is quite apt to edit this book. As he was trained in physiology, he appreciates how powerful scientific method is. We are talking about experimental design e.g. control vs experimental group where by all known variables are kept in both group except only one variable of interest. So whatever measured as an outcome differences would be the result of that variable of interest. This study design is used all the times in the lab, but, of course, this design cannot be used to study what happened in the past (it’s already happened) or population (manipulation is unethical etc.). So case-control fits perfectly for the purpose as it’s not experimental per se, but more observational (which Diamond called it “natural experiments“).