Organizational Influence on Scientists’ Efforts to Go Public

Abstract

This article contributes to the debate on the influence of organizational settings on scientists’ media contact. Drawing on a quantitative survey of researchers (n = 942) from 265 German universities, the results indicate that a large proportion of scientists from all disciplines participate regularly in the dissemination of research findings. The authors provide evidence that scientists’ media efforts are influenced by how they adopt their university’s desire to be visible in the media, as well as by the university’s PR activities. The increased orientation toward news media is discussed in the light of the new governance of science within Europe.

Keywords

organization communication public relations quantitative analysis new governance of science medialization of science

Introduction

While a growing number of studies focus on scientists’ reasons for “going public” (Dunwoody, Brossard, & Dudo, 2009; Gascoigne & Metcalfe, 1997; Martín-Sempere, Garzón-García, & Rey-Rocha, 2008; Peters, Heinrichs, Jung, Kallfass, & Petersen, 2008; Poliakoff & Webb, 2007; Tsfati, Cohen, & Gunther, 2011), little attention continues to be paid to the potential influence of scientific organizations on scientists’ media efforts. Nonetheless, some scholars have indeed recognized the increasing importance of universities promoting themselves and their personnel (Engwall, 2008; Jacobson, Butterill, & Goering, 2004; Kyvik, 2005; Peters, Heinrichs, et al., 2008; Rowe & Brass, 2011). Recent evidence suggests that “the role of organizations in science communication has become more important relative to individual initiatives and responses to media attention” (Peters, Brossard, et al., 2008, p. 271). These findings contrast with other studies that focus on the influence of differing research areas or academic disciplines (Jensen, 2011; Kyvik, 2005; Rödder & Schäfer, 2010).

The aim of this article is to investigate the role of universities in shaping scientists’ willingness to go public, by contrasting it with the role of the academic disciplines. To the best of our knowledge, this is the first study addressed to this particular topic on the empirical basis of all universities within one country. Although existing studies are important to understanding recent developments, their findings are largely descriptive. In this article, we compare the effects of organizational and disciplinary factors through statistical analysis. We make use of neo-institutional theory to sort our data and interpret them against the backdrop of a fundamental change in the organizational governance of science.

New Governance of Science in Neo-Institutional Perspective

By 1990 at the latest, public administration in general and the higher education system in particular were permeated by the neoliberal concept of New Public Management (Braun & Merrien, 1999). In most OECD (Organisation for Economic Co-operation and Development) countries,¹ universities transformed themselves into entrepreneurial organizations (Clark, 1998). The new governance of science is accompanied by ongoing and increasing pressure on universities to demonstrate their legitimacy, not least in the media (see Jansen, 2007). Germany, on which we concentrate in the empirical part of this article, can be described as a “European latecomer” (Kehm & Lanzendorf, 2007, p. 158) in implementing governance reforms. In recent years, competitive structures were created to reinforce these reform policies (Kehm & Lanzendorf, 2007). The pace of reform differed across countries, but “in any case, reforms are rather far-reaching and can be assumed to have fundamental impacts on the work of universities” (Kehm & Lanzendorf, 2007, 2007, p. 154). This is especially true for central European universities, which were set free from formerly extensive state control but still depend largely on public funding. Hence, universities have gained more autonomy, although at the cost of increasing pressure to garner public support for their stagnant public base funding and to raise additional third-party funding (Friedrichsmeier, 2012). Therefore, they have to meet the expectations of various stakeholders (Delanty, 2001) and compete with a diverse range of other public service organizations in a radically transformed organizational field. The most prominent approach to conceptualizing the potential effects of new challