Empirical Research on Human Resource Management as a Production of Ideology

The article argues that empirical research on Human Resource Management creates a one-sided, distorted image of the reality of work and thus generates ideology. Such an ideology legitimises HR practices and favours the interests of entrepreneurs and managers. This assertion is illustrated and discussed using the case of empirical research in HRM in German-speaking countries, although the ideology assertion should also be valid for Anglo-Saxon countries. It is shown that HRM research mainly follows employer objectives; it primarily analyses performance-related variables. The surveyed HR practices focus on " High Performance Works Systems " , while other HR practices are largely ignored. Mainly organisational elites (managers, experts and other highly qualified employees) are surveyed as personnel and provide information about the situation in companies. Empirical research paints a unitaristic picture; depicting the employer and the role of HR management positively. Deviations from an employee-friendly HR strategy are overlooked or seen as relatively rare anomalies.

So few reports.So few questions.(So wenig Berichte.So wenig Fragen.)(according to Bertolt Brecht and Volker Braun)

Research objectives and approach
The thesis of this article is: the empirical research in Human Resource Management 1 creates a wrong image of the reality of work.This image serves as the affirmation of relevant actions and values related to human resources -especially in the interest of entrepreneurs and managers.In short -current empirical research in HRM is generating ideology.
I contend that research in Human Resource Management (HRM) mainly focuses on economic aims and employer objectives and rather neglects employee interests as a starting point for research.Mainly those variables are analysed which refer to work performance.Priority is given to "High Performance Work Systems", i.e. clusters of HR practices including comparatively favourable features for employees.Less favourable HR practices, which are by no means characterised by a careful handling of "human resources", are omitted.HR strategies which run contrary to employee interests and can be observed at discount chains like Aldi, Lidl and Schlecker (see e.g.Voss-Dahm, 2009;Bormann, 2007) or at companies like McDonald's (Royle, 2000) -to take just a few examples here -are ignored to a large extent.When these strategies are brought into focus, they seem to appear as rather scarce and disappearing phenomena in the general view, and which are practically swept away by the competition for "valuable human resources", the so-called "War for Talent" (von der Oelsnitz, Stein, & Hahmann, 2007).The information generating the empirical view, i.e. the image of corporate HR management, mainly derive from the organisational elites which also form the group that empirical HRM research focuses on.All these points together generate an euphemistic view of corporate HR management which also contributes to firewalling given practices against criticism and changes.
Such ideological HRM research 2 would clearly contradict the ideal conceptions, which may be shared by many, if not most people engaged in empirical HRM research (for aims and objectives of HRM research cf.e.g.Matiaske, 2004;Drumm, 1993;Martin, 1994): The purpose of empirical research should be to formulate (theoretically based) empirical hypotheses and test them in order to describe and explain reality.Empirical statements can be interesting as descriptions of conditions, changes or relations between elements of reality and can be consulted in order to corroborate theories (or as some would argue: in order to develop theories on the basis of empirical data).In addition, they could form the starting point of the development of practical respectively social-technological statements or the criticism of conditions.Most scien-1 When speaking of HRM research here and in the following, empirical research is always implied.
The term "ideological research in HRM" is shorthand for "research in HRM generating potential ideological knowledge".management revue, 22(4), 367-393 DOI 10.1688/1861-9908_mrev_2011_04_Nienhueser 369 tists may presumably agree that HRM research should generate a true view of the real working conditions and HR management -despite different methodological views.There might be a certain agreement on these assumptions regarding the purpose of empirical research. 3 Finding an answer to the question of value issues (normative statements and their meaning and relevance for empirical research) is more complicated: Whose interests constitute the starting point for empirical analyses; whose interests do the results serve?Every scientist is inevitably selective in terms of choosing his or her research question, theories and methods etc., and valuations are therefore unavoidable: "Like any other scientific activity or any other praxis social science research depends on different value perspectives.… Rules for orientation are related to certain objectives.Decisions constantly have to be made within their framework depending on the corresponding valuations, decisions on the selection of scientific problems, suitability and usability of hypotheses and theories, as well as the adequacy of statements, and relevance of observations and acceptability of methods" 4 -those are the comments of Albert (2000, p. 47), a proponent of the postulate of value neutrality.Scientific research in application-oriented sciences such as business administration as well as research in Human Resource Management is effectively more application-oriented and less oriented towards basic research (i.e.developing and testing abstract theories).It is obvious that the objectives of users of these findings generate and determine research projects, that scientists unconsciously or consciously adopt certain goals as their own, or even just use them as a hypothetical basis.Therefore, some of these projects will reflect the interests of employers while others will reflect employee interests. 5If the value foundations (as necessary and unavoidable basis for decisions on research questions and theories as well) led to research projects always being approached mainly from the perspective of a single interest group, this one-sidedness would -assuming a conflict of interests -disadvantage the corresponding other groups.This would not comply with the regulating ideal of pluralism as model of scientific epistemology (Spinner, 1974).
If it were the case that empirical HRM research produced a one-sided, biased, and therefore false image of reality, which, in addition, would also be suitable to unilaterally affirm certain actions and values in particular for companies and managers and that the findings generally did not aid in improving conditions for the employees, then it would be justified to say that HRM research generates ideology.
I would like to emphasise one point here, which I will also refer to later: My concern is the collective body of statements of empirical research and which image of the re-3 The ideal conception of empirical science has been expressed rather simply here; each individual point may surely be in dispute on closer inspection, e.g. a dispute between a rather inductive and a deductive orientation of empirical research, or between those scientists following "Critical Theory" and such scientists following "Critical Rationalism".

4
All translations from German to English are mainly done by Anke Wiggers (thanks a lot) and in part by me, Werner Nienhueser.All mistakes are mine.

5
To simplify matters I am ignoring other interests, e.g. of consumers.In addition, I am arguing on the assumption of a conflict of interests between capital and labour.ality of HRM and working life this reflects.Individual statements like: "x per cent of employees are satisfied", "y per cent have resigned", "z per cent have a high level of stress in the workplace" or "the correlation between satisfaction and performance in the sample is r = 0.XY" may well apply, or not -but this is not the bottom line.It is more important to say how selective, biased and one-sided the entirety of the statements is that constitute the "image" of HRM (in the broader sense) and why such an image comes about.I will therefore not label individual statements and studies as ideological ones.
I will proceed as follows: First I will outline the framework of reference, on which my analysis is based.This includes an explanation of the concept of ideology which I use.After that, I will go into the methodological and empirical basis of my analysis as well as some problems related to it (Section 2).This is followed by the actual analysis (Section 3).I attach importance to those areas of research in particular that are related to the exchange relationship in companies.Exchanging labour against wages (the reward-effort bargain) and the legitimisation of this exchange relationship make up the substance of what corporate HRM deals with.In Section 4 I will summarise all findings and try to describe the image that empirical HRM research draws of "Human Resource Management", and what the implications are for justifying the constitution of the exchange between capital and labour.

