There is a significant problem that can affect government decision making that I have not heard much about before.

This might be because Conservatives are not looking for it, and the Left are seriously taciturn about anything that critiques decision making processes within political, rather than commercial organisations? I don't know exactly. Maybe I haven't looked hard enough.

I would argue that decisions made for large numbers of people cannot be rationally based without recourse to statistical methods; from here the problems are two-fold.

Firstly, statistical methods work on the basis of best fit. Which is immensely useful in the design process, ergonomics and so forth, for its ability to turn an incomprehensible mass of data into a single number that can be used in a design.  My main objection to it is that the number of outliers can be large or small, and the decision about how many outliers is acceptable comes down to assumptions which may or may not be stated.

In short, a highly anatomical chair design will injure people whom it fits poorly. How badly they are affected is largely a function of how much clearance, or slack, there is in the design. This is essentially the extent to which the ergonomic design is modified to accommodate those who differ from the mean case that was the original basis for the design. The most egalitarian design is a straight backed chair, which could be said to fit everyone equally badly.

There are definitely times when a less anatomical design may cause less suffering in total, even if the straight-backed chair is less comfortable for someone within half a standard deviation of the mean, depending only on the lack of diversity of the sample. In short, the more diverse the population, the more un-tailored must be the solution to minimise the dis-ergonomic effect on those not close to the designated mean.

This extends to a variety of policy decisions, some of which fit a great many people very badly indeed.

Which brings us to the second problem: most people are intuitively very poor at understanding base-rate statistics even when technically trained, and appallingly bad when untrained. My experience with government in Australia suggests that in the public sector, Arts majors predominate, whereas in business, quantitive skills are far more prevalent.

This leads to policy being informed by officially sanctioned narratives rather than empirical data, and often, by heavily dialectic reasoning, bordering on reasoning by analogy.

There is no way to to quantify fit with a qualitative narrative. If the policy starts that way, it may be subject to numerical analysis only when it is being budgeted for; at this point, it is all about costing, the quantity most measured is monetary.

This excludes a lot of very important variables from consideration, especially those variables significant to the design.