Archive for category Scientific Research
Sometimes you come across scientific papers that leave you in a cloud of amazement. While reading the paper you go through a particular kind of transformation which leads you from “do they mean that serious?” to “they do mean it serious!”. Stephen Haber and Victor Menaldo wrote such a paper titled “Rainfall, Human Capital, and Democracy”.
Moderate amounts of rainfall, the authors claim, have stable democracy in their tow, however, it has to be not less than 550mm per year and not more than 1300mm. Immediately, you start to ask, what the heck should connect rainfall to democracy? Haber and Menaldo provide the answer. They have a theory and it goes like that: Moderate levels of rainfall enable cultivation of storable crops and legumes, storable crops and legumes enable economies of scale, trade in crops and legumes and accumulation of wealth. Wealth can be invested in accumulation of human capital. Intelligent people, people high in human capital that is, are able to control their government more efficiently than people low in human capital are, hence, democracy settles in. Don’t believe it? Here’s the original: “Rather, we are saying that a society made up of grain-growing family farmers is likely to have gone down a path of development that produced a high level and a broad distribution of human capital. When democratization happens in such a society, for whatever reasons, it is likely to stick because a broad swathe of citizens will have the knowledge and sophistication to enforce their rights and hold politicians accountable” (21).
Don’t object to that reasoning by pointing at Nazi Germany or ancient democracy in the dry country of the Greek, it won’t do. The theory, developed by Haber and Menaldo is one of short-range. It is designed to only apply to the world after the II. World War. Before, rainfall seems to be in no connection at all to the regime of government, hence, you can’t blame too much or too less rain for fascism or authoritarian systems. Also, skip the obvious critique that it is quite odd to place the nucleus of democratic development in farmer families given, e.g., that most if not all democratic movement started in big cities, where farmers are not that common. Anyway, let’s overlook that and concentrate on the way, Haber and Menaldo try to prove their claim, let’s look at the methodology and the measurement.
Haber and Menaldo’s model mainly draws on three variables,
- A polity-score for all countries spanning the period from 1965 to 2009 and telling you the level of democracy for which a particular country qualifies;
- Annual average precipitation within 100 miles of a country’s largest city;
- Human capital measured as newspaper circulation per capita in 1965;
Well. Results are not too surprising, democracy flourishes most with moderate rainfall and high newspaper circulation.
Since I first looked at the way, Haber and Menaldo operationalized their “theory”, I wondered about a lot of things. E.g., what would happen if precipitation is measured within a range of 10 miles of a country’s largest city? What impact do decreasing newspaper circulations have on democracy? Not to think about artificial irrigation, is it the best way to spread democracy? And, if you want to spread democracy in, say, Afghanistan, will the export of a number of family farmers to Afghanistan do the job? Questions over questions…
On page 4 of their paper, Haber and Menaldo come close to realize that maybe their research does rely on too much assumptions that are far-fetched and cannot be proven when they write: “Moderate levels of rainfall tend to generate societies with social structures that are conductive to the consolidation of democracy: they do not guarantee that every country with a moderate level of rainfall will be democratic”. However, what’s correct in one direction, must be correct in the other direction as well. Hence, absence of moderate levels of rainfall does not guarantee that every country with more or less than moderate rainfall will turn out to be autocratic, non-democratic that is. However, this renders the entire paper meaningless. Science is about making firm connections, stating that if A is given B will follow. To say that if A is given B might or might not follow, isn’t a scientific statement, it is a tautology. There is no place for tautologies in science.
A short while ago Ben Goldacre asked Why don’t journalists link to primary sources?” when they report about scientific research.
I will add to this question by asking, why don’t journalists report the real results but always seem to bend them?
Today, while brushing my teeth I learned that my life is in grave danger. Why, because often I work more than 11 hours a day. According to the chap who read the news, more than 11 hours of work a day would increase my risk of coronary heart disease. Wow, time to change your life! Start working part-time…
The news spread like malignant cancer.
ABC was more concrete: “Study Links Long Working Hours to 67 Percent Hike in Risk of Cardiovascular Problems”.
The Press Association was a bit more lavish and blamed “Long hours” for an “increase” of “heart risks” .
While The Los Angeles Times warned that “Working longer hours may make the boss happy, but it could take a toll on your heart” .
Though all journalists have the same study in mind their assessment of what the results of this particular study are, differ quite considerably.
The respective study is published in the Annals of Internal Medicine, by Mika Kivimäki and nine co-authors and it is titled: “Using Additional Information on Working Hours to Predict Coronary Heart Disease (CDH). A Cohort Study.” The adjective “additional” is the important thing here. It says, there is something else present, apart from working hours. The authors included long working hours in a model to predict prevalence of CHD. Already in the model, we find other factors that increase the risk of CHD, like “age, sex, total cholesterol levels, high-density lipoprotein cholesterol levels, systolic blood pressure, and smoking habits” (p.458).
Say these latter mentioned variables form the base model to predict a person’s risk of falling victim to CHD over the next 10 years, would this person furthermore work more than 11 hours a day, the risk he already faces would improve by a further 4,7% percent. Thus , an accurate interpretation of the study’s results means that people who are male, above a certain threshold in age, above a certain total cholesterol level, and a certain density of lipoprotein cholesterol levels and who have a high blood pressure and smoke run an 4.7% increased risk of getting CHD when working for more than 11 hours.
So, no need to start working part-time unless you are a, high blood pressured smoking male, who is high lipoprotein dense, rather old and high on cholesterol.
Maybe journalist should start reading articles rather than abstracts or, as Ben Goldacre proposed, they should link to primary source, like that:
Kivimäki, Mika et al. (2011). Using Additional Information on Working Hours to Predict Coronary Heart Disease. A Cohort Study. Annals of Internal Medicine 154(4): 457-463.