Analytical framework, conceptual and methodological definitions 2.1 Analytical framework
The basic idea of the analytical framework 6 is first of all that the socio-cognitive conditions of reproduction of empirical research affect and influence views of corporate HRM.Second, these views themselves develop impacts which I define as ideological.The images of corporate HRM are not generated solely based on results from empirical research.To do so would overestimate their influence.Images of exchange between capital and labour are more strongly affected by general world views, which in turn were generated by socialisation and selection processes.People who are imprinted by a liberal (market) world view will have a different perception of the corporate exchange relationship than someone whose world view focuses on the importance of irreconcilable extremes between capital and labour (Figure 1).
The assumption that the socio-cognitive conditions of reproduction influence certain views of corporate HR management is of particular significance for my essential argument.They affect the "production" of the cognitive (to some extent also normative) "raw material" of our view of HRM.I include the following items under the term socio-cognitive conditions of reproduction: (i) selectivity with regard to topics and the type of questions of empirical HRM research, (ii) valuations in the interest of the employers, (iii) the focus on certain research objects (e.g.mainly on specialists and executive staff) as well as (iv) selectivity arising from restricted access to informa-6 A theoretically based explanatory model of the development and effects of ideologies would be preferential to an analytical framework, but there is no space here for developing such a model.tion (e.g. when mainly those companies are studied where good working conditions and little conflict already exist).In addition, I also consider a theoretical foundation which includes one-sided world views as socio-cognitive conditions of reproduction.Except for the last point all of these items serve as criteria for my subsequent analysis of empirical HRM research.I am excluding the theoretical foundation here because I would like to place more emphasis on the fact that "pure" empiricism itself can be ideological -I will refer to that point later on.I call these conditions socio-cognitive because I consider the relationship between the individual level and the social level as important.World views are individually cognitively rooted.The selection of a research question, for example, is a cognitive process, which is affected by general world views.I am less concerned though with the individual level and that single researchers are selective -they necessarily have to be because selection cannot be avoided -but rather with the selectivity of the social aggregation, the group of empirical researchers as a whole.This selectivity is a social phenomenon based on individual cognitive processes, but which nevertheless cannot be reduced to these.Instead it is social processes which play an essential role, e.g.communication strengthening selectivities, external or self-selection of individuals who are not sufficiently socially/politically adapted, etc. Due to the non-additive aggregation of individual selectivities an image (possibly several images) of the exchange relationship between employees and management (resp.labour and capital) is generated.I use the term of conditions of reproduction to define the interrelation, the mutual strengthening of already existing cognitions, theoretical concepts on the one hand, and questions as well as results of empirical research on the other.

The concept of ideology
The term ideology is often used as a polemic fighting word discrediting all those value-information-systems as ideological that do not correspond to one's own views.Terry Eagleton's drastic formulation gets to the heart of it: "… nobody would claim that their own thinking was ideological, just as nobody would habitually refer to themselves as Fatso.Ideology, like halitosis, is in this sense, what the other person has" (Eagleton, 1991, p. 2).As we will see, such a purely denunciatory concept makes no sense, and neither does the wide understanding that conceives ideology generally as a more comprehensive value-knowledge system, as a world view or system of thought.
Studies dealing with the concept of ideology focus mainly on its development and use.Theories of ideology in the sense of a system of assumptions on the generation and effects of knowledge grouped under the term of ideology are rare (Eagleton, 1991;Sorg, 1976, Lieber, 1985, Hauck, 1992, Ritsert, 2002; see also the discourseoriented approach of van Dijk, 1998) 7 .We can differentiate between three types of definitions of ideology (Lieber, 1985;Eagleton, 1991;Gröbl-Steinbach Schuster, 2011; on the difficulties of a definition also Jost, 2006).
A first type of the notion of ideology neutrally defines any thought or, even better, value-knowledge system of social groups.According to this view, ideology is a cognitive construct which contains both valuing and explanatory elements (see Jost, Nosek & Gosling 2008).This notion of ideology is, however, only another way of describing general thought systems shared by certain social groups.If one were to conceptualise the term in this manner, one could do without it and replace it with one of the aforementioned terms.
Building on the first type, a second type adds that ideologies develop a stabilising effect on societies or, more generally, social systems, by giving people a sense of orientation and endowing them with a sense of cohesion.An example for a corresponding definition is as follows, "…ideologies may be … defined as the basis of the social representations shared by members of a group.… ideologies allow people, as group members, to organize the multitude of social beliefs about what is the case, good or bad, right or wrong, for them, and to act accordingly" (van Dijk, 1998, p. 8).
A third type, which is most aptly represented by the ideology-critical perspectives of Critical Rationalism, addresses the truth or falsehood of statements and the possibility of their falsification (see e.g.Degenkolbe, 1965;Schmid, 1972).According to this view, a statement is ideological when it becomes impervious to criticism by virtue of being devoid of meaning.Although the political effects of ideologies are not ignored (see e.g.Salamun, 1988), the analysis concentrates more on the characteristics of statement systems and less on their effects.
My view of ideology is in line with this third definition, but I emphasise more strongly the effects or functions of distorting or obscuring reality and the legitimisation of sovereignty on the collective level (similarly Elster, 1986, p. 462).An analysis based solely on the characteristics of the statements (such as being devoid of meaning) would be too superficial as I assume that it is the body of statements in particular as it is assembled through selection, which has the ideological potential and it does not necessarily have to be the empty phrases that carry the ideology.7 Jost, Nosek, and Gosling (2008) postulate the end of the "end of ideology" and show that social psychology in particular is now increasingly addressing ideologies, how they came about and what their impacts are.They are ultimately looking for a theory of ideology which explains certain impacts that justify the system.Thus, I will use the term "ideological" to refer to the body of statements which (i) suggests truth but is at the same time difficult to falsify and even immunize themselves to falsification, (ii) creates an image that is at least partially false and furthermore, (iii) has the function of obscuring or affirming certain interests of assignable social groups.So, my concept of ideology lies between a purely neutral one, which could be replaced by that of a world view or something similar and a denunciatory-negative variation (as referred to by Lieber, 1985), which assumes that reality and certain valuations are intentionally obscured by the dominant parties.The intent to obscure need not be excluded; however, it is not a required definition attribute of ideology, as a set of statements can unintentionally produce positive effects for one group (and negative ones for others). 8 In the following, I will explain in more detail a few of the aforementioned points and present the methodical conclusions.First, I will discuss my assumption that it is necessary to analyse not individual statements, but the body of statements as a whole.Second, it is necessary to clarify what is meant by empirical HRM research and what body of statements I am referring to.Third, the relationship between the characteristics of statements and their effects in a social context must be clarified as statements have no effect in a social vacuum.

