What are involved in the social sciences

as well as online

Hans Peter Henecka

content

1. General aspects
2. Research logic of empirical investigations
2.1 Context of discovery
2.2 Context of justification
2.3 Relationship between exploitation and effects
3. Research methods
3.1 The observation
3.2 The survey / interview
3.3 The experiment
3.4 Action research
3.5 Content analysis
3.6 Sociometry
3.7 The biographical method
3.8 The secondary analysis
3.9 The panel investigation / survey
4. Method problems
4.1 Data collection artifacts
4.2 Data evaluation artifacts
5. Literature

Illustrations:

Fig. 1: The course of empirical investigations based on the research logic "

1. General aspects

The forerunners of today's methods of social science can be traced back in the history of science to the social statistical procedures in the 17th and 18th centuries, where attempts were made to discover social laws in the social, economic and political "mass phenomena" using the "political arithmetic" customary at the time. In the 19th century, inspired in particular by the large-scale studies by Le Play in France and Booth in England, the study method was developed in order to record the living conditions of working-class families resulting from early industrialism and to present them with detailed figures. In Germany it was not until 1908 that the "Verein für Socialpolitik" carried out systematic surveys on the "selection and adaptation of the workforce". At the same time, at the beginning of the 20th century, the survey on the social overview study (survey method) was further developed, which was methodologically refined from the 1920s, especially in the USA, within the framework of community surveys.

While empirical social research was institutionalized and continuously expanded in North America in the period that followed, theoretical and socio-historical analyzes dominated the German social sciences until the Weimar period. The first attempts at developing empirical social research then suffered a severe setback from the National Socialist regime. Therefore, if, in the absence of appropriate research traditions, social research in post-war Germany was initially based more on historically related individual case analyzes or on qualitative methods of data collection and analysis, the development of quantitative empirical methods was soon pushed forward by a tendency towards planning belief in political practice and a correspondingly increasing need for secured data Opinion research data. In the last few decades, German social research has succeeded in connecting to the elaborate standard of the empirical methods developed in the USA through its university anchoring or through the creation of university-related research centers as well as through the establishment of market and opinion research institutes working on a commercial basis. Access, evaluation and use of social science data is now also subject to rapid research changes and further developments in this country.

One of the main problems of empirical social research is still the relationship between methodological research practice and social science theory formation. The rapid development and differentiation of research techniques harbors the risk of the methods becoming independent in the sense of naive empiricism, so that it is less the scientific problems that define the research strategy than, conversely, the accessibility of data and possible applications for research techniques that define the social reality to be examined. In addition, there is a non-uniform understanding of theory among the social scientists themselves. For example, researchers who are strictly empirically oriented use social science methods to back up theories that they themselves understand as systems of empirically verifiable statements about limited facts. Representatives of a more critical-dialectical research direction, on the other hand, strive for a critical assessment of social phenomena in their overall socio-political context in addition to such descriptions and explanations of causal relationships.

Another basic problem is that the full social reality can never be fully grasped, but scientific knowledge is always selectively aimed at a section of the "unmistakable diversity" (Rickert) of the real. It follows from this that one not only has to specify the object of investigation and become clear about the way of its empirical "recording", but also orientate it, together with the selected methods, to a theory or to research hypotheses that are to be subjected to an examination. Without testing it against reality, theories remain speculative and easily succumb to sterile dogmatism. On the other hand, mere collecting of facts by means of a sophisticated methodology, but without theoretical guiding principles, is usually just as sterile, because in this case it remains unclear what to look for and what to look for. The task of the social sciences is to use theory-based methods to make correct statements about important issues.

In this context, the controversy about the advantages and disadvantages of quantitative and qualitative methods, especially since the so-called "positivism dispute in German sociology" (Adorno et al. 1969), should be seen. Quantitative (so-called "positivistic") methods, which strive to describe "objective reality" based on scientific paradigms, a standardized and controllable data acquisition and its evaluation with statistical counting and measuring methods, were sometimes extremely pointed and often polemical alternatives by supporters of a "Emancipatory social research" (critical-dialectical theory, Frankfurt School) contrasted with qualitative (so-called "interpretative") methods by means of which the contents of social reality should be analyzed in a hermeneutical way.

