Traditional statistics is strong in devising ways of describing data and inferring distributional parameters from sample. Causal inference requires two additional ingredients: a science-friendly language for articulating causal knowledge, and a mathematical machinery for processing that knowledge, combining it with data and drawing new causal conclusions about a phenomenon.

Today we preach that science is not science unless it is quantitative. We substitute correlations for causal studies, and physical equations for organic reasoning. Measurements and equations are supposed to sharpen thinking, but, in my observation, they more often tend to make the thinking noncausal and fuzzy. They tend to become the object of scientific manipulation instead of auxiliary tests of crucial inferences.
Many - perhaps most - of the great issues of science are qualitative, not quantitative, even in physics and chemistry. Equations and measurements are useful when and only when they are related to proof; but proof or disproof comes first and is in fact strongest when it is absolutely convincing without any quantitative measurement.
Or to say it another way, you can catch phenomena in a logical box or in a mathematical box. The logical box is coarse but strong. The mathematical box is fine-grained but flimsy. The mathematical box is a beautiful way of wrapping up a problem, but it will not hold the phenomena unless they have been caught in a logical box to begin with.

One can ask two different kinds of questions with regard to the topics of study in psychology as well as in other sciences. One can ask for the phenomenal characteristics of psychological units or events, for example, how many kinds of feelings can be qualitatively differentiated from one another or which characteristics describe an experience of a voluntary act. Aside from this are the questions asking for the why, for the cause and the effect, for the conditional-genetic interrelations. For example, one can ask: Under which conditions has been a decision made and which are the specific psychological effects which follow this decision? The depiction of phenomenal characteristics is usually characterized as “description”, the depiction of causal relationships as “explanation.”

In theoretical science, the question is—What are we to think? and when a doubtful point arises, for the solution of which either experimental data are wanting, or mathematical methods are not sufficiently advanced, it is the duty of philosophic minds not to dispute about the probability of conflicting suppositions, but to labour for the advancement of experimental inquiry and of mathematics, and await patiently the time when they shall be adequate to solve the question.

Isn’t the social scientist’s use of the null hypothesis simply the application of Popperian (or Bayesian) thinking in contexts in which probability plays such a big role? ... since an output variable such as adult IQ, or academic achievement, or effectiveness at communication, or whatever, will always, in the social sciences, be a function of a sizable but finite number of factors... Putting it crudely, if you have enough cases and your measures are not to-tally unreliable, the null hypothesis will always be falsified, regardless of the truth of the substantive theory.

The postulation of a principle of causality, “to every effect there is a cause,” has been a continuing central problem for philosophy (Popper, 1972). Its role as a source of contention in modern science (Jauch, 1973) is epitomized by Einstein’s remark that, “I can’t believe that God plays dice.” Many of the arguments about the application of the principle are very relevant to systems science and to problems of system identification and machine learning, on the one hand,and to epistemology and behavioural psychology, on the other. In current system science the theory of causal deterministic systems is most well developed and generally applied, while the theory of modeling with alternative structures, e.g., stochastic automata, indeterminate automata, products of asynchronous automata, etc., has not been developed to the same degree.

Many psychologists, sociologists and especially anthropologists and psychiatrists raise serious objections against routine attempts to "extend the methods of the physical sciences" to the study of man. These objections cannot be dismissed simply on the grounds that they are not constructive; for inherent in the objections may well be a conviction that there can never be a "behavioral science" as scientists understand science. Whether there can be such a science or not will be decided neither by citing successful applications of "scientific method" to carefully circumscribed sectors of human behavior nor by pointing out what has not yet been done. Therefore on the question of whether a can in principle be constructed, we shall take no sides. That some kinds of human behavior can be described and even predicted in terms of objectively verifiable and quantifiable data seems to us to have been established.

Ideology, legitimacy, authority: Such topics used to interest academics no end. The mania for quantifying the social sciences and the attendant decline in foreign-language acquisition have changed things. I remember Sovietologists lamenting this trend when I was at school in the 1980s. It’s gone much, much further since. What cannot be researched with the proper statistical-numerical methodology is thought beneath serious analysis. On the rare occasions when a non-quantifier takes the microphone at a conference, the Gradgrinds lean back with an indulgent smile: Time for some light relief.

The problems of society are our problems. And we go through research to proffer solutions to the problems of humanity. In that sense, what we say is that, if we carry out these research, we must ensure that the research or the outcomes of this research are impacting the people for whom the research are carried out.

Research in statistical theory and techniques is necessarily mathematical, scholarly, and abstract in character, requiring some degree of leisure, detachment, and access to a good mathematical and historical library. The importance of continuing such research is very great, though it is not always obvious to those whose interest is entirely in practical applications of already existing theory. Excepting in the presence of active research in pure science, the application of the science tend to drop into a deadly rut of unthinking routine, incapable of progress beyond a limited range predetermined by the accomplishments of pure science, and are in constant danger of falling into the hands of people who do not really understand the tools that they are working with and who are out of touch with those that do. ... It is in fact rather absurd, though quite in line with the precedents of earlier centuries, that scientific men of the highest talents can live only by doing work that could be done by others of lesser special abilities, while the real worth of their most important work receives no official recognition.

As long as research uses sample data to make inferences (which can hardly be avoided), the problem of interpretation cannot be avoided, regardless of the particular statistical method or even methodology we use. In everything we do, we are forced to make decisions based on assumptions formed in part from our experiences with the behavior of others, which in turn reflect only part of the whole picture. These assumptions reflect conscious or unconscious values (such as giving people the benefit of the doubt, presuming the innocence of an accused person, etc.). This reality does not entail that we cannot strive for scientific detachment once these fundamental pre conditions have been set and the research proceeds. Within the particular “research design,” we may still strive for and insist upon scientific detachment, even though our values have necessarily influenced the design itself.

In the last decade's intensive study of all sorts of social and economic time series, it has become clear, it seems to me, that the usual time series technique is not quite adequate for the purpose which the social investigator is pursuing... We want to find out on more or less empirical grounds what is actually present in the series at hand, that is to say, what sort of components the series contains.

The solution of the most important problems of the theoretical social sciences in general and of theoretical economics in particular is thus, closely connected with the question of theoretically understanding the origin and change of 'organically' created social structures.

The most important questions of life... are indeed for the most part only problems of probability. Strictly speaking it may even be said that nearly all our knowledge is problematical; and in the small number of things which we are able to know with certainty, even in the mathematical sciences themselves, the principal means for ascertaining truth—induction and analogy—are based on probabilities.