The collectivity of statements
As previously mentioned, I am concerned with the overall picture that the collective body of statements of empirical research brought forth.Individual statements play a role only as elements of the overall picture and must be viewed in context with others.Empirical results contribute to the formation of a picture of the working world; we look for patterns, a shape, and in the search for a coherent picture, we also consider empirical results.Thus, all statements in HRM research can be empirically true in and of themselves, but still present a false overall picture when certain aspects do not appear systematically.This means that, in an extreme case, ideology can be based on a set of statements which are individually true, but the overall picture can be false, i.e. it does not accurately represent reality.In comparison: assuming we want to create an "overall picture" of our life and our relationship to our partners, children, etc. and we would use only photos taken on holiday, Sunday outings and similarly relaxed situations.With this approach, we would hardly get an accurate picture.Who takes pictures of themselves and their adversaries in the midst of a marital row or a conflict with their pubescent children? 9 In particular, the fact that at least individual statements of 8 I also do not understand ideology to be a false consciousness; therefore, it is not about individuals as the medium of certain value-knowledge ideas and a differentiation between ideologists and non-ideologists is unnecessary.At the same time an explanation of the formation and the effects of ideologies that is well-grounded by action theory does not get around making statements about the producers, recipients and beneficiaries.This project of a theory of ideology can at best be touched on here.-Ideologies are neither "reflections" of material circumstances as (vulgar) Marxist interpretations suggest.9 This idea is from the cabaret artist Eckard von Hirschhausen.
an overall picture are usually accurate can complicate the criticism of the falsehood of the overall picture. 10ethodically, this means that it is difficult to prove or to illustrate the element of the ideological, as I understand it, using individual studies.Therefore, I will also mention individual studies as examples which are representative of substantial pieces but which cause distortion when assembled to an overall picture.A test of the ideology thesis would be possible using a systematic quantitative analysis of the contents of a specified number of publications.However, I am attempting here more of a qualitative appraisal.For this analysis, I generally refer to three types of materials to illustrate and support my contentions.First, I draw on quantitative analyses concerned with empirical research in the area of HRM, which analyse the totalities in a way I approve and not (only) the individual analyses or their individual statements.However, these analyses of publications refer almost exclusively to English-language journals published outside of the German-speaking area (especially Keegan & Boselie, 2006;Wasti, Poell, & Çakar, 2008).There is little evidence, though, that circumstances in Englishlanguage research are fundamentally different to Germany, but this is an assumption to be discussed.Second, I rely on those empirical studies published in the last ten years in particular in the "Zeitschrift für Personalforschung"11 (German Journal of Research in HRM).And third, I take into account empirically-oriented papers or presentations given at conferences of the "Kommission Personalwesen" 12 from 1999 to 2009.

Concept of empirical HRM research
The term "empirical HRM research" includes the body of statements generated in the science system of German-and English-language research, which is based on systematically collected data.In doing so, it is irrelevant whether data are collected using qualitative or quantitative methods, and whether a large or small number of cases are "The Zeitschrift für Personalforschung (ZfP) is one of the oldest HR journals worldwide.It was founded in 1987 … and currently is the only HR journal outside the US and the UK covered by the SSCI" (http://www.zfp-personalforschung.de).
12 "The commission "Personnel Management" within the "German Academic Association for Business Research" was founded in 1973.The commission deals with questions of describing, explaining and designing human work in organizations, especially in firms.The more than 100 members of the commission represent a broad range of theoretical perspectives: Economical, sociological and psychological as well as political, pedagogical or ergonomical approaches.To enhance research and communication between theory and practice, workshops are organized every year" (http://pers.vhbonline.org/index.php?id=197).analyzed. 13Here, I am taking German-and English-language research together, because this research constitutes the relevant discourse.In the analysis, I will sometimes differentiate the country-specific contexts.Regarding the content, I will distinguish HRM research from other veins of research through the fact that it by definition concerns HR management.I understand research into HR management to be abstract analyses of questions of the reproduction and use of an adequate company working capacity.Institutionally, I refer to research which is undertaken within business and management studies, especially in the specialisations of HRM, organisation and management.HRM research results produced in other scientific communities -such as sociology, psychology, medicine, engineering, etc. -are treated here only in passing.

Ideological effect and the meaning of the social context
In the following analyses, I do not examine the effects of specific statements.This is necessary, but cannot be done here.Rather the characteristics of the collective body of empirical statements (and individual statements) and how they come to be are in the foreground.Nonetheless, reasons will be provided as to why certain effects are to be expected.Characteristics of statements (e.g.being devoid of meaning) unfold their effect neither by themselves nor in a social vacuum, but rather are received and used by individuals and groups in social contexts.Statements from empirical research, e.g. about the effects of employee participation, trickle down through the media (such as the press, scientific and practical or political publications, radio, television, the new media) into the cognitive structures and constitute consciousness, as they deliver the material for the construction of an image.In a conflict between capital and labour, such images and elements of the images are used by certain actors as arguments in order to directly (in a company) or indirectly (e.g. via the law) influence the exchange relations within companies.Not every actor can use empirical results to the same extent; in addition to the corresponding prerequisites regarding qualifications, it especially depends on access to the areas of decision making.This means that, in order to estimate ideological effects, one must take into account not only the statements and their characteristics, but also the social context, which means that one must make certain assumptions, particularly about its power structure.This is not only about material but also about symbolic power (Bourdieu, 1992, p. 153), which reverts to the accepted value-knowledge systems and is communicated via language: conclusions of empirical research are generated against the backdrop of particular thought systems, where they encounter more or less fertile ground, but also reinforce them.Concerning the power structure, I make a simplifying supposition; I act on an ideal-typical assumption of an asymmetric distribution of material and immaterial, symbolic power in favor of the employer.
Regarding the effects in a business context, the question is to what extent the core bodies (management in particular) in companies take notice of empirical research or use it in one form or another to push their interests.I argue here that scientific 13 The focus does not need to be on the description and explanation.Many scientists consider empirical research as a means to and end -it is done in order to influence business practice.On this the (few) trade union-oriented researchers in the field of business administration agree with those who would rather serve capital interests.findings are mainly used for a symbolic legitimisation of decisions in the company (Pfeffer, 1981).That information considered to be necessary for the process of decisionmaking is often supplied by watching other businesses that are comparable and perceived to be successful -in contrast to justifying the decision to third parties (Nienhüser, 1998), where the assistance of business consultants plays a role (Kieser, 2006).If empirical findings in general were to offer a kind of legitimisation-quarry where companies could help themselves as needs arise, then these could be employed as a means in the manifest or latent conflict with stakeholders.This would be particularly easy for management if the totality of findings were already tailored to such a demand by integrating more the interests of managers and capital owners than those of other stakeholder groups beginning with the set-up of the study.At the same time this would make it more difficult for other stakeholder groups to utilise the empirical findings (in their totality).
In summary: statements of empirical HRM research have a particular ideological potential when they offer a power advantage in conflicts irrespective of their truthcontent.

