Autonomy in Journalism: How It Is Related to Attitudes and Behavior of Media Professionals


Armin Scholl and Siegfried Weischenberg

[WJMCR 2:4 September 1999]


Abstract|Introduction|Theoretical framework and hypotheses


Autonomy is a main characteristic of professions. Social system theory suggests observing journalism in terms of self-referentiality and external referentiality. In our study “Journalism in Germany,” we could identify a particular self-referential group of journalists, which differed from the rest of the sample regarding role perception, unusual reporting practices and assessment of press-releases. Data provided further evidence for a more complex and adequate perspective on journalists’ attitudes and behavior.


The history of communicator analysis indicates that most of the research in this field has focused either on professional roles or on communication processes within newsrooms.1 While the professionalism approach has primarily been concerned with role orientations in journalism2, the gatekeeper approach has observed and reconstructed editorial decisions for the selection of news.

The gap between these two more or less unconnected research directions often has led to one-sided reduction of media production to either aspects of journalists’ attitudes or processes in a “black box” which seems to work without any obvious contact to environment.

But there is one concept bridging the gap between the two approaches, which is autonomy. The degree of autonomy characterizes both journalists’ actions and the structure of organizational systems, where media coverage is generated.

Autonomy is one of the key terms of the professionalization approach.3 It means freedom to shape journalists’ work without being controlled by internal and external powers. Besides expertise, commitment to the job and responsibility are used as the main indicators for professional orientations. This aspect is particularly crucial, insofar as journalism differs from “real professions” by its legal preconditions.

There are a lot of political, economic, organizational, and technological constraints which restrict journalists’ autonomy almost naturally. This assessment corresponds directly with the evidence that gatekeeper research offers.4 It emphasizes organizational self-regulation instead of individual self-determination or even defines the organization itself as the gatekeeper.5 Gatekeeper research started off with describing news selection as only determined by one single personand ended up with explaining news production as a result of cybernetic mechanisms within a media institution.7 All these individualistic, institutional or cybernetic perspectives have in common that media production has been described as autonomous processes.

Autonomy is, however, a Janus-faced factor in journalism. Journalists need a great amount of scope to do a good job because their autonomy is directly connected to freedom of the media. Journalists (individuals) ought to be free in selecting information and in covering stories; newsrooms (organizations) ought to be independent from external influences, such as commercial or political constraints; media systems (society) ought to have guaranteed press freedom and ought to be free from all kinds of censorship. For that reason, the degree of autonomy has been investigated in most of the 21 national surveys on journalism that have been published recently.8

But this is only one side of autonomy because autonomous journalism may also lead to being cut off from its audience or even to despise the audience.9 The notion of “fourth estate” expresses the fear of an illegitimate power of journalism.10 Although autonomy is a necessary condition of free journalism, it is not a sufficient condition for a free society because too much of it is often suspected of developing the journalists’ tendency to cut themselves off from society and opposing established institutions. Therefore, responsibility for “deciding what’s news”11 should not be monopolized by journalistic professionals�and it is in fact no longer, since Internet has opened the gates for all kinds of information.12

The concept of autonomy is also fundamental for modern social system theory, which has an increasing influence on communicator analysis (in Germany and elsewhere). This approach, however, prefers the term self-referentiality, which lays stress on the relationship between systems and their environments.

The purpose of this paper is to demonstrate that a system theoretically guided by observation of journalists’ autonomy provides a more precise insight into the processes which constitute journalism. Based on empirical data, we will explore how autonomy and self-referentiality are related to certain attitudes and behaviors of media professionals.

Theoretical framework and hypotheses

Previous research has used a narrow concept of autonomy, which defines autonomy as the absence of external influence and as independence from the environment. Autonomy can be observed in terms of causal relationship modelling external influences as cause (independent variable) and internal changes or reactions as effect (dependent variable). This theoretical perspective methodologically is often based on surveys of journalists, who can only report several external influences. The measurement of actual instead of reported or estimated autonomy or influence would require analysis both the environmental input into the system (e. g. public relations press releases and informal pressures on journalists) and the system’s output (media coverage).

As a consequence, we used a social systems theoretical perspective, which starts observation with the system’s operations themselves instead of the environmental influences on the system. The causal relationship has to be replaced by a self-referential perspective. The German sociologist and social system theorist Niklas Luhmann defines the autonomy of a social system as the system’s capability to manage or to self-regulate the relationship between self-reference and external reference. A (social) system is not called autonomous if it is entirely independent from its environment; rather it is autonomous if it is able to select certain areas in its environment from which it gets certain impacts and in like manner or at the same time is able to shield from influences of other environmental areas. In this sense autonomy includes both, self-referentiality and external referentiality.13Self-referentiality (or auto-referentiality) should not be conceptualized as a logical contrast to external referentiality (or allo-referentiality) because both kinds of references can empirically also be modelled as a mutual relationship.14

A broader concept of autonomy implies methodological advantages. Although autonomy cannot be reduced to self-referentiality, self-referentiality is a necessary and sufficient condition for autonomy. So it is convenient to operationalize the theoretical concept autonomy with the empirically observable concept of self-referentiality. The “self” can be measured in three dimensions: interpersonal relationship (e.g. with other journalists), relationship within the organization (e.g. with colleagues in the newsroom), and relationship with other media (e.g. with colleagues working for other media).

In journalism research, the status of (perceived) autonomy has been changing either as independent or dependent variable. In order to test consequences on attitudes and behavior of media professionals, we have fixed autonomy and self-referentiality in journalism as the independent variable. As a consequence, we do not try to explain, why journalists are more or less autonomous, but explore who the self-referential journalists are, which professional attitudes they have and which behaviors they perform. As dependent attitude and behavior variables we have analyzed journalists’ role perceptions, journalists’ images of their audience, their usage of unusual reporting practices, and their attitudes toward public relations sources.

We have suggested that self-referential journalists insist to a great extent on individual autonomy, and that they claim a complete journalistic responsibility to decide what news is. They view any constraints on their work as dysfunctional for media coverage. As conclusion from previous research and from our reflections on self-referentiality, we have raised four specific hypotheses on correlations between self-referentiality and the four dependent variables mentioned above:

  • Self-referential journalists perceive and perform their roles more communicator-oriented, but less service-oriented than journalists normally do. Therefore, they emphasize a more political role and try to affect rather than to serve the political system and the public.
  • Self-referential journalists have a more negative image of their audience than journalists normally have. They regard their audience as passive and uninterested, because they do not really care.
  • Self-referential journalists are less concerned about applying unusual and illegal reporting practices than journalists normally do. Their principal concern is to get information not regarding ethical deliberations.
  • Self-referential journalists are skeptical toward their sources, particularly toward organized sources like public relations. That is why they assess press-releases more negatively than journalists normally do.


The data used for this analysis base on the nationwide representative study on “Journalism in Germany.”15 The population under study included the full-time and the free-lance editorial staff of newspaper journals, news agencies, freesheets, magazines and radio and television stations. The complex sampling procedure consisted of a disproportional three-stage stratification, which could be corrected by weighting the sample with the correct parameters of the journalistic universe. The net sample finally included 1,498 journalists in East and West Germany. Although the design and questionnaire followed the example of Johnstone et al.16and of Weaver/Wilhoit17, the German study slightly differed from the U.S. surveys in defining journalism in a more extensive way. In additional to “hard news” journalists, we included infotainment journalists; we also interviewed journalists working for freesheets and other peripheral media; and we picked up professional freelancers. The interviews were conducted in person for two reasons: the questionnaire was too long for telephone interviews, and the telephone network in East Germany was � at that time � not sufficiently developed.18

For the purpose of this paper, we selected several variables as indicators for self-referentiality (cf. table 1): From a theoretical point of view, self-referentiality in journalism can be classified into (a) interpersonal, (b) organizational and (c) inter-media dimensions.

Table 1: Theoretical classification of several dimensions of self-referentiality in journalism.
theoretical dimensionempirical indicator(s)/variables(s)
interpersonal dimensionindicators for self-referentiality: 

1) mother or father also journalist, 2) journalists as one of three best friends, 3) percentage of journalists in private circle of acquaintances, 4) journalists in private circle of acquaintances 

indicators for external references: 

5) non-journalistic persons in private circle of acquaintances, 6) being in office of a non-journalistic organization
organizational dimension 

a) political integration 

b) professional integration 

c) work integration
indicators for self-referentiality: 

a1) difference between journalist’s political attitude and editorial policy, a2) difference between journalist’s political attitude and colleagues’ political attitudesb1) member in a journalistic organization or association, b2) statute for the editorial staff, b3) measures for promoting women journalistsc1) internal newsroom reactions on journalist’s work, c2) internal newsroom influences on day-to-day job, c3) frequency of story checkingindicators for external references:c4) external reactions on journalist’s work, c5) external non-journalistic influences on day-to-day job
inter-media dimensionindicators for self-referentiality:1) number of regularly used media, 2) number of regularly used politically left media, 3) number of regularly used politically right media, 4) reactions from other media’s colleagues on journalist’s work

With interpersonal self-referentiality we mean the fact that journalists have a lot of colleagues or other journalists in their private circles of acquaintances. If the journalist’s father or mother is also a journalist we took this as an indicator for self-reference, too. Indicators for interpersonal external references were non-journalistic persons in private circle of acquaintances or the fact that a journalist is in office of a non-journalistic organization.

Organizational self-referentiality included political, professional and work integration in the newsroom. Political integration means the absence of a difference between the journalist’s own political attitudes and those of his or her colleagues or of the editorial policy of the journalist’s media organization. A journalist is professionally integrated if he or she is member of a journalistic organization, association or union or if the media organization has established a statute for the editorial staff and particularly promotes women journalists, etc. Work integration refers to certain specific reactions or more general influences on the journalist’s work by his or her colleagues within the newsroom and to how often story checking is practiced. External referentiality means that journalists get various reactions from external non-journalistic sources, persons or organizations.

Taking inter-media self-referentiality into consideration is due to our theoretical approach, which supposes that journalism is a societal system. Although other media organizations are competitors for the newsroom and belong to the environment of the newsroom (which can be described as an organizational system), all media organizations are elements of the societal system journalism and belong to it. Professional reactions of other media’s colleagues on the journalist’s work are a kind of self-reference within the system-theoretical framework. The same is true for journalists using other media coverage for professional reasons, e.g. in order to get hints for new topics and issues. The more a journalist uses other media for reporting and investigating, the more this operation may be called a self-referential one.


The analytical strategy consisted of three steps: First, dimensions of self-referentiality could be explored with the help of a principal component analysis. Second, journalists who work within a self-referential structure could be separated from other groups of journalists with the help of a confirmatory cluster analysis. Therefore, we fixed the centroid or profile of one cluster on above average values of several variables indicating self-referentiality in journalism. Third, we were looking for differences between the self-referential cluster and the total sample19in sociodemographic variables, professional education, journalistic work, role performance, images of the audience, use of unusual reporting practices and use of press-releases. These comparisons were used to describe the self-referential cluster in detail and to test the hypotheses stipulated above, while the previous principal component and cluster analyses prepared this research objective stepwise.

Table 2: Principal component analysis of self-referentiality variables in journalism.
Indicators for self-referentialityfactor 1factor 2factor 3factor 4factor 5factor 6factor 7factor 8factor 9communality
number of regularly used media.90        .85
number of regularly used politically left media.80        .71
number of regularly used politically right media.75        .69
journalists as one of three best friends .78       .62
percentage of journalists in private circle of acquaintances .72       .61
journalists in private circle of acquaintances .52   .39   .45
external reactions on journalist’s work  .78      .68
reactions from other media’s colleagues on journalist’s work  .77      .62
difference between journalist’s political attitude and editorial policy   .82     .70
difference between journalist’s political attitude and colleagues’ political attitudes   .79     .66
measures to promote wormen journalists    .81    .72
statute for editorial staff    .79    .69
internal newsroom reactions to journalist’s work     .70   .59
frequency of story checking     .66   .56
external influence on day-to-day job      .85  .78
internal newsroom influence on day-to-day job      .68  .64
being in office of a non-journalistic organization       .80 .68
non-journalistic persons in private circle of acquaintances       .47 .50
mother or father also journalist        .87.81
member in a journalistic organziation or association       .35.53.53
explained variance10.
Only factor loadings greater than or equal to .30 are documented in the table.

