Why Empirical Research Isn’t Scientific

In this day and age, science is considered a field for empirical study. The word “science” is almost synonymous with “empirical.” Actually, it runs deeper than that: whatever is not explicitly empirical is automatically dismissed as unscientific, ideological storytelling or masturbation dressed in scientific wording. This was not always the case, but that is not the point I wish to make here. What I would like to do is discuss the implications of empirical science and why its adoption as The True Science is so dangerous.

There are problems with empirical science, and I would even go so far as to claim the implications of a distincly empirical science are a lot more problematic than the “storytelling” of scientific methods not primarily empirical. The reason for the problem of empiricism is partly due to the well-known induction problem. Just because something has happened in a certain way in the past doesn’t mean it will happen the same way in the future. The likelihood of future phenomena being like they were in the past may be greater for the natural sciences than in the social sciences; but even if the small step from past to future in the social sciences is more of a giant leap than a step, it is still fundamentally problematic even in the natural sciences to use historical empirical data to “foresee” the future.

The reason for this is that there is seldom the case that one simple rule is applicable on a phenomenon without influence from other, probably unknown, parameters. It is almost never the case that a phenomenon tomorrow will take place under the exact same circumstances as an event tomorrow. We may think this is the case, but concluding that we know this is the case – even in a controled laboratory – is a bit premature (or, rather, ignorant).

Some “laws” in the natural sciences seem to be applicable on certain phenomenon and have been “verified” in a number of cases already. But what this means is not that the law is necessarily correct, only that the law comes close enough to explaining what is really going on. There is no way we can tell with 100% certainty that we have found the true explanation, even if it were the case that the theory has been verified a billion times. Or even if it hasn’t been falsified.

But this is not to say that we cannot know anything, which seems to be a popular belief nowadays in fields adopting e.g. postmodernist “theory.”

My concern is not the natural sciences, even though empiricism in such “law-based” sciences is problematic. My main concern is the social sciences, in which empirical research as the hegemonic view of science is literally killing the accumulation and creation of new knowledge. Economics is no doubt the field that has come furthest of the social sciences, and  it is therefore in worse shape than the other disciplines.

The reason empirical research worries me is that it tends to be without aim. A scientist conducting empirical research is a slave to the data collected; he can not come to any conclusion that does not fit the data and he also cannot verify that the conclusion drawn based on the data is correct.

Imagine a physicist conducting an every-day experiment when teaching a class the rules of gravity. He drops his pen and expects it to fall to the floor. But imagine the pen instead floats in the air for a while before it rushes to the floor. This would seem to falsify the law of gravity, and the whole class could see it. But the explanation may be as simple as the class on the next floor is doing an experiement with powerful magnets. The problem with blind empirical research is that there is no way of saying that the data is wrong or that there is another explanation.

The example of the pen may seem ridiculous, but it is important and can be applied on basically any experiment or study. What we do not anticipate and control for is hardly ever noticed, and so the data we have collected – even if the data is itself correct – may lead us astray. Induction, which is the same as purely empirical research, leaves us clueless of whether there was something wrong in data collection or data handling. What we don’t know can totally change the outcome of empirical research; actually, empirical research presupposes that we do not know and that we try to control for our “biases” as far possible. What the scientist believes, thinks, etc is supposed to not affect the data.

Then apply this on a social science such as economics, where that which is studied consists of people. Studying a simple market transaction, say the purchase of gum, and collecting all the data we possibly can could tell us something. For instance, we might find out that in a certain neighborhood gum is sold to 10% of customers at a price of $1.25 whereas in an adjacent neighborhood the same kind of gum in the same kind of convenient store is sold to 20% of customers at a price of $1.65. Adding another store, a little further down the road, we find that 25% of customers buy that same type of gum for $1.99.

With the customers being basically the same, and the stores are the same as is the gum, we must conclude that the higher the price the more people buy gum. This doesn’t make sense at all, so most of us would discard the findings and claim that there is something wrong with the data. Perhaps the stores are all along a route to a small airport where people usually buy gum to chew during take-off to avoid discomfort due to rapidly changing cabin pressure, and that could explain the strange findings?

If economics were empirical we couldn’t tell that something ought to be wrong. Imagine a number of empirical studies such as this one, and we suddenly have a verified false theory. Or, even worse, we have a number of conflicting results and have to way of knowing what is the right explanation. Economics, just as any social (and natural) science, is dependent on knowing what to look for, how to look for it, and what to expect from the data. Without such theories of what should be the case we are blind and do not know where we are heading or whether we are going in the right direction.

This is partly what is happening in the social sciences today, where the different disciplines increasingly overlap and study the same things. But they come up with different results even though the studied phenomenon is the same. The reason conclusions in e.g. sociology, economics, and political science can be different even if the same phenomenon is studied depends on the perspectives and theories prevailing in the respective fields. This fact would be used by an empiricist as an argument for the pure empirical research – to let the data tell the story without human bias. But what this shows is exactly the opposite.