Analysis
The analysis starts first of all with the question of which topics are higher priority in empirical HRM research, which are lower priority and which ones are not addressed at all.This includes what the guiding questions in HRM research are and which questions are omitted, which dependent and independent variables are used and how these are interpreted with regard to content (conceptualised).
Second we need to ask which interests and values matter in HRM research and have a selective effect.Is HRM research specifically characterised by a one-sided capital or employee orientation or is there a broad variety of values so that we can speak of a pluralistic (or at least a dualistic, i.e. oriented in the values of capital and labour) value landscape?
Third, I raise the question of which "human resources" are the object of HRM research.Is it mainly the "upper echelons", (senior) executives and other organisational elites such as academically qualified professionals or is research also interested in the broad masses of employees?What should be clear is that a concentration on the second group does not necessarily imply "employee orientation" in terms of the aim of realising employee interests.Interest orientation is the object of the previous point and has to be analytically separated from the question about the object of the analysis.
Fourth, there is potential for selectivity and bias in the fact that access to certain companies is systematically more likely than access to others.There is the possibility that companies with conditions that are more favourable for the employees or "presentable" are more likely to grant researchers access.
Fifth, access to companies might be granted and perhaps we are even asking the "right" questions.But who is answering those questions?Is it also the elites of the organisation or do we get information from other groups as well?
There are two other points which I will not cover despite their importance.For one I will largely leave out the question about those theories that guide empirical re-search.Of course this is an important aspect which I will cover fairly briefly, not only for reasons of space, but mostly because the connections between theories and valuations, images of man, ideologies, etc. are also explored in other studies (see e.g.Budd & Bhave, 2010), while the ideology content has hardly been analysed empirically.I will also defer another point as it deserves a more in-depth coverage as well and which is furthermore not directly linked to selectivity in empirical research but to the practical reference of the empirical results: often practical "conclusions" are drawn based on empirical results or "design recommendations" given etc.Logically there is, however, no implication for practical matters from empirical results (alone).The derivation of practical statements does not necessarily require values, but at least hypothetical targets.Practical conclusions, however, would be crypto-normatively charged if theyintentionally or unintentionally -were to appear as the logical consequence of the empirical findings, although they would in fact only apply under the proposition of specific (implicitly hypothetical) objectives.As this issue would need to be covered in more detail, I will defer it here.

Selectivity in research topics and questions
The overall results of empirical research -i.e. the image that results from the basis of all findings put together -depend on the topics that were selected and the questions used to approach the respective topic (see also Habermas, 2001).Those topics and questions that are not addressed or asked cannot contribute to the image.Topics and questions of HRM research can be presented abstractly in a diagram like this.Let us look first at the chain of cause and effect along the darker (bold) arrows.With this marking I want to emphasise that these connections are studied more intensely than others.There is first of all the focus on the question of how HR practices affect the composition of staff on the one hand and their behaviour on the other.Both the individual level as well as the social aggregation "workforce" can be the subject of analysis.Normally HRM research is not interested in the effects on individuals but in the effects on staff in general.The analysis of effects on individuals is rather a means to an end.Second, there is the question of how HR practices (via the impacts on behaviour) influence the behaviour of businesses.Third -I am returning to the beginning of the cause-effect chain -one could ask for the determinants of HR practices and clarify whether and why businesses exhibit differences in this respect.Fourth, HRM research could study the consequences of HR practices beyond the corporate context on society (e.g.whether and how corporate social policy replaces that of the state, the impact of employee behaviour beyond the organisation (e.g. the effects of a longer daily presence at work on family life).Fifth, those reasons for employee attributes or structures and behaviour lying outside HR practices (e.g.demographic changes or "incidents" which happen outside the company and change individual behaviour) can be the subject of HRM research.Which of these connections and blocks of variables does HRM research actually look at, which are studied less, and how are variables conceptualised and interpreted?

Focus on performance
In empirical studies work performance directly or indirectly plays the central role.So the focus is not on any kind of employee attributes or behaviour, but it is rather the reference to work performance that influences the necessarily selective conceptualisation.Performance is either a variable that is empirically collected in a study and incorporated in the analyses.Or if it is not the subject of an empirical analysis, it is nevertheless very often used as an argument to justify the necessity of a study.Performance can comprise performance-related employee behaviours or an individual performance result or the possibly affected (economic) performance of a firm.The analysis by Wasti et al. (2008) offers a first indication of the dominant position of performance as the subject of studies in HRM research.It is based on the content analysis of 268 articles on the topic of "Human Resource Development" (HRD) which were published in 14 journals in Europe (102 articles) and the US (166 articles) in the period 1990 to 2003.It becomes evident that performance is the central aim of the studies.
Moreover, there is a limitation to economic parameters in the conceptualisation of firm performance.Boselie, Dietz, & Boon (2005) 14 have analysed a total of 104 English-language articles regarding the connection between HRM and performance.Regarding the indicators for firm performance, the focus is on finance-oriented parameters which are important for shareholders such as productivity, profit, turnover, etc. (Boselie, Dietz, & Boon, 2005).In contrast, performance parameters referring to employee objectives such as job security, good remuneration or work-life balance are hardly found in the empirical studies that (Boselie et al., 2005) analysed.Other aims of target groups such as customers or social objectives such as environmental protection are hardly addressed.Of course there are publications, e.g. on the topic of "Corporate Social Responsibility" (Preuss, Haunschild, & Matten, 2009); empirical studies, however, are scarce.
It is particularly noticeable that in the introductions of publications with an empirical focus, the meaningfulness and necessity to study object x is almost ritually justified with the argument that x ultimately contributed to an improvement in competi-14 With the following percentage values I mainly refer to the table by Boselie, Dietz, & Boon (2005) which was not published in the article in 2005 but only on the internet (http://www.tilburguniversity.nl/faculties/fsw/departments/HRS/research/themes/HRperformance/HRM_and_Performance_list_July_2005.pdf).Nearly all calculations of percentage were done by myself.tiveness, a statement which also implies that x increases work performance.The following quote is highlighted as the motto and precedes a study on the "social skills" of insurance employees (Graf, 2002): "Nowadays no company that wants to survive on the global market stands a chance without the social skills of its employees" (Faix & Laier, 1996, p. 41, quoted by Graf, 2002, p. 377).Another study puts it slightly more intricately: "Owing to the pressure for change in globalised market constellations and accelerated technology development, great significance is attached to innovation competency for the sustainable competitiveness of businesses…" (Kriegesmann, Kerka, & Kley, 2006, p. 142).Another example: in an essay about measuring the "participatory culture" in organisations (Martins et al., 2008) state in their introduction regarding the motivation: "We suppose that one way to induce and maintain high work motivation and a positive attitude towards the work and the organization is by employee participation" (Martins et al., 2008, p. 196).However, it might at least be conceivable to justify the measurement of the participatory culture not with potential motivational effects but with the aim of introducing industrial democracy in companies.
While the determinants of performance -particularly those controlled and influenced by management -are frequently studied, only few analyses address the effects of performance delivery and the respective "incentive systems" for employees (an exception is e.g.Breisig et al., 2010).Certainly there are surveys on stress and the like but then these are usually not justified with employee interests but with the negative impacts on future performance, absenteeism, etc. (see e.g.Chang, 2008). 15

Restriction to the firm level
Let us look further at the connection between the spheres inside and outside the workplace.There we can also note a dominance of the performance perspective.Questions about the impacts of social issues have been asked more frequently in recent years, e.g. about the effects of demographic developments on the availability of workers (Bieling, 2010;Basler, 2009;critically Ebert & Kistler, 2007).However, the effects of HR policies beyond the firm's boundary are studied less frequently, e.g. the effects of corporate working time arrangements on family members and when they were studied then also with emphasis on the "competitive advantage" (Badura, 2004).Apparently such questions are only included if, for example, the family situation affects the availability and motivation of employees; then it is topics such as work-life balance (Kaiser, 2011) and the integration of the family in questions of posting expatriates and repatriating them (Spieß & Stroppa, 2010).Questions that do not refer directly to the usability of a company's manpower are in turn less frequently picked up on, e.g. the effects of HR management on health after retirement, on mortality, costs for the family or for the health system (Pfeffer, 2010).That individual businesses have little interest in the social consequences of their actions (as long as they do not have to internalise the costs), is consistent with the logic of capital valorisation.That empirical HRM research collectively imposes a similar constriction of its perspective on itself does by no means correspond with a pluralistic view (not even in a dualistic version).
Consideration should be given as to whether these or other propositions about HRM research also apply to work and organisational psychology.

Concentration on high performance work systems
One-sidedness can also be noted in the surveyed HR practices which, in turn, are supposed to affect performance by influencing the psychological condition and behaviour.A study by (Boselie et al., 2005) shows that out of the 633 mentions of HR practices the largest proportion with 13% is "training and development", followed by performance-related pay (11%), performance appraisal and personnel selection procedures (8% each).Less frequent and of lower significance in research are practices like indirect participation (via trade unions, works council, etc.) (below 2%) or familyfriendly HR policies (1%); staff reduction is also hardly addressed (Boselie et al., 2005).The way I see it there is a similar pattern in German-language HRM research: it is about the means and result of human capital reproduction, not about the analysis of the conditions under which decisions about them are taken (e.g. more or less participatory).
Empirical HRM research pays particular attention to a certain pattern of practices that is often referred to as "High Performance Work System" (HPWS 16 ).This pattern is ideal-typically characterised by the following attributes: job security, selectivity in personnel selection and preference of internal as opposed to external recruitment, participation of employees in decisions and extensively informing them about the economic situation of the company, small differences in wages and status, profit sharing and in-company training (cf.Pfeffer, 1994, pp. 30-65). 17When following Pfeffer and also the empirical studies on HPWS (cf.representatively the influential study by Huselid, 1995, in the overview also Huselid & Becker, 2011), the image of a "one best way" is created -that only HPWS can lead to success, that successful companies prefer this strategy and others (should) join them.This way the latent conflict between capital and labour is eased as it can be claimed that employers and employees are equally interested in such HR strategies or at least should be.
It is easy to lose sight of the fact that an interest in capital valorisation can make it practical not to follow an HPWS HR strategy, particularly with job tasks that are simple, can be easily divided into small sub-tasks, and require little experience.Because of this ignorance an image of working conditions and dealing with employees is created which suggests that certain HR practices are in widespread use and generally aim at job satisfaction and good working conditions.This image may apply to highly qualified groups of employees or those that are difficult to substitute but not for all employees and all firms.
Huselid and Becker's argumentation is very revealing (Huselid & Becker, 2011).Both are frequently quoted and therefore influential authors, who themselves have contributed a lot to HPWS research.I therefore allow myself three longer quotations: 16 An early overview of research and origin of the term can be found in Cappelli & Neumark (2001; see also Boxall & Macky, 2009).17 Relevant workforce strategies, however, have not been allocated only the term "HPWS".There are in fact numerous terms for quite similar strategies.HPWS, for example, shows considerable parallels to the system of the internal labour market described by Lutz (1987).
"Since the early 1990s, well over 300 articles on HR strategy have been published in the academic literature (and many more in the popular press), drawing from fields as diverse as economics, HR management, industrial relations, sociology, and strategy.Taken as a whole, the primary conclusions of this line of research have been that the financial returns to investments in high-performance work systems (HPWS) are both economically and statistically significant …" (Huselid & Becker, 2011, p. 422).
A few paragraphs later doubts are raised about the validity of the statement about the relationship between HPWS and financial success: "After nearly two decades of empirical research support for a positive HPWSfirm performance relationship, why do we still observe such substantial differences in HR management "quality" across firms …? Said differently, if the financial returns to HPWS are so substantial, why aren't more firms using them?" (Huselid & Becker, 2011, p. 423, I have omitted the authors' references, which mostly refer to their own works).
They come to the (extremely normatively distorted) conclusion that apparently there are different HRM strategies as there were different jobs ("strategic" vs. "nonstrategic jobs") and "top-talented" employees were not needed in all firms and for all jobs: "The essence of our argument is that some jobs are more valuable (strategic) than others, and they should be managed accordingly.This means that disproportionate investments must be made in strategic roles, ensuring that the organization places top talent in these positions.For less value-added roles, the organization must make informed decisions about the right level of talent that it needs in these roles, and the answer might well be that top talent is not needed in many jobs.Attracting, selecting, developing, and retaining world-class talent represents a very significant investment for most organizations, and the harsh reality is that most organizations simply do not have the time or resources to do this for all organizational roles.This means that a decision about where and how to invest must be made, which we believe should focus predominately on strategic jobs" (Huselid & Becker, 2011, p. 424).
It is admirable how the authors present the overcoming of their own ignorance of reality (in this case the existence of HRM strategies which, to put it briefly, are characterised by short-term employee retention, recruiting from the external labour market and relatively bad working conditions) as scientific progress.Nonetheless they are right, although a recourse to theoretical approaches could have provided the same insight earlier: several theories dealing with HR strategies distinguish between completely different strategies.They state that HPWS do not occur consistently and other strategies are by no means anomalies.They thereby draw a picture which is not a priori positive for the realisation of employee interests, as the mainstream HPWS discourse suggests.Transaction Cost Theory with its base in microeconomics (Williamson, 1984) as well as the Labour Process Theory, which partly reverts to Marxist approaches (cf.representatively Thompson & Newsome, 2004) comes with entirely different reasons to very similar statements about the differentiation of HR strategies and the different ways of dealing with employee segments: Transaction Cost Theory tells us that in a situation where no firm-specific human capital is needed and where the control of work performance is unproblematic, an extensive HR management, a respectful treatment of human capital, i.e. employees, is inefficient from an economic viewpoint and can therefore hardly be expected empirically.Labour Process Theory makes similar predictions.Furthermore, in contrast to the HRM research mainstream, it negatively frames HPWS; it emphasises the normative integration as particularly clever because hidden control techniques are employed which superficially correspond with employee interests.A HPWS is a solution to solve the control problem when other control strategies don't work, the intention is the socialisation, the self-control, of employees.In contrast to mainstream research Labour Process theorists point out the downside of "responsible autonomy": self-exploitation, destruction of solidarity and so on.Both theories would in any case explain why we cannot find the "High Performance Work Systems" in all firms.Yet empirical HRM research has been mainly interested in this very system type and neglects the others.
So Huselid and Becker come to see what they have not seen so far.One is tempted to say: a late insight but better than the ignorance that persisted for a long time.Immediately they limit the potential reach of the findings by normatively distorting their descriptive statements: with regard to the usage of resources they recommend "focus[ing] predominantly on strategic jobs" (Huselid & Becker, 2011, p. 424).Now non-strategic jobs are placed into the centre of attention but the recommendation to management is to largely ignore them.
To sum up: When the larger part of empirical HRM research refers only to those parameters in the shareholder interest, when research is guided by a constricted performance construct and a picture of HR management is drawn that is too positive compared to reality because of the focus on High Performance Work Systems (even though perhaps unintentionally), then this is a distorted, one-sided social construction of the perception of reality.The focus on those HR strategies which are called HPWS here and overlooking other forms of "dealing" with employees casts a positive light on reality and makes changes for the benefit of employees in non-HPWS less likely.