(1) The principal component analysis resulted in two main findings (cf. table 2):

(a) There are more independent components or factors empirically found than theoretically presumed. Self-referentiality obviously is not a single dimension, but consists of different facets. In addition, the empirically found dimensions do not exactly correspond to those theoretically presumed. Two differences are worth being scrutinized:

We did not expect that external reactions and reactions of colleagues working for other media on journalist’s work load on the same factor. Certainly journalists think of their colleagues working for other media as competitors or, theoretically speaking, as not belonging to the same journalistic system. This is due to the fact, that journalists do their job within an organizational context, i.e. within a newsroom.

We did not expect either, that being member of a journalistic organization or association would correlate with having a journalist as father or mother. We expected, however, that it would correlate with a newsroom statute for the editorial staff or with the existence of measures to promote women journalists, which has proved not to be the case. Being organized in a professional association or union seems to be a more personal concern, whereas unionist achievements within the newsroom is a more political or organizational concern.

(b) There is no simple contrast between self-referentiality and external referentiality. On the one hand, both variables do not correlate negatively, as they should if they were the two poles of a unidimensional scale, but they actually do load positively on the same factor. This is the case with journalists believing as well that the journalists themselves or the newsroom as journalistic organization has most influence on media coverage as reporting a high influence of external sources, such as political parties, enterprises, trade unions, economic associations or other societal institutions. In other words, a high external pressure corresponds with a high internal integration within the newsroom. Of course, this correlation cannot be interpreted in a causal manner.

On the other hand, both variables load on different factors and are independent. This is the case with external and inter-media reactions (factor 3) and with internal reactions on journalists’ work and with frequency of story checking within the newsroom (factor 6).

(2) In addition to the substantial findings, the principal component analysis was used to reduce the variables to independent dimensions. Redundant variables could now be weighted less in the subsequent cluster analysis than those variables loading highest on their factors. The results of the principal component analysis helped us selecting relevant variables for the next step of data analysis.

Primarily, the cluster analysis was not applied to explain as much variance of the analyzed variables as possible, but to create one specific cluster with the profile of high self-referentiality in several dimensions.20 The profiles of the remaining clusters were of only minor interest, so we just processed a two clusters solution. The first trial to define the self-referential cluster as a profile with the above average values of all variables indicating self-referentiality failed, because the algorithm could not find any journalist who fitted to the given profile.

As the self-referential journalist did not exist, we had to fix a more moderate centroid including only relevant variables, which was finally defined as follows (cf. table 3):

Table 3: Profile of the self-referential cluster.
Indicators for self-referentialityScales of variablesMeans of self-referential cluster
Overall means
mother or father also journalist0-10.030.06
journalists as one of three best friends0-10.530.51
percentage of journalists in private circle of acquaintances0-99 (%)29.4825.15
journalists in private circle of acquaintances0-10.93**0.84
non-journalistic persons in private circle of acquaintances0-41.010.84
being in office of a non-journalistic organization0-30.12*0.20
difference between journalist’s political attitude and editorial policy0-71.711.69
difference between journalist’s political attitude and colleagues’ political attitudes1-52.802.85
member in a journalistic organization or association0-10.490.56
statute for the editorial staff0-21.55*1.37
measures for promoting women journalists0-31.010.91
internal newsroom reactions on journalist’s work0-22.00***1.46
reactions from other media’s colleagues on journalist’s work0-11.00***0.33
external reactions on journalist’s work0-52.81***2.05
internal newsroom influences on day-to-day job1-54.36***3.80
external non-journalistic influences on day-to-day job1-53.573.48
frequency of story checking1-54.75***3.65
number of regularly used media1-2411.83***8.36
number of regularly used politically left media0-83.71***2.44
number of regularly used politically right media0-61.78***1.29
Variables in italics were fixed above average for the profile of the self-referential cluster.
*, **, *** = significantly different from overall means (p < 0.05, p < 0.01, p < 0.001)

Journalists to be grouped in the self-referential cluster had to get more inter-media reactions as well as reactions from their own newsroom on his or her work than the journalists in the remaining cluster do. Furthermore, they had to report more influences on coverage of sources within the newsroom, had to get their stories checked more frequently and had to use more other media for professional reasons than journalists generally do. In addition to this theoretically given profile, cluster analysis algorithm empirically grouped journalists in this cluster who more often have journalists in their private circle of acquaintances, who more often have a statute for the editorial staff in their newsrooms, and who get more external reactions on their work than journalists in the remaining cluster do.