While this radical methodological opposition soon turned out to be sterile and, in this form, also irrelevant for research practice, the current discussion hardly applies to dogmatic fundamental questions of a fundamental superiority of this or that research approach, but rather to pragmatic questions of the appropriateness of qualitative and / or quantitative methods for a specific one Research problem taking into account common basic principles such as validity (validity), reliability (reliability) and intersubjective verifiability. For example, qualitative methods have proven their worth especially in the field of microsociology or organizational research, while, conversely, social reporting, social structure analysis or electoral research still have to work primarily with quantitative methods, as the main thing here is the generalization of data is about populations and precise determinations of distributions. In addition, application-oriented research shows that an adequate understanding of specific social problems is usually only possible through the combined use of different, complementary, but also mutually controlling qualitative and quantitative methods in the sense of a method mix or a so-called methodical triangulation can be reached.

2. Research logic of empirical investigations

In practical application, social science methods should usually be embedded in a research-logical context, the structure and process sequence of which generally consists of three phases, some of which are consecutive in time: the context of discovery, justification, and utilization and effect.

2.1 Context of discovery

Which social problem is "discovered" and selected for an empirical project cannot be justified from the research logic, as this is often determined by extra-scientific or pre-scientific coincidences. It is all the more important that social researchers develop a "feeling" for which problems represent important and meaningful facts with regard to theoretical and methodological progress or with regard to results that can be used in practice. For this purpose, on the one hand, thorough studies of the relevant literature are used, on the other hand - if feasible - a practical, exploratory introduction to the selected "field" is recommended. A professional social scientist, for example, will first have a look around a city district and talk to some residents before he begins the actual theoretical and methodological conceptualization of his renovation study.

2.2 Relational context

As an introduction to the actual empirical investigation, the social problem to be investigated must first be "translated" into a (testable) social science question. In practice, this step takes place with the search for plausible and provable assumptions about social relationships, dependencies or regularities. Since these are generally variable quantities that are only delimited more precisely by the question posed by the individual researcher, one speaks of variables. In addition, it is usually not only two or more variables that come to the fore as objects of investigation, but the relationships between the individual variables are also determined more closely. A hypothesis always contains a statement about which variable is to be regarded as a causal factor (independent variable) and which as an effect factor (dependent variable). However, since most of the variables are in a complex interrelated relationship and can therefore be understood as dependent or independent variables depending on the perspective and the question, it is up to the researcher to decide which aspect to investigate primarily. For example, "political participation" can be viewed as an independent variable for the dependent variable "democratization of societies"; On the other hand, "political participation" could itself be examined as a variable dependent on certain socialization experiences, personality factors and the position in the social system. The term "independent" is therefore not to be taken literally: it only refers to the theoretical frame of reference of a certain hypothesis and means that a certain variable influences the other variable, but does not examine it further in the context of the hypothesis for its own dependencies becomes.

The real importance of research hypotheses lies in the incentive to review their statements. Therefore, the theoretical terms used to represent the facts to be investigated must be concretized on the empirical level, i.e. H. be defined and specified as clearly as possible using indicators ("operational definition"). However, since the respective research operations are also determined by appropriate indicators, it is necessary to examine them for their validity and reliability. The indicators are actually supposed to measure what they claim to be measuring or, under the same conditions, should produce the same values ​​with repeated measurements of the same facts. Social issues, which by means of "receptors", i. H. that can be systematically recorded with natural sensory organs or with artificial means are both objective facts (e.g. age, education, occupation, income level, family size, party affiliation) as well as subjective data (e.g. ethnic identity, ecclesiasticalness, attitudes to changing values, etc. .) as well as real actions and behaviors that are also relevant from a social science perspective in connection with subjective data. While objective facts are relatively easy to operationalize in their concreteness and thus evade any interpretation, subjective data are much more difficult to transform into operational definitions using indicators and thus to make them accessible to an intersubjective control.

In connection with the definition of the method or the controlling and supplementary methodical triangulation, a decision is usually also made on the observation units. For example, one can choose a qualitative case study to describe a certain "typical" social context in detail, in parallel or alternatively, however, also - via a sample taken from the population according to certain statistical selection procedures or by examining the population itself - Strive for representativeness that allows the results to be generalized.