Plurality through dissent-oriented research or monistic consensus orientation?
Now there are also researchers who do not embrace business goals normatively and who are particularly interested in the "dark side" of HRM.In English-speaking countries a new field of management research is being established which calls itself "critical": "Critical Management Studies", which by now has its own "division" within the Academy of Management."The Critical Management Studies Division is a forum… for the expression of views critical of unethical management practices and exploitative social order.Our premise is that structural features of contemporary society, such as the profit imperative, patriarchy, racial inequality, and ecological irresponsibility often turn organizations into instruments of domination and exploitation" (http://group.aomonline.org/cms/About.htm).Critical management studies (CMS) aims at a radical change in management practices, as this is supposedly one way to reduce social injustices and ecological destruction reproduced by management (Forbes, Willmott, & Adler, 2007).CMS can mainly be found in organisational research and is less prevalent in the field of HRM research.It would be going too far to consider the mere existence of a research field that regards itself as being critical to be an indication of a plurality of research.However, the question arises how prevalent and influential this research is.Journal analyses by (Keegan & Boselie, 2006) and (Wasti et al., 2008) indicate that only a small fraction of the articles in total can be called "critical": (Keegan & Boselie, 2006) have analysed 1,674 articles -though not only empirical ones -in nine scientific, English-language journals from Europe and the US over the period of 1995 to 2000 with regard to their "dissent orientation".The authors (referring to Alvesson & Deetz, 1996) base the distinction between consensus-and dissent-orientation on the fact whether scientists support dominant discourses or rather seek to disrupt them (Keegan & Boselie, 2006, p. 1495). 18The central finding is that on the whole dissent-oriented studies are rare.In HRM journals the authors of the study rated a share of 1.2% as dissent-oriented.In the second group, the organisation journals, only a small share addresses HRM issues (34 out of 983), however about 35% of these were rated as dissent-oriented.In summary we can say that English-language HRM publications are strongly consensus-oriented.The findings of (Wasti et al., 2008) point in a similar direction.The share of essays where a "left-wing ideology" can be identified was 8% at the most, depending on the journal category.This share was lowest in US-american journals .Wasti et al. labelled those articles as "left-wing" that dealt with "capitalist hegemony, exploitation of low educated workers, lack of workplace democracy" (Wasti et al., 2008(Wasti et al., , p. 2160)).They furthermore observe that dysfunctional organisational processes like "disenfranchising, ostracizing, harassment, bullying" attract little interest.The share of the 268 articles addressing those questions is well below 3%.When publications in English-language journals and areas are characterised by consensus orientation, what does it look like in the German-speaking area?A review of conference topics and articles in the Zeitschrift für Personalforschung (ZfP, German Journal of Human Resource Research) yields similar results to the findings of the aforementioned publication analyses.In the German-speaking area there is presumably an even stronger consensus orientation (though there are exceptions, see website www.http://www.kritische-organisationsforschung.de), as in Great Britain a larger part of those doing research in HRM do not have a degree in business administration but in sociology (cf.also Muller-Camen & Salzgeber, 2005).Selection and socialisation in business administration lead more to a consensus orientation of the scientific community than those disciplines that are more closely affiliated with the social sciences.Furthermore, empirical research in CMS is under the suspicion of positivism (Alvesson & Deetz, 2000) -at least its quantitative types -and is less widespread, so that there are probably few from this field to be found in the group of empirical researchers.

18
According to Alvesson & Deetz (1996, p. 197) dissent orientation is first of all characterised by mistrust (towards management) and second by the perception that dominant points of view (and theories) are always those of the ones in control and in their interest.Third, dissent orientation includes the assumption that order is to be seen as dominance and oppression of conflict, and that conflicts about the form of order are the norm; science and also theories are considered political means of change.
Research that is critical towards the objectives of capital and management does not equal Critical Management Studies.However, other critical perspectives are hardly to be found (anymore).In the Federal Republic of Germany a field of research that claimed to be specifically employee-oriented (Bosch, 1980) had been of a certain significance since the mid-70s, but disappeared at the beginning of the 90s at the latest.They had in any case had little backing from the field of business administration.Here an (indirect) indication for the decreasing significance of an employee-oriented science and correspondingly oriented empirical research can be seen in the changes in the discussion and presentation of "Arbeitsorientierte Einzelwirtschaftslehre" (AOEWL) (labour-oriented business administration) (Projektgruppe im WSI, 1974) (project group at the Institute of Economic and Social Research) in the course of time.The AOEWL was developed as a counter concept to capital-oriented business administration and was discussed critically in the discipline of business administration.Today the concept has been all but forgotten: while Schanz (1997) presented the AOEWL alongside other science programmes in previous editions of an introductory business studies textbook, that section is missing in later editions (for more detail, see Freimann, 2006).
All in all there is no pluralism of interests in empirical HRM research.On the contrary, as in the general discussion, studies indicate that there is a monism of interests.

