What does the empirical theory say?

Basic concepts of empiricism

Whatever positivism may advance against metaphysics is pointless.
Whatever metaphysics may propose in self-defense against positivism is wrong
Wolfgang Stegmüller (1954)

The scientist's aim in conventional research is to uncover cause-effect relationships that contribute to an understanding of the phenomena he is observing. Science is therefore every intersubjectively verifiable investigation of facts and the systematic description and explanation of the facts examined based on it. A science traditionally consists of

  • theory: Descriptions, models, explanations,
  • Empiricism: Facts, observations and
  • communication: intersubjective review

We can distinguish between two "types" of scientists or two approaches:

The empiricist argues that one only has to collect enough individual observations to be able to recognize the complex structure of relationships between this isolated information. The empiricist rejects the anticipation of these relationships and is thus in opposition to the

theoristwho starts from a reference system and allows himself to be guided in his research by the implications derived from it (argument of economics!)


The discrepancy between the two attitudes expresses not only a different assessment of the theory, but a fundamental difference in the question of the methodical treatment of observational data:

The empiricist proceeds inductivei.e., he moves from the facts to the formulation of legal statements that describe the facts.

The theorist proceeds deductivethat is, it assumes a general principle or "vague premonition" and looks for evidence to support the principle.

In scientific practice, i.e. in most cases, theories are the result of the interaction of both methods. Scientific theories determine the framework in which empirical investigations are then possible. Without an empirical examination, the theoretical framework remains without content, but an investigation without a framework remains equally incomprehensible.

Popper (1979) differentiates between two forms of carrying out empirical research: by means of the Bucket or headlight theory. In which way scientists collect data, whether by observation, experiment or questioning, in one way they encounter problems of a general nature that is independent of the instrument. According to the headlight theory, the information-containing hypotheses are initially formed and only afterwards are attempted to test them for their truth content through empirical research or experiments, i.e. reality is examined in a headlight-like manner. Popper's criticism is mainly directed against the bucket theory, in which empirical observations are collected as if in a bucket, i.e. it is merely about an accumulation and collection of facts in which hypotheses are only formed after the observation and depends on the observation results. With his headlight theory of science, Karl Popper pointed out the fact that not only laypeople are prone to prejudice, but also that every scientific description of facts is selective: what the headlight makes visible depends on its location and the way it is used how we adjust it, its intensity, color and, of course, how we decide what should be illuminated by it.

The Function of theory is to combine the available factual evidence into a somewhat structured whole. To do this, it needs (hypothetical) constructs, (explicative and descriptive) hypotheses, (independent, dependent and intervening) variables and laws.

Hypothetical constructs

Processes as they take place in the organism, in the atom or in the galaxy are not accessible to direct observation. What we can observe are the effects of these processes. Knowledge of the effects and the impulses that trigger them, as well as the reference system of relevant laws and findings, enable conclusions to be drawn about what is not visible. These constructs are done with the rest of the theory Hypotheses connected. A hypothesis is a conclusion - derived either from theoretical or factual statements - that concerns the relationships between observable and non-observable quantities, processes or interactions. One hypothesis is either explicit (explanatory)i.e., it makes predictions of the kind that a specific phenomenon will occur together with other phenomena as a concomitant of a Causal relationship exists between them or it is descriptive (descriptive), i.e. it says something about phenomena that are on the same level of abstraction as a specific phenomenon.

Popper (1979): "Observations are secondary to hypotheses".

Hypothetical constructs have a narrower scope than hypotheses. Their use in theory and research gradually helps the construct to a more precise definition, its connotations and its dependence on other factors become clearer. These factors are also known as


Variables are variable quantities that can be used to describe or explain phenomena. One further differentiates:

independent variables: causative factors that are manipulated in an experiment to have their effect on dependent variables to observe, that is, the researcher determines the changes in the effect in order to arrive at conclusions about the process he has postulated. If the elements of the relationship are determined with sufficient accuracy, he can make specific, exact predictions about events and behaviors. He's working now intervening variables: these are processes that can be determined relatively precisely, although they cannot be observed themselves. These are the processes which one must assume in order to explain a connection between the changes in the observed dependent variable and the changes in the relevant observable independent variable. These intervening variables are a form of

scientific laws: Laws are terms for invariant relationships of any kind. They can refer to the connection between two events, or they are general enough to extend to groups or classes of events. (They have a narrow or wide scope). Laws are more stretched than constructs or intervening variables, because these are each limited to processes of a certain category. The usefulness of scientific theories depends on the clarity and accuracy of their language.