The actual data collection takes place in the form possible and determined by the selected method. The data is then coded and mostly evaluated using relevant social science computer programs. Finally, the interpretation should discuss the methodologically obtained results against the background of the theoretical considerations expressed in the research question, whereby it applies that the results obtained are only provisionally "verified"; they are regarded as "proven" as long as all other alternative hypotheses that are currently seriously discussed are "falsified", but the thesis represented has withstood "attempts at falsification" itself. If empirically based statements are refuted at a decisive point by newer or more extensive research results, they must be rejected, modified or relativized depending on the quality and extent of the objection.

2.3 Relationship between exploitation and effects

In order to convey the generally relevant findings of a research project to the interested public and to make them usable for social practice, a corresponding graphic processing of the data and a transfer of the results from the technical jargon of the social sciences into the public language is often necessary. If a researcher does not want to take the incalculable risk of someone else arbitrarily disposing of his data and results, he will be well advised, in particular to critically monitor the processes of socio-political impact and the exploitation of his findings.

Fig.1 The course of empirical investigations based on research logic
View image in original size

3. Research methods

A large number of research methods are now available for social science research practice, which are often arranged and used in a research-specific mix. The most important methods are shown below:

3.1 The observation

The observation (Bb) of social facts and the corresponding description of the observed are often at the beginning of empirical research. Quite different social situations and contexts can be recorded, such as whole cultures (ethnography), but also religious sects, social movements, citizens' initiatives, decision-making processes in local politics, the behavior of teachers and students in class or the development and dynamics of a conference group. Since basically all perceptions are selective, scientific Bb must also assume that social situations can never be fully grasped in all details. In contrast to the unsystematic, arbitrary and usually also unconscious selection of environmental stimuli in the naive everyday observation, the empirical selection process in the systematic Bb is deliberately planned through a precise definition of the observed actions, events and content using appropriately standardized category schemes. Nevertheless, due to the fact that the observer's perceptual coloring or narrowing of the perception is relatively difficult to control and eliminate, due to his personal attitudes or pre-scientific expectations with regard to the examined object, dangers to the reliability and validity of Bb results can arise. Usually, the Bb of the immediate linguistic behavior will be in the foreground as an analysis unit, but also extra-linguistic behavior (such as volume, modulation, speaking time, vocabulary), non-verbal behavior (such as certain gestures, facial expressions, eye play and general body language) as well as specific personal, Spatial and time-related behavior are also examined.

Methodologically, Bb can be set up in quite different ways. In the participating Bb, the researcher tries to integrate himself into the group to be examined in order to gain the deepest possible understanding of the subject under investigation through a high degree of participation, which, however, also harbors the risk that the reliability of the Bb will suffer due to excessive identification can. In the case of the non-participating Bb, the social researcher approaches his Bb object from the outside for reasons of practicability of access or also out of concern for his own objectivity. Furthermore, Bb can be controlled or uncontrolled; In other words, they can be obtained on the basis of a relatively narrowly predefined Bb structure (e.g. to record quantifiable elements) or in a more relaxed and impressionistic way that serves the general purpose of the investigation (e.g. to record primarily qualitative characteristics) become.Furthermore, they can be carried out openly or - if bias effects are to be expected in the test subjects - also covertly, whereby legal and ethical aspects must be taken into account for the latter.

3.2 The survey / interview

Systematically questioning people by means of casual individual discussions, group discussions, questionnaires or interviews is still the most popular and most frequently used way of empirical social researchers to collect socio-scientific and socio-political relevant data.

Surveys (Bf) can be made orally or in writing. In the case of the oral questionnaire, an interviewer (usually trained) asks questions to a person or to a group called together for this purpose; the answers are usually recorded immediately by the interviewer. Since the 1970s, however, written Bf has dominated, primarily for reasons of pragmatic research and as a result of improved and more interview-friendly survey techniques ("total design method"), in which the test subjects fill out a questionnaire independently, i.e. without external mediation or consultation. This requires in particular that the questions asked are phrased precisely and at the same time (with regard to the target group) as understandably as possible.