The reason they reach different conclusions is that they are guided by perspectives and theories that are all wrong – but they are wrong in different ways. Only through studying the same phenomenon from different perspectives trying to explain what is going on, and through inter-disciplinary research, can the truth be found. Empirical research can never tell us what is going on, but only what we were able to record at a certain moment or time period.

The empirical researcher might say that empirical research is not totally without theory, which means that it is guided by theory – but that theory can be tested empirically and that theories that don’t fit the data should be discarded or at least modified. But this presumes that we can control the setting fully and that we know that what we find is everything and correct – it presumes the researcher is God.

The researcher is not God, and that is the very reason research must make errors and mistakes. Because we are not omniscient we must be allowed to make mistakes yet learn from them. This goes as much for a five-year-old growing up as it goes for the scientific community. We know that whatever facts we have established empirically can be wrong; we know that we might have overlooked something. The results of science must make sense, and science makes sense only if we begin with theory. The pen floating in the air for a minute doesn’t make us wonder what happened to gravity, but what – at that very moment – we’re not seeing because a pen floating in the air doesn’t make sense.

It doesn’t make sense because we know that things tend to fall to the floor rather than float in the air, and we know all other things in the room are indeed “behaving” the way we expect them to. So the pen doesn’t falsify our understanding of gravity – it is the result of some phenomenon/parameter we are not seeing.

The same is true, and even more so, for social sciences, where we study human behavior. In the case of the convenient stores we know, a priori, that no human being would buy gum unless that which he gives up to get the gum is worth less than the value he places in the gum. So the fact that gum is sold at greater volumes in one store at a greater price than in other stores means there must be a reason why customers in the former value gum more. Maybe they have more money? Maybe they expect greater utility from the gum for some reason? Our not knowing means there is something we need to find out. If we were looking at the data without a theory to interpret and assess what we find, we would have to discard what we know is right. We know that people cannot give up something that is of value to them for something that is of less value to them – unless there is something making them act irrationally.

The “storytelling” of non-empirical sciences may indeed be incorrect, but so can any conclusion – because the researcher is not God (and if the researcher was God, of what use would doing research be?). But non-empirical research can still tell us about the true state of the world without necessarily being verified by data. Non-empirical knowledge is the starting point, and empirical research may show us that we are unknowingly omitting something important – if we don’t see what we expect to see there obviously must be something missing: either our expectations are based on a false understanding of the world, or there was something making the data unreliable (collection or registration errors, or some other phenomenon taking place at the same time that we didn’t foresee and “control for”).

The truth is that empirical research cannot take us forward – it can only provide data of the past. Empirical research cannot even tell us about the past without interpretation which necessarily makes use of the imagination of the researcher – that which he already knows and understands. The researcher’s understanding is the key to making new knowledge, whereas the data can only strengthen the researcher’s case through making the explanation convincing to others.

There is a reason most contemporary disciplines have been further fragmented into sub disciplines, e.g. economics into micro economics, macro economics, etc. There is also reason why these sub disciplines generally don’t mix. The reason is not that they are on different levels or study different phenomena, but that economics as a primarily empirical research has lost its understanding of the human being. Through endless simplifications and idealizations (in the Weberian sense) micro economics have gone its own way in the study of certain phenomena, guided by data alone; macro economics has taken the same methodological path, but has ended up somewhere else. They are no longer compatible, which causes friction and frustration and therefore makes it necessary to keep them apart.

But are micro and macro economics different “sciences”? No, they both study economic phenomena resulting from human [inter]action and should therefore be fully compatible; if they were true, they would offer the same kind of explanations to the studied phenomena and be able to provide a greater understanding for how human beings interact and what to expect from future interactions. Instead, we should expect to see further fragmentation as research conducted in micro economics will show “great divides” between economists researching somewhat different phenomena or using different techniques.

With an empirically based research agenda scientists will soon find that their findings don’t go well together with other scientists’. Since “data don’t lie” researchers of the same type will create sub groups with different “proven” theories in each group explaining the same kind of phenomena. This fragmentation will, theoretically, go on until each individual researcher has his own discipline in which his own understanding, based on the data, is “law.”

Economics doesn’t need more unguided empirical research (I’m exaggerating somewhat here), but a general understanding for what it studies and what to expect from future empirical research. Such understanding cannot be provided by data, but must be the result of introspection and reasoning based on experience and supported by observation. Data is a means to strengthen one’s theory or understanding and a way of showing others interested in the same kind of phenomena the benefits of your explanation.

Data and statistics can presumably show what really happened, but only to the extent that we had already tried to explain the phenomenon – our understanding of the world is crucial in the selection, collection and interpretation of data. Statistical analysis can at best tell us that we’re missing something, but it cannot tell us what is missing, what we don’t understand, or what to expect from this kind of phenomenon in the future. It is but a tool we can use to make sure we haven’t overlooked something.