The language plays an important role as soon as a theory prepares itself term to outline or define. All definitions are based on elementary terms that do not require any definition. These, in turn, can be abstract or concrete. The empiricist defines abstract terms using concrete terms. The theorist makes the abstract his elementary concepts and defines the concrete concepts by means of the abstract classes of which they are elements.

The practice of defining abstract terms by resorting to more concrete processes has much to do with the process of operational definition together. The Operationalism is an aid to the definition and limitation of the scientifically permissible terms and problems. He does not guarantee the scientist that his results will be free of errors or that his theory will be valid. The operationalization gets its meaning in the formulation of terms and the enabling of communication and comparability between scientific systems. It is particularly problematic when there is a glut of such definitions for a single term, which is not uncommon in psychology, educational science and other social and / or human sciences. While the descriptive function of the theory discussed so far raises problems of definition and language, its explicative function leads to problems of a logical nature. The first difficulty lies in the temptation to think teleologically: roughly speaking, it means that one ascribes the role of the cause for some current behavior to a future event, that is, the cause of the effect follows. While this logical "footfall" is easy to see through in most cases, anthropomorphic or animistic interpretations are widespread. Here, analogies are misused to depict relationships as if they can actually be represented or paraphrased in this way. An analogy must therefore always be identified as such, for without it it is of course impossible to construct a theory, and not even to create a model of reality.

Under a model one understands a carefully constructed system of proposals, constructs, analogies, hyotheses, which at the same time offer an ordering scheme for them. Just like constructs, models have the character of suggestion and assumption. The creator of a model uses it in all his research and modifies it until the original metaphor has become a system of legal statements. The benefit of such models is that we can explain human behavior as if mechanical principles played along. What remains essential, however, is that it is just one "as if" is and not about the reality himself.

The Conventions of Theory Construction should help the scientist to a reference system of laws, which he can use for the discovery and formulation of his information. We make a difference Fundamental lawsthat state truths in invariant and universal forms. They apply to all things within a certain category, they are unspecific in this sense. We do not find such in the social and / or human sciences, rather the statements are statistical lawsthat state what is probable under which circumstances, but what does not necessarily have to be. Its object is a matter of mathematical probability, but not of absolute certainty. All laws must Truths testify, regardless of whether they are found by observation or by inference.

A final but essential point is that practical requirementsthat a theory must fulfill:

Scope of the theory: A theory must encompass a broad field of appearance and be able to process the relevant (present and future) data without internal contradiction. Economics of Theory: that statement is the most expedient, which requires the fewest assumed assumptions in order to be transformed into exact predictions. It cannot be more extensive than the phenomena it tries to explain. Verifiability of the theory: it must give other scientists the opportunity to understand or check their findings. Predictability of the theory: it has to be original, so it has to be able to tell us something that we don't already know without it; it must not stop at a description. Flexibility of the theory: it never comes to an end and must be able to process new evidence. It must be able to absorb new information, but must not at the same time be used for all possible and impossible data.

The intersubjective verifiability (Communication) means that the empirical facts must stand up to an intersubjective examination as well as the conclusions within the framework of a theory. "Intersubjective" has multiple meanings:

  • in principle observable for all people,
  • in principle repeatable and all
  • Conclusions must be understandable for others, i.e. obey certain logical rules of argument.

example: There are numerous reports about UFOs. If we classify the witnesses as credible, it is a matter of "observations" and to that extent also of "facts". There is also a theory: UFOs are extraterrestrial machines for intergalactic traffic. For many UFO supporters, this theory is confirmed by their individual observation. So far, however, it has not been possible to confirm the observations intersubjectively; the observations are one-sided (only visible phenomena, there are no verifiable effects: extraterrestrial objects, etc.). The associated theory is by no means the only possible one. There are also psychological theories about UFOs (perceptual disorders, hallucinations, projections of the collective unconscious, etc.). The same applies to many phenomena in psi research.

