Social Statistics & Moral Inferences

Last week, I wrote about the problem of understanding what the social sciences provide. My intention was to theorize the thinking involved in social science research and reflect on what this means for its outputs. While somewhat critical, my overall position is not that the social sciences are vapid but merely misunderstood – people misinterpret what such forms of study reveal and consequently misuse and misinterpret their implications. Now, however, I wish to be somewhat more polemical. I think that the manner in which social sciences are invoked, especially at the level of individual decision making, is rather problematic and needs to be questioned. People invoke statistics and act as if they are explanatory instructions about what any given individual ought to do – and I simply think this is erroneous and even destructive. Though I maintain a critical tone in this, I should not be understood as dismissing the social science as such; rather, my ire should be understood as directed against a specific manner in which the products of the social sciences are used that has become fashionable. Social sciences may provide intriguing or even valuable information, but their findings are often misunderstood as deterministic prescriptions which leads to flawed moral decision-making.

Inevitably, I must repeat in some manner what I have already said. The most simple is that the ‘facts’ of the social sciences are frequently misunderstood to be akin to the ‘laws’ of physics. Most often, the social sciences merely reflect the self-understandings of a given people at a given moment, and such self-understandings are certainly mere contingencies that can be changed with time. Insofar as the findings of the social sciences may extend further than this, they arguably become more biological or neurological as they plumb depths more profound than mere social realities; an example of this could be tendencies in sexual differentiation, wherein men and women have cross-cultural proclivities that seem to be unaffected by local customs – though this requires literally global data that is incredibly difficult to ascertain. The second is that the social sciences’s findings tend to be too unstable to have any prescriptive force. As I explained in my prior post, scientific formulations require profoundly strictly parameters in order for their fundamentally descriptive statements to be considered capable of prediction. The predictions of astrophysics are reasonably predictive because the relevant variables are fairly easy to know and expect, whereas meteorology does not have such an easy go; this consideration is only made more difficult to manage with the social sciences.

Now, the practical implications of such realities may not be so obvious, but I think an example will be illustrative. I have heard some folks argue that we should not expect lifelong, monogamous marriage because of the present rates of divorce; in the US, for example, I have heard some people conjecture that it is as high as 70% but my research has led me to believe that it is closer to 50%. In either case, whether we think that it is merely half or a bit more, the invocation of this statistic is used to suggest that we should not expect our own marriages to last. People are fickle, as (allegedly) shown by this statistic, so this is an iron law to which we should not expect ourselves impervious. Just as we cannot defy the reality of gravity, so too must we accept the harsh reality that permanent marriage is just a foolish institution invented by archaic ignoramuses of prior civilizations; we need to update these institutions to match our present understandings of the real way in which people come to love each other (or not).

This, I hope you agree, is obviously stupid. But if you so happen to think this is a legitimate manner of justifying actions or do not understand why this is stupid, then we need some reasons. For now, I will relegate myself to just three: 1) statistical generalizations are not specific instances; 2) social patterns are not agentic reasons; and 3) the existence of different generalities based on alternative beliefs. In each case, I will continue to harp on this argument gleaned from marital statistics, but I hope that my point will be understood as applying to a broader type of claim. Anyone who takes a simple, sociological ‘fact’ (or that of any other social science) and tells you that they can derive a moral imperative from it is simply incorrect; such information might be able to inform, but its instructive capacity is non-existent.