Bf can also be more or less standardized. In the completely open and free, "real" interview (I), only the subject area is roughly given; The interviewer who works exploratively is required to have a high degree of conversational skills as well as knowledge of the research goal and the theoretical framework. The narrative or depth I, in which the I-partners are asked to spontaneously tell stories about their biography, their role in certain political processes or the like, for example, with minimal guidance of the conversation, is considered to be the Bf-form increasingly used in qualitative projects . The directed or centered I requires a guide for the subject areas to be addressed; however, there is no differentiated "battery of questions". The desire for the greatest possible formalization ultimately leads to the structured I, in which the wording and the order of the questions are precisely defined, or even to the completely standardized I, in which the exact reaction behavior of the interviewer is also specified.

Contrary to popular belief, setting up a usable questionnaire and properly handling the collected data without the advice of social science experts is a very difficult task.

The actual construction problems and evaluation questions are so varied that in the meantime a professional methodology of the interview has developed, which provides a system of rules to eliminate sources of error as far as possible and to make the research process transparent and controllable.

Problems and sources of error arise mainly from the fact that the applicant is a communicative process in an unusual situation, which can be influenced depending on the subject of the survey and the relationship between the interviewer and the person questioned (reactive method). In addition, only statements, possibly attitudes and opinions, but not actual behavior are shown at Bf. Since Bf results in particular are increasingly used as aids to argumentation and decision-making in politics, administration, planning and the media, everyone who is confronted with Bf results in any way has to be aware that such findings only ever before the respective person Background of their serious, open and concrete survey and evaluation problems can be assessed.

From a critical public, Bf are increasingly problematized "politically" with regard to data evaluation. For many citizens, the processes of processing the data collected are no longer transparent. Despite the data protection act and the guaranteed anonymization of data, the fear of "data traps" and uncontrollable "checks" is growing, while at the same time the willingness of many citizens to provide information is correspondingly reduced.

3.3 The experiment

Following the example of the classical natural sciences, the experiment (E) is considered to be the most reactive form and at the same time the safest method for establishing causal relationships in the area of ​​social phenomena. According to a previously established test plan, at least one independent variable is varied according to plan and under controlled conditions, and the effects of these changes in a dependent variable are observed and measured.

The essential characteristics of the E are predictability (control), repeatability (replication) and variability (manipulation) - criteria which, however, are often difficult or impossible to meet, since in the social area "artificial" or "isolated", that means that exact laboratory-like experimental situations can only be produced to a very limited extent. Not least for this reason, social science studies are more often carried out while largely leaving the natural environment intact (field experiment). In these "natural" (ex-post-facto) experiments, the change in situation due to the presumed cause has already occurred due to social processes. For example, the sympathy curve for the X party (dependent variable) during different phases of an election campaign based on interviews or subsequent file studies or content analyzes of certain current events and changed election spots, with the participation of the top candidate in a television discussion, i.e. from changes in the "independent variables" , can be explained causally. While laboratory experiments can usually claim a high degree of internal validity with regard to the examined situation, field experiments in natural situations have the advantage of higher external validity, as their results allow generalizing statements about real social conditions and relationships. On the other hand, it has recently been objected to these classic experimental procedures that most concrete social phenomena can be recorded less as a cause-and-effect sequence, but more as a systemic network - a demanding new experimental research program, which has so far only been carried out on the model level via computer simulations could be realized.

3.4 Action research

Action research (Af) was established in Germany under the influence of critical theory as an innovative research program that sought to combine the acquisition of knowledge with a simultaneous improvement in the social conditions examined (e.g. slum rehabilitation, dismantling xenophobia). The ideal case for this direct link between research method and practical social reform is a cooperative social structure of relationships between researchers and those affected. The researchers are at the same time observing and actively acting participants in the social processes to be examined and changed, while those affected by the program (e.g. the slum dwellers) for their part not only act as "data providers" but as "subjects" actively and continuously in the planning , Implementation and evaluation of the research. Af is particularly useful when certain social problems can only be solved with the cooperation of those affected. Since the social field is consciously changed by this research practice, a hypothesis test with this method is not possible.

3.5 Content analysis

The content analysis (Ia) is a method developed especially in mass communication research. It assumes that written, oral or even visual communications have an important social function and that the symbols conveyed in the communications that have become manifest are indicators of attitudes, opinions, values, tendencies and intended effects, of prejudices or other non-directly identifiable properties of the respective author represent. The focus of the Ia is the question: Who tells whom, what, how, with what intention and with what effect? Analysis objects can be minutes of meetings, contracts, speeches, letters, diaries and autobiographies, certain newspaper articles or entire volumes of newspapers and magazines, posters or advertisements, brochures and newsletters, hit texts or graffiti on house walls, subcultural slogans or bumper stickers, certain radio or television programs, etc. .