Objections to Popper's Critical Rationalism

Thies (2004) points out in the form of three theses that all the objections that can be formulated against Popper have already been considered by himself, whereby he has further developed his conception himself. "Popper's critical rationalism is primarily directed against empirical concepts, above all against the logical empiricism of the Vienna Circle (Carnap, Neurath, Schlick et al.). Carnap was then forced to make some revisions, such as clarifying his ideas of inductive logic. But Popper's criticism of empiricism remains right: there is no such thing as an "innocent eye", every observation is loaded with theory, and every term contains theoretical assumptions. Popper later coined the following picture for this: Human knowledge is not like a bucket, but a spotlight (in "Objective Knowledge", Hamburg 1998, 61ff., 354ff.). Therefore, theories are not derived from experience (this happens at best in the " context of discovery "). However, experience remains the supervisory authority that can bring about falsifications; the critical examination of theories (in the "context of justification") is empirical. In general, the objection to Popper is that his image of the sciences is too "idealistic". In particular, falsifications must not be imagined too easily.

  • One can learn from Duhem that several theories are always connected in a network-like manner (holism). Popper also speaks of an “empirical-scientific system.” If one observation turns out to be negative, the question still remains as to which part of the entire ensemble should be abandoned.
  • A single black swan does not refute the whole "swans-are-white" theory. There are always explanations possible as to how this negative observation can come about: a unique genetic mutation, a peculiar disease, etc. Not every unsuccessful attempt in the Physics teaching is bringing down modern science, and obviously a decision has yet to be made whether to accept or reject a falsifying basic set.
  • It plays a role here whether another theory is already available that can better explain emerging "anomalies" and all previous observations. One must take greater account of the competition between different theories.

Popper admits that he makes conditions that he cannot justify. Critical rationalism is based, he consequently concludes, on a "moral decision" ("The open society and its enemies", Vol. 2, Chapter 24, Tübingen 71992, pp. 271 and 281). So if Popper claims that science should search for truth (truth as a regulative idea), but that the truth of a statement can never be proven, Popper is unable to justify why science is after
Should seek truth (self-application argument). According to his own assumptions, Popper's idea of ​​science can only be empirically proven better than another - but, as Kuhn and Feyerabend showed, that is by no means the case. "Thies sees the only way out as a transcendental expansion of Critical Rationalism in this sense : "Truth is not just the regulative idea of ​​scientific action, but also an inescapable requirement. In addition, further a priori assumptions go into the sciences, the validity of which can be justified. "

Expertise vs statistics

The empirical research on prognostic validity shows that a forecast based solely on statistical models is on average better than the expert judgment. I'll do it very simply. Statisticians have found that characteristics A, B, and C are strongly related to behavior X. Now let's take a hundred experts who are supposed to predict whether a hundred test persons will show behavior X. Independently of this, we measure characteristics A, B and C and predict behavior X based on a mathematical model. The probability that the prediction based on the mathematical model is correct is significantly higher than the probability that the experts are correct. This is shown by fifty years of empirical research on the validity of expert judgments.It has been shown, for example, that the professional success of trainees is better predicted even with a nasty intelligence test alone than with the judgment of experienced HR managers.

For example, it has been shown that combining personality tests with a few sociodemographic characteristics is better at predicting the likelihood of recidivism of offenders than the judgment of experienced forensic psychiatrists. The list could be continued indefinitely, it is always like that. The prognostic validity is always increased if psychological instruments such as tests, questionnaires, assessment centers are used and the values ​​are related to socio-demographic variables using suitable mathematical models. So it is by no means the case (...) that the statisticians provide the rough orientation and the experts then have to do the fine-tuning. As a result of this "fine-tuning", they often even worsen the forecast quality. It is a fact.

People usually overestimate their ability to make judgments. Psychologists and other psycho-experts should actually know that because the relevant studies are not kept secret. They can be borrowed from any university library. Now I have the assumption that this self-respecting disregard for reality could be responsible for economic as well as psychological reasons. In psychopathology, distortions of reality for the purpose of self-esteem regulation are called narcissism. Another characteristic tendency of narcissists is the devaluation of people who question their own pathologically inflated self-esteem. What this can mean in practice was astonishing to see a few years ago when the science journalist Rolf Degen published his book on "Psychological Mistakes". In that book, he claimed that psychotherapy was basically placebo. He was not claiming anything new for the specialist. Robyn Dawes, for example, had previously published something similar in his book House of Cards. A method-critical evaluation of empirical psychotherapy research to date does not allow any other conclusion. Dawes and Degen are no nuts, they only use the methods that they have learned as psychologists and which are generally considered binding in specialist circles. Degen countered an anger from psychotherapist circles that can hardly be described in words. Dawes also had to take a lot, but the Americans are obviously not quite as irrational as the Germans. In my opinion, these drooling attacks cannot be explained by economic interests alone. Psychological reasons are also likely to play a role here. Schmidbauer's theory of the helper syndrome as a form of narcissistic personality disorder seems to me to be an entirely suitable explanation. (...)