Which persons are we dealing with?
Empirical HRM research is mainly about the organisational elite or highly qualified employees.There is good evidence for that in the English-language literature.According to the aforementioned publication analysis of Keegan/Boselie, 37% of the arcticles deal explicitly with the core workforce (Keegan & Boselie, 2006, p. 1501).Relatively few publications in scientific journals refer to "diversity", i.e. to groups such as women, homosexuals, non-whites, disabled people, marginalised (sub-)cultures or poor people (here I am using the list of Wasti et al., 2008Wasti et al., , p. 2160): depending on the disciplinary origin of such articles the proportion is between 3.3% and 12.7%) (Wasti et al., 2008(Wasti et al., , p. 2160)).Even more concise is the finding that of the 639 essays published between 1997 and 2007 in the "Academy of Management Journal", a mere 7% deal with "´lower-echelon employees´ (i.e., non-managerial/nonprofessional staff" (Bamberger & Pratt 2010, p. 665).
In order to assess reasonably whether there is a similar tendency in German HRM research, I have analysed the conference programmes of workshops that were hosted by the commission "Personnel Management" in the years 1999 to 2009.I gathered the number of those presentations or papers given at the conferences where specific groups of employees or people were the object of the analysis (whereby only the titles were considered for counting purposes).There may therefore well be a bias because of those cases where the employee group was not mentioned in the title.Moreover I have integrated not only empirical analyses but all presentations in the following Table 1.Of the 65 presentations where groups of (employed) persons are mentioned, 54% explicitly refer to managers, experts and freelancers, where managers account for the biggest group with a total of 34%.26% of mentions are allotted to employees in the narrow sense, although there are also some groups comprised in this category that we can for good reasons still count as experts or at least highly qualified (e.g.employees in commercial law firms or the so-called "equal opportunity actors").In any case it is only a small part of the conference contributions dealing with the largest part of staff.
When looking only at the 25 presentations with the word "empirical" in the title, six of them are explicitly about managers, executive staff and experts and one each is about works councillors and temporary workers.
The results indicate that a large part of HRM research in total, but also of empirical research deals mainly with managers and experts or other highly qualified staff.The empirical findings move the working conditions of the organisational elite to the foreground: they form an image composed of single studies, which lets the conditions of other employee groups be overlooked or forgotten.
In the following section we will see that this group is not only the focal point of the research interest but also controls the access to information.

Controlling the access to information -companies and interviewees
Further bias can be created through information sources used in data collections.This applies on the one hand to the access to profit organisations.Which companies do researchers want to gain access to and where do they get it?On the other hand selectivity can play a role in choosing the key informants in the company.
At first there is again the situation in English-language publications: a very large proportion of the studies evaluated by Keegan & Boselie (32% after all) uses data from "large multinational companies" (Keegan & Boselie, 2006, p. 1501).Drawing on the issues of the Zeitschrift für Personalforschung (Journal of Human Resource Research) from the past ten years, this impression is confirmed for the German-speaking area.The reason might lie in the interest of HRM researchers but also in the possibility to gain access.
When it comes to HRM, working conditions and the impacts on employees, companies will be more willing to participate if they expect favourable results for themselves.Fischer (Fischer, 1989) points out that job satisfaction surveys were mainly carried out in those companies where there was a "minimum consensus" between staff and management.There may be the "risk of producing data suspected to be artefacts as the lack of motivation and orientation of the interviewees encourages trivial remarks borne by conventionality.The contradictory results arising from numerous studies of job satisfaction research may to a large extent have been induced by these facts" (Fischer, 1991, p. 195).
It can be argued that there are now better data sets available (which go beyond the theme of satisfaction surveys), e.g. the Socio-Economic Panel (Wagner, Frick, & Schupp, 2007) or the IAB Establishment Panel of the Institute for Employment Research (IAB) (cf.Janik & Kohaut, 2009), the European Survey of Working Conditions (Parent-Thirion, Macías, Hurley, & Vermeylen, 2007) or the Works Council Survey of the Institute of Economic and Social Research (WSI) (Schäfer, 2008), which are in parts freely accessible for scientific purposes.There is no doubt that these are good surveys, large scale, with thoroughly constructed and tested samples, partly designed as a panel and so on.But even with good data sets, such as the IAB Establishment Panel, biases due to interests cannot be eliminated in the survey.One critical point is the panel attrition rate.As employers' associations are needed to call on the businesses to participate in the IAB survey (Janik, 2009, p. 18), employers and their associations control at least partly an important resource: the willingness to participate.Employers' associations can -and will presumably -use this power base to influence the compilation and formulation of questions as well as their evaluation (by the IAB), after all they are represented in the Board of Governors of the Federal Employment Agency.It can also not be fully ruled out that researchers practice self-censorship in anticipation of employers possibly intervening in conflict-laden questions and thus certain research projects or data analyses are not carried out in the first place. 19In addition bias occurs because of refusal to respond: large firms refuse more often than small firms to participate again in the IAB establishment panel (Janik, 2009, pp. 17f.).Refusal is higher in those companies that did not specify their wage bill in previous surveys.We can 19 For this proposition I can cite no reference other than confidential information.management revue, 22(4), 367-393 DOI 10.1688/1861-9908_mrev_2011_04_Nienhueser 387 therefore assume that low-wage businesses, which are also unfavourable for employees in other working conditions, participate less frequently 20 and that this positively exaggerates the view of the "average" business situation based on the findings of the IAB panel.
Findings of a study by Harrison & Cycyota (2006) indicate that the response rate might be influenced by a more or less strong support from employers' associations.They analysed 231 studies where managers were interviewed with regard to the reasons determining the differences in the response rate.For example, there is evidence that the response rate is higher when there is a positive reference to a network where the respective company or person surveyed is a member.It is also conducive for the response behaviour when the survey is supported financially or by other means by an employers' association, when researchers can refer to names of well-known managers, who have already participated in the survey or when the interviewees are personally known.In the multivariate analysis the network effect turns out to be the strongest (beta = 0.32) (Harrison & Cycyota, 2006, p. 140).Conversely this means that the probability of response is lower when employers' associations do not support a study or when researchers themselves are not well-connected with businesses.This reduces the probability of questions, access possibilities and therefore findings which might not cast a positive light on businesses.
Who is interviewed in the companies?This is not the same as asking which groups of persons are the research objects.But when we assume that those who are the subject of research respond themselves, 21 then the results outlined above indicate that for the most part managers, members of the executive committee, senior staff and experts are surveyed.According to the study of Boselie et al. (2005) it is primarily managers and HR managers respectively who are surveyed.49% of all mentions in the articles are allotted to these groups plus those managers in surveys where more than one person was interviewed.Individual employees make up 12% of the mentions, employee representatives 4%.The analysis of 164 journal articles relevant to work, personnel, organisational or vocational psychology by Solga and Blickle of the years 2004 and 2005 shows likewise that managers are strongly overrepresented in the samples; and studies with production workers are rather rare (Solga & Blickle, 2006).
Altogether it can be assumed that there is a bias because of selectivity in the access to collecting the data, in the response rate, and also in the selection of the interviewees, which are rather to the benefit of the employers: the conditions in the companies seem better than they are, because "good" companies are more likely to participate and because the organisational elite presumably tends to have a more positive 20 The economic situation (changes in employment) and industry classification have no significant effect on non-participation (Janik, 2009, pp. 17f.).

21
It is not necessarily the case that those who the statements are about are the ones who are interviewed.Krystek, Becherer, and Deichelmann (1995) interviewed HR managers about the number of employees who had "mentally resigned".This would not be a problem if this statement were considered to be the managers' assessment of staff's condition and not be rated as the true description of the state of mental resignations without further examination.view of labour relations, which is not least influenced by their own conditions.And finally they look better because surveys integrated in the social networks of companies tend to allow only less critical questions.