Problems can arise with regard to the establishment of the contextual reference or rather the situational reconstruction of the meaning of the research material. Questions relating to reliability can also arise, as the assessment of the categories of the investigators can vary both among themselves and over time. Certain elements of the understanding of the text, such as ironization, alienated speech and the like, are relatively difficult to grasp analytically. Since the basic text material is often quite extensive, practical problems have arisen time and again with regard to the representativeness of the text selection (sample problem) and the quantitative data processing, but the problem of text volume has largely been solved today through the development of automatic methods of text coding and text analysis.

3.6 Sociometry

Sociometry (S) is a measurement method for the quantitative recording and representation of certain affective and / or functional aspects of interpersonal relationships in a group. Originally developed in the USA by Jacob L. Moreno in the 1930s with social therapeutic intentions, this method has now analytically proven itself wherever tensions between formal (e.g. prescribed by work organization) and informal (i.e. spontaneously and voluntarily occurring) group structures and relationships grow up. The S is particularly used in industrial sociology and in the educational field, for example in the social fields of action (workshops, offices or school classes) to highlight the importance of informal communication and information processes and the corresponding positions of authority or informal management structures for the general group climate as well as for performance of the group and individual group members.

With the help of a relatively easy to carry out "sociometric test", the social interrelationships between the group members are measured using the criteria of preference, indifference and rejection. This is achieved by asking the group members in writing about their interaction preferences or factual interactions, which other members of the group like or dislike them most, as partners in certain situations (e.g. at work, in their free time, on vacation) or who they actually interact and communicate with and with whom they don't. So all kinds of election acts are asked, some of which can manifest themselves in an undifferentiated manner, some of which directly refer to the affective dimension of popularity / sympathy ("Who would you consider your best friend?") Or to the functional level of performance ("With whom Would you like to work together on this task? ") within the group. Often the relational expectations of the group members are also ascertained through corresponding questions about the presumed results of the sociometric test regarding oneself ("What do you think, how often you are mentioned in this context?"). The frequency with which individual group members unite positive or negative elections is tabulated in a "sociomatrix", on the basis of which individual group positions can be characterized in more detail statistically. In addition to such statistical parameters for the analysis of sociometric results, the sociomatrix is ​​usually converted into a so-called "sociogram", with a variety of options for graphical representation (column, circle, profile, coordinate sociogram, etc.).

3.7 The biographical method

The biographical method (BM) aims to reconstruct the social reality of everyday life by analyzing biographical material and the evaluations, opinions, attitudes and events included therein. These, already from sociological classics of the Chicago School such as W.I. The qualitative method used by Thomas and F. Znaniecki in the early 1920s, according to which objective social conditions are reflected in subjective phenomena of consciousness, only had the status of an auxiliary instrument that gained material and generated hypotheses until the 1970s. In particular in response to macro-sociological, systems-theoretical and quantitative research approaches as well as under the renaissance of phenomenological sociology and the spread of symbolic interactionism during the 1980s, biography research developed as an independent, interdisciplinary research field with BM as the central analytical technique.

On the one hand, written résumés, memories, letters, and diaries, which are then evaluated in particular for content analysis, are considered material of the BM On the other hand, in recent years, narrative and in-depth interviews have been increasingly used as a supplement or alternative, in which the telling of life stories is considered a mode of material and data acquisition.

The advantage of this procedure is its practicality; What is critical, on the other hand, are the modifications or distortions of autobiographical reality possible through subsequent rationalization of the test subjects. It is true that the human ability to (re) construct one's own biography "meaningfully" as a result of constantly changing reference points and reflections in the course of one's life cannot be blamed on fundamental errors. But it cannot be determined hermeneutically when earlier events are relativized to later ones or when unpleasant insights are replaced by more pleasant ones, for example in the sense of reducing cognitive dissonance. Although the way in which social reality is subjectively processed is of fundamental relevance for social science issues, the assessment of the productivity of the BM with regard to the reconstruction of social specimen life courses is controversial.