The first issue is that statistics are comprised of a variety of specific instances, and the statistics always glean a mere generalization from the data. For example, a set of ten people might be asked their annual income: five say ten-thousand dollars while the other five say one-hundred-twenty -thousand dollars; the average income is, then, sixty-five-thousand dollars, but this represents no specific person (indeed, it is quite far from any person given person in the set). Alternatively, to use the marriage example once more, even if we presume that the divorce rate is, in fact, 70%, this means that three tenths of population are not getting divorces. From this information, there is no way to determine which side of that divide one’s own marital situation will fall on. This may have to do with a confusion between probabilities and likelihoods. These two notions get confused in common parlance, but – understood in a stricter sense – they quickly become distinguishable from one another. A probability, properly understood, is a statistical notion about how frequently something will occur in an essentially mechanical or causal sense; an easy example is that of flipping a coin – we should expect that one thousand flips of a coin will yield approximately five hundred occurrences of each side being face up. Likelihood, however, has more to do interpersonal interaction and the  proclivities that people tend to exhibit. It is only on the side of likelihood that someone might predict a divorce; perhaps a a man is well known to have had a history of commitment issues and being flaky – we might therefore doubt the longevity of his marriage. This is dictated, therefore, not by the statistics but by the self-understanding of the agent in question – and this is the crucial next claim.

Insofar as probabilities are hardly the same as the specific instances from which the generalizations are derived, so too are the ‘facts’ of social sciences profoundly distinct from the specific reasons agents have for making the choices that underlie the statistical outcomes. To take a somewhat jocular example, I remember during my childhood that, whenever I did something questionable due to peer pressure, my parents would quip: “Would you jump off a bridge if all your friends were, too?” My smart-aleck response would be: “It depends on why.” This flippancy was not, however, without its wisdom. Involved in my parents’ statement is a presumption that the simple mass of people doing something is what causes me to do it, but this is simply wrong. Instead, a group of twelve year old boys might be doing something stupid or dangerous to show off their strength and bravery to one another or even to the pretty girls in their class; the reasons for the action must be taken into account, not just the mass of people doing that thing. (Of course, this can be a factor in people’s decision making: if someone thinks everyone is doing a specific thing, they may presume it is the right thing to do – but this itself is based on an assessment of reasoning as opposed to probability.) What becomes important in the marriage example is what the respondents believe about themselves and their circumstances that results in them either remaining in or abandoning their marriages – and this is utterly distinct from the statistical question. Someone who is concerned about the strength of his own potential marriage should look to these lines of reasoning as a comparison for his own self-assessment and not the statistics that are abstracted from those richer self-understandings.

This reality of diverse self-understandings brings us to the last problem with statistical reasoning: the fact that different generalizations can be brought about by isolating different variables representing different general beliefs. Though I mentioned a more recent number of divorce sitting around 50% above, a 2015 Pew Survey had the number around 30%; when they controlled for Catholics, however, that number dropped to 25%. Now, though this is not definitively the case, it could be that a Catholic understanding of marriage as being an indissoluble bond with your spouse and God has an effect on the members of that union – divorce simply does not emerge as an option in the same way. Similar statistics are, from what I understand, common among different religions that all place a premium on lifelong marriage. Of course, further controls could be provided that might affect the variables: education, upbringing, economic status, etc., which all have an effect on one’s self-understanding. Though I am still skeptical of any of these markers being determinative or even informative, the point I wish to derive is that no study will provide you with an assessment of your specific self-understanding. It simply is not possible. The studies themselves suggest that different beliefs cause different outcomes, but there can be no way for a study to ever capture and make generalizable the totality of your individual views – the statistics cannot be relied on for this sort of agentic deliberation. Something else is needed, and the statistics are likely just distractions.

In brief, when discerning a personal decision of great importance, statistical generalizations are the wrong place to look for guidance. This is not a flaw of the social sciences themselves but of the way they are often misapplied—a problem whose origins we might debate another time. Certainly, much of ‘pop psychology’ and ‘click-bait studies’ appear foolish, but I worry we dismiss them for the wrong reasons. The usual critiques focus on their superficiality or flawed methods rather than recognizing the deeper issue: the essence of social science studies cannot provide moral imperatives. Whether you are debating which degree will secure the best career, whether to marry, or which diet promotes the longest life, you must turn to reasoning grounded in your own particular circumstances—not the flimsy trends of generalized populations.