The self-referential cluster estimated by the algorithm finally consisted of (only) 93 journalists. It can be described with the following characteristics: There are more journalists in this cluster than in the whole sample working for the public TV (15.2% vs. 6.2%), but not a single journalist working for freesheets (0% vs. 10.9%). More of them than overall frequency cover a cultural (20.2% vs. 11.9%), but less a local and regional beat (11.7% vs. 18.4%). There are no significant differences in position in the newsroom hierarchy, in years of education, or in gender. Self-referential journalists more often only have a training on the job (28.3% vs. 20.0%), but do not differ from average in academic ways of professional qualification (study of journalism, communication science or another subject).

They are three years younger (34 years vs. 37 years), have two years less professional experience than average (8 years vs. 10 years), and they work three hours more a week (49 h vs. 46 h). With regard to specific journalistic occupations, we found out that journalists belonging to the self-referential cluster do less selecting (37 min. vs. 49 min.) but more organizing and administrative work (90 min. vs. 69 min.) than the average journalist does. There are no further differences in spending time for reporting, writing or editing activities.

Test of hypotheses

For the test of the hypotheses we compared means of several items of role perceptions, reporting practices, images of the audience and assessment of the quality of press-releases, again between the self-referential cluster and the total sample of journalists.

(1) Role perception included 21 items, which could be combined into different independent dimensions, such as political, idealistic, entertaining and service-oriented, and information journalism.21 As table 4 depicts, the first hypothesis can be mainly verified. Indeed, self-referential journalists support particularly some politically oriented professional aims, such as “controlling politics, business and society,” “being an adversary of public officials,” “investigating claims and statements of the government,” “being an adversary of business” and “influencing the political agenda.”

Table 4: Role perception, reporting practices, and assessment of press-releases*
Items**Means of self-referential cluster
Overall means
Role perceptions:
providing analyses and interpretations of complex problems4.304.01
criticizing bad states of affairs4.023.73
developing intellectual and cultural interests of the public3.723.45
being an adversary of public officials by being constantly skeptical of their actions3.342.95
controlling politics, business and society3.302.94
presenting the own opinion to the public3.112.80
investigating claims and statements made by the government3.112.73
being an adversary of business by being constantly skeptical of their actions2.962.64
influencing the political agenda and set issues on the political agenda2.832.45
Reporting practices:
using confidential government documents without authorization2.992.58
getting employed in an organization to gain inside information2.892.53
paying people for confidential information2.552.30
agreeing to protect confidentiality and not doing so1.231.37
Assessment of press-releases:
information in press-releases is reliable3.043.36
press-releases are well prepared2.703.06
there are too many press-releases3.573.32
press-releases tempt to uncritical reporting3.152.85
* Table 4 documents only significant differences between means of self-referential cluster and of complete sample (p < .05). There were no significant differences in items of images of the audience.
** All scales range from 1 (do not agree at all) to 5 (do agree entirely).

In addition, there are four roles emphasized by self-referential journalists, which do not strictly perform political functions, but which are also communicator-oriented: They approve to a significant higher degree of “criticizing bad states of affairs,” “presenting their own opinion to the public,” “providing analyses and interpretations of complex problems” and “developing intellectual and cultural interests of the public.”

The second part of the hypothesis says that self-referential journalists are less interested in service-oriented coverage, e.g. “providing entertainment and relaxation” or “helping people in their everyday lives.” This part does not approve to be true because there are no significant differences between self-referential cluster and total sample. We interprete this finding as another indicator for the multi-dimensional character of self-referentiality. Even if self-referential journalists do favor an active role performance, including political or influential journalism, they do not necessarily neglect other functions of journalism. There is not a single item of which self-referential journalists significantly approve to a less degree than the total sample. In sum, they belong to an ambitious species of journalists.22

(2) Images of the audience can be divided in three dimensions: civic characteristics, e.g. politically (un)interested, (ir)responsible, (un)committed, well/badly informed, or (un)educated; ideological aspects, e.g. conservative/progressive, (in)tolerant, narrow-/open-minded, politically left/right; and social class related descriptions, e.g. poor/rich, (un)influential or self-confident/anxious. In no case self-referential journalists differ significantly from other journalists. Obviously, self-referentiality is not correlated with certain characteristics of images which journalists develop of their audience. This finding can also be interpreted within the frame of multi-dimensionality of self-referentiality. Self-referentiality does not imply a specific negative or even cynical view of the audience. Both variables are simply (statistically) independent.