3.8 The secondary analysis

The secondary analysis (Sa) is a method of using the mostly quantified data or data collections (primary analysis) already collected by other researchers or institutions for your own questions. These can be official (e.g. Federal Statistical Office) or non-official data sets from completed surveys or empirical studies that are sooner or later the scientific public in corresponding data pools such as the Cologne Central Archives for Empirical Social Research (ZA) as a kind of collective property be available. Especially in the field of international comparative research, the SA for checking and accumulating results has gained in importance in recent years.

Since the respective data material is dependent on the research objectives of the primary surveys, independent problem formulations and questions under a new theoretical reference system are, however, subject to content and methodological limits at the SA. Your advantage lies in the saving of time and costs that were incurred for the primary survey and the preparation of the data material.

3.9 The panel investigation / survey

In order to analyze changes in attitudes and behaviors over time, panel research (PU) is used, particularly in market and opinion research. Once selected, a group of people is examined at different times using the same survey instrument. The PU thus enables a diachronic analysis (longitudinal section) of the examination variables of interest under changed environmental conditions or new external influences. With a PU, the short-term and long-term effects of social change in particular can be empirically mapped. Since 1984, for example, the "Socio-Economic Panel" has been carried out in the Federal Republic of Germany, which was expanded to include the new federal states from 1990 onwards and which now involves the changes in living conditions and ways of living through the annually repeated survey of all members of around 8,000 households is continuously observed and analyzed both in the course of the life of individuals and in the context of their family or partner relationships.

Critical with the PU are the high costs and the so-called panel mortality (failures of test persons, refusals to answer, etc.), which can trigger representativeness distortions and require time-consuming "panel maintenance". Other panel problems are effects that are caused by repeated questioning of the same test person and that can lead to changed reaction behavior. A variant of the PU is the Survey (S), which as an overview study or trend analysis is also referred to as "fake" PU. In contrast to the PU, the S is a repeated (replicative) investigation of the same problem with a time interval, but with different samples. This means that the PU is no longer able to identify changes in individual individuals. Only changes between the respective groups of respondents can be ascertained.On the other hand, the S is recommended in comparison to the PU from a research-economic point of view. S-examples are the election studies that have been carried out since 1961 in connection with the federal elections, which provide information about the conditions of voting behavior and voting-determining factors, the welfare surveys that have been carried out at irregular intervals since 1978 on the development of the welfare of the population resulting from objective living conditions and subjective evaluations , the general population survey of the social sciences (ALLBUS) carried out every two years since 1980 or the family survey carried out for the first time in 1988 and supervised by the German Youth Institute.

4. Method problems

As a result of deficiencies or unintentional errors in the application of social science methods, facts can be suggested, i.e. so-called research artifacts are caused, so that the research result is deceptive and no adequate answer to the original research question is possible through the result statements. Experience has shown that such "art products" are often based on sources of error in the area of ​​data collection and analysis.

4.1 Data collection artifacts

Since most social science studies focus on social interactions and communication between at least two people, it can lead to mostly unconscious and difficult to control distortions or falsifications in the planning and execution of experiments - for example by the person of the experimenter: under such examiners effects are extremely difficult influences understood by external characteristics of the experimenter such as age, gender; Habitus (experimental effect), but also through unconsciously communicatively transmitted signals and cues of a paralinguistic (tone of voice, emphasis, pauses, etc.) or mimic-gestural type (smile, nodding, visible relaxation when reaching the "correct" frown, nervous clearing of the throat) in the case of the "wrong" answer, etc.) are conveyed (experimental bias). In particular, the "experimenter bias" can significantly influence the research situation through the subliminally communicated expectations of the experimenter, his knowledge of the investigation intentions, etc. and "unintentionally" lead to a confirmation of the hypotheses.

The effects of the halo effect in the process of observation are also known. Analogous to the "halo" - the halo of sun and moon, which supposedly allows exact weather forecasts - the true assessment of individual properties of people (or objects) is erroneously guided by the subjective "radiation" or "overexposure" by other prominent features in the vicinity observed facts or from an implicit "overall impression" on the part of the investigator. Characteristics that emanate and lead to selective perception processes are often age, gender, appearance, clothing, social origin, denomination, occupation, language / dialect, skin color, etc. the categorical classification of certain social facts errors occur more than randomly in the sense of expectations and hypotheses. There are also potential tendencies on the part of the test subjects not to react adequately and flexibly to stimuli offered in certain situations, but rather to follow certain patterns or schemes (response sets). In particular in the case of questionnaire surveys with given yes / no answers, yes-saying tendencies can be identified more often, regardless of the content of the questions presented. In a similar way, test subjects can, in contradiction to their true convictions, align their expressions of behavior and opinions in the sense of social desirability to the expectations of the social environment they suspect, for example of the test director or the client.