Conclusion
My aim was to analyse which image of the exchange relations between capital and labour is drawn by empirical HRM research.It is generally possible, but practically very difficult to look at the totality of the individual studies as parts of a puzzle, to put them together and to describe this inductively generated image (and also to analyse its effects).I have therefore chosen to analyse the selectivity in the generation of empirical research results.The assumption is that relevant selectivities lead to an incorrect overall picture, that ratings are excluded from discussions and that generally ideological potential is developed.First let us recap: (1) HRM research mainly follows employer objectives; it primarily analyses, directly or indirectly, performance-related variables.(2) The surveyed HR practices have a clear focus on "High Performance Works Systems" (systems with comparatively favourable attributes for the employees).(3) Critical analyses that either distance themselves from profit-oriented business goals or employer objectives or which are neutral towards them are rather rare, specifically employee-oriented research is hardly to be found anymore.Studies analogous to Critical Management Studies are practically nonexistent in the German-speaking area.All in all there is nothing that indicates that the political spectrum in German-language empirical HRM research is characterised by plurality, as we find in Great Britain, even though it is not particularly strong there.(4) It is mostly the organisational elite that is surveyed as personnel.(5) This group also controls the access to information and provides information about the situation in companies.
What matters is not only whose objectives are pursued and what is said, but whose objectives are not considered and what is not said by that.If, for example, the effects of HR practices were assessed with regard to the achievement of employee objectives, one might find that their objectives were realised only to a small extentcompared to those of the capital side.Additionally it might likely show that certain practices (such as some performance-related pay systems, cf.Breisig et al., 2010) even have a negative effect.Surveys based on employee objectives might also bring alternatives to the current HR practices more to the fore.Relevant objectives, questions and answers, however, are not employed.Yet this selectivity is hardly considered problematic in HRM research.One reason might be the assumption that HRM researchers share the assumption that it serves employee interests best if those HR practices are applied that maximise economic success.Apparently it is suggested that there is an objective interest of employees as well as investors and managers in the economic success of the company, which employees might not necessarily be able to realise subjectively, unlike management.According to this view, management acts in the employees' interest and contributes to the common good, even if some practices might not comply with the primary, i.e. subjective interests of employees.
The individual selectivities must not be viewed in isolation as they interact and reinforce each other.Let us now add the following points to the focus on capital inter-ests -that empirical research focuses more on "High Performance Work Systems" and that practices which are considered positive have a greater chance of being studied and therefore mould the image.This leads to a positive view; low-wage labour relations, bad working conditions, few opportunities for participation, avoidance of trade unions and collective agreements etc. disappear as elements of the picture.Then, when "key witnesses" of positive HR practices can be produced, the relations appear to be even better than they really are.It is these key witnesses and their reports that shape the results of empirical HRM research: employee elites are more frequently interviewed than other employee groups; it is about their conditions, the HR practices concerning them, that we find most reports.HR scientists will hardly deny that there are employees who are exposed to bad HR practices (and working conditions).Why does the non-observance of these strategies, practices and labour conditions not generate a cognitive dissonance pushing for more insight or even change?One reason might be that many HR scientists (doing empirical research) attach a kind of utopia to it -they imply a progress model of change that looks like this: (i) Science knows what good HR practice has to look like.(ii) There is one model of good HR policy and that is called High Performance Work Systems (or something like that).Good means: good for employer and employee objectives.(iii) In practice some parts of this model might still be unknown, but it will be implemented via clarification from science or through market pressure.Market competition plays a major role in this concept.It enforces the common good.It is suggested that the "War for Talent" (von der Oelsnitz et al., 2007) will be lost if a company does not practise "modern" HR management (meaning one which is good for all). 22And who wants to lose "wars"?So an image is generated that the conditions change all by themselves because of market's higher reason that forces unreasonable businessmen or managers to do good deeds for all staff -or eliminates them or their companies from the market.Those who want to recruit and retain qualified and motivated staff who like to perform over their limit and over the normal level (showing "Organizational Citizenship Behaviour"), will need to give in to these constraints.However, it is easily overlooked that these assumptions might well apply to highly qualified employees and corresponding workplaces, but not for the entirety of all types of work and employees.Yet the picture that empirical research collectively draws is not fully naïve: it contains some dark spots which are apparently seen as difficulties and which always come up when there is change (which is generally welcome or was even initiated): we analyse work-life balance problems caused by the demands of flexibility, difficulties with commitment in the integration of newly selfemployed, repatriation problems with highly qualified staff returning from abroad and so on.All these questions focus mainly on highly qualified employees and on improving the conditions of this group.
That at least some specific industries (e.g.basic services, meat-processing, construction and so on) can stay competitive with few highly qualified workers, via low 22 The restriction to highly qualifieds is evident."Der Talente-Krieg: Personalstrategie und Bildung im globalen Kampf um Hochqualifizierte" (The war for talent: HR strategy and education in the global battle for highly qualifieds) is the full title of the book by von der Oelsnitz, Stein, & Hahmann (2007).wages etc., is simply ignored (or sometimes even normatively approved like Huselid & Becker, 2011).
The underlying analytical framework outlined in the introduction also includes the statement that empirical findings become effective in terms of reality construction particularly when they are in line with existing world views.The effect would be especially strong if the picture produced by empirical research were in harmony with a general view of HR management and corporate exchange relations.This is exactly the case.I contend that empirical research is reproducing an existing unitaristic idea (cf.particularly Legge, 2008;Budd & Bhave, 2010) of the exchange between capital and labour.While other perspectives -a market-oriented, a pluralistic and a radical perspective on the exchange relation (Budd & Bhave, 2010) -mainly emphasise the clash of interests, the unitaristic exchange relation is characterised by the assumption that there is an overlap of interest and by the idea of a long-term partnership between employer and employee.In this image HR management has the function of evening out the (few) clashes of interests but also of identifying employee interests and ways to satisfy them.This idea of the role of management is like that of the head of a family who acts in the well-understood interest of his wards.It is part of this picture that employee interests, in their own interest, have to lag behind those of management.
If my thesis of ideology production applies, we should be able to note that empirical research produces or enforces that kind of image that best reflects employer interests.This is the unitaristic image; it depicts the employer positively, just as it does the role of HR management.Conflicts are considered resolvable, and the existence of state regulation, laws or trade unions is not ignored, but has at best a marginal role for the events in the company.Deviations from an employee-friendly HR strategy are seen as relatively rare anomalies which are about to disappear.
So empirical HRM research in its entirety produces an ideological knowledge base that distorts perception and corresponds with a unitaristic image, which it confirms and enforces or at least does not question.This raises the question of why such a collective one-sidedness of HRM research came about -but that is a different topic.

Figure
Figure 2: Causes and effects of HR practices Determinants of Human Resource practices