(3) Self-referential journalists are also supposed to be unscrupulous with regard to reporting practices. Previous findings suggest that German journalists have refined attitudes toward unusual methods to get information: They legitimize some of them moderately depending on the circumstances, which we called “aggressive methods,” and consequently disapprove of others, which we labelled “unscrupulous methods.”23

Although this difference seemed to occur particularly in Germany24, we found it again, when correlating reporting practices with self-referentiality in journalism. Self-referential journalists agree with some aggressive methods to gain information to a higher degree than the total sample does, but do also disagree with one specific unscrupulous method more than journalists normally do. So they gradually approve more than all journalists together of “paying people for confidential information,” of “using confidential government documents without authorization” and of “getting employed in an organization to gain inside information” and in like manner disapprove of “agreeing to protect confidentiality and not doing so.” Even more than journalists normally do, self-referential journalists feel a difference between unscrupulous methods and unusual methods which they nevertheless consider legitimate under certain circumstances.25

In sum, we do not have to reject this hypothesis entirely, because self-referential journalists do indeed agree to unusual reporting practices more than the total sample of journalists. The reason for this aggressive comprehension of their work, however, does not seem to emerge from ethical immaturity or from a lack of ethical consciousness because the self-referential cluster disapproves of all unscrupulous methods at all and of one of it even more consequently than all journalists together. Again, self-referentiality does not imply any kind of self-closure or a lack of sensitiveness toward the environment. What is true for autonomy, is also valid for self-referentiality: A high degree of self-referentiality corresponds with a high degree of environmental attention.

(4) Finally, we expected self-referential journalists to be skeptical of public relations sources. Respondents should therefore assess nine statements toward the quality of press-releases. Indeed, the self-referential cluster proves to be less pragmatic and more critical than the average journalist. Self-referential journalists consider press-releases less reliable and worse prepared than journalists normally do. In addition, self-referential journalists complain about the existence of too many of them more intensively and suspect more often to be tempted to uncritical reporting by press-releases. Although differences between self-referential cluster and total sample are only significant for these four of all nine items we asked respondents to assess, the tendency is clear, and the hypothesis is verified for the most part.


For a long time, researchers thought that journalism could be described in terms of contrasting categories. Particularly journalists’ role perceptions were analyzed in simple dichotomies, such as neutral vs. partisan, and also journalists’ political attitudes. In the case of autonomy, which is a key term of different approaches in the field of communicator analysis, we could demonstrate that social system theory can help to observe journalism in a more complex and adequate way. Our study has shown that autonomy is a multidimensional construct.

In sum, autonomy does not imply that journalists have no contact with the environment, but that their relationships with environment are more selective. Eventually, autonomy does not mean a system’s independence from external impact, but is the system’s internal selection of which external impacts are allowed, to put it in terms of social system theory. It seems reasonable to replace an “either … or” logic by an “as well … as” logic. We found some empirical evidence for that in the case of journalistic profession.

Appendix: Methodological aspects of the study “Journalism in Germany.”
type of media organizationnewspaper journals (including freesheets)
news agencies and media service bureaus
general-interest and special-interest magazines
public and private broadcast stations and networks
mode of employmentpermanently employed editorial staff and professional freelancers
type of beatsnews people and infotainment journalists
sampling proceduredisproportional three-stage stratification
sample sizeN = 1,498
mode of interviewingface-to-face interviews
questionnairecomparable to Johnstone et al. (1976) and to Weaver/Wilhoit (1986, 1996)
fieldworkApril to September 1993

About the Authors:

This article was presented as a paper to the Mass Communication and Society Division at the Annual Convention of the Association for Education in Journalism and Mass Communication, New Orleans, LA, August 4-7, 1999. Scholl is associate professor at University of Muenster, where Weischenberg is professor at the same institute and director of the journalism department. They can be reached at:

Universit�t M�nster
Institut f�r Kommunikationswissenschaft
Bispinghof 9-14
48143 M�nster
Phone: 0049-251-8324261; Fax: 0049-251-8328394
e-mail: and 

Leave a comment

Your email address will not be published. Required fields are marked *