4.2 Data evaluation artifacts

The preparation and further processing of information that has been collected and encrypted to form data is usually carried out in the research process using statistics. Typical translation problems in this mathematical area arise here from the change from the linguistic to the numerical level, so that, for example, a critical check must be made as to whether the intensity difference between the answer categories "never" and "rarely", which is statistically exactly the same and expressed by identical numerical relationships, is real is exactly the same as the one between "often" and "very often". What is "often" for one respondent may only mean "more often" for another. In addition, pseudo-accuracy is simulated in many tables when several places after the decimal point are given as a percentage, where one place after the decimal point is "exact" and even percentages rounded to whole numbers are often sufficient.

Statistical methods are therefore not simply applicable automatically; rather, in order to circumvent artifacts, it is necessary to re-examine for every social science investigation to what extent they are suitable as "models" for the facts under investigation. Are z. If, for example, differences in the behavior of different categories of people, e.g. between men and women, Catholics and Protestants, adolescents and adults, can be determined in an investigation, a so-called significance test will have to be used to decide whether these differences are just random or just as large (ie statistically significant) that one can make statements about more general hypotheses on their basis. Due to methodological misunderstandings, however, significance tests are often not used as a formal decision-making model about hypotheses, but rather as a measurement method for designating the quality of certain effects.

Finally, it is not uncommon for correlation coefficients to be interpreted causally as a measure of the statistical relationship between two or more variables. For example, from a given relationship between X and Y, it is concluded that X is the cause of Y. Apart from the fact that Y could theoretically also cause X under certain circumstances, on closer examination some established relationships between phenomena only turn out to be a sham correlation, so that With every combination of features one has to check again and again whether an initially hidden, but then to be controlled, third, "intervening" variable (Z) refutes the assumption of a causal relationship between X and Y or at least partly explains it.

Practice shows time and again that statistically significant relationships often do not offer any causal explanatory models, but can only be ostensibly conveyed through calculations. This means that the process of evaluation and, above all, the interpretation of data includes steps to examine the collected material under other interrelationships and relationships than those initially recognized as meaningful. In order to avoid artefacts, an explicit presentation of the selected theoretical perspectives that can be verified by the scientific community as well as a differentiated documentation and discussion of the methodological decisions made in the research process should meanwhile become a standard.

5. Literature

Adorno, Th.W., et al. (1969): The Positivism Controversy in German Sociology. Neuwied, Berlin.

Dollase, R. (1976): Sociometric Techniques. Weinheim, Basel.

Dreier, V. (1994): Data analysis for social scientists. Munich, Vienna.

Engel, U. / Reinecke, J. (1994): Panel analysis. Berlin, New York.

Friedrichs, J. (1990): Methods of empirical social research, Reinbek.

Froschauer, U. / Lueger, M. (1992): The qualitative interview. Vienna.

Früh, W. (1991): Content analysis. Munich.

Grümer, K. W. (1974): Observation. Stuttgart.

Kohli, M. / Robert, G. (Eds.) (1984): Biography and Social Reality. Stuttgart.

Költringer, R. (1993): Survey Data Validity. Vienna, Cologne.

Kriz, J. (1981): Method critique of empirical social research. Stuttgart.

Kromrey, H. (1998): Empirical Social Research. Models and methods of data collection and data evaluation. Opladen.

Lamnek, S. (1995): Qualitative social research. Weinheim.

Moser, H. (1977): Methods of action research. Munich.

Noelle-Neumann, E. / Petersen, Th. (1996): Everyone, not everyone. Introduction to the methods of public opinion polling. Munich.

Quick, R. / Hill, P.B. / Esser, E. (1993): Methods of empirical social research. Munich / Vienna.

Stelzl, I. (1982): Errors and pitfalls in statistics. Bern, Stuttgart, Vienna.

Zimmermann, E. (1972): The experiment in the social sciences. Stuttgart.