Wolfram Heinrich replied to this statement:

The MPU expert or a forensic or other psychological expert should not predict the behavior of a hundred subjects, but of one. The statistical findings about the group of people to which the test subject belongs are important information, they
but are - logically - not enough. I have to bring together my knowledge of the whole and my knowledge of this one special person. The statistics have to leave me in the lurch because it * can * no longer be accessible to them.

It's (...) about a special person. And to take your example from above: My job as an appraiser is to find out to what extent the characteristics A, B and C in Mr.
N. are pronounced in order to then predict his future behavior X. (...)

source: Heinrich Drendorf and Wolfram Heinrich in the newsgroups "de.sci.psychologie" (Sun, 6 Aug 2006 11:32:12)

Highlighting and cuts (…) from me; W.S.

How can one deduce the individual case on the basis of empirical data?

Of course, one cannot deduce the individual case from a random sample, which "nomologists" do not claim. However, it has generally been shown that predictions based on mathematical models (e.g. regression equations) are more valid than "clinical", "naive" or heuristic predictions for psychological decisions. In other words, conclusions are not drawn from a random sample on the individual case, but from a random sample on which the validity and reliability of measurement methods were tested, on other random samples. This also applies to all psychological tests. They do not prove themselves for the individual case, but for the collective. If, for example, suitability for certain jobs is diagnosed with tests or assessment centers, then with a larger number of applicants and positions only the company systematically benefits, whereas unfortunately some applicants who would be perfectly suitable are rejected and others who are not suitable are accepted become. The benefit is only shown on average, not in individual cases. So if one finds in a representative sample that 95% of people with the characteristic pattern X are suitable for a position, then that does not mean for Mr. Meyer, who has this characteristic pattern, that there is a 95% probability that he is suitable for the position. It only means that if I draw another sample from the same population, I can expect a 95% probability of suitability for feature pattern X (expected value). And then and only when I primarily want to optimize my decisions for aggregates, when I want to hire the largest possible number of suitable people on average, I am allowed to use such statistics.

If this requirement is met, then mathematical models are more powerful than case-by-case methods. That too should make sense. We know nothing at all about the probability with which Mr. Meyer is suitable for the position, although he is a bearer of features because of his bearer of features. We only know: If we only hire people with the combination of characteristics X, then we can expect (assuming a sufficiently large number of recruitments) that 95 percent of these people will be suitable for the position.

Incidentally, this phenomenon can not only be observed in the psychological field, but simply everywhere, e.g. B. in the stochastic determination of requirements in materials management compared to the "pi-times thumb method" (to pick out just one application at random).

Shortened article by Ulrich Gresch in the newsgroups "de.sci.psychologie" 11/13/2007 6:52 pm) Cuts and reformulations by me; W.S.

annotation: This representation corresponds to an empirical-nomological-scientific understanding of science that still dominates today, but which has been questioned by some scientists (especially science theorists) for various reasons in recent years. In particular, the problem of truth, induction and deduction, causality, scientific language and laws were critically examined.

tip: We recommend that psychologists use the Internet-accessible scripts by Uwe Mortensen (emeritus of the Westphalian Wilhelms University), who has basic texts on his homepage (http://www.uwe-mortensen.de/) Philosophy of science as pdf files offers:

Thankfully, all scriptures are constantly being corrected and expanded. There are also others Pearls of the internet to Test theory and test construction, statistics and probability theory, also downloadable as pdf files. Even if the justifications and deductions require mathematical and statistical knowledge, the methods and their meaning and objectives in the context of scientific psychology are written in clear and easily understandable language in which one can recognize the committed teacher.

Contents overview of research methods in psychology and education

content :::: message :::: news :::: imprint :::: data protection :::: author :::: copyright :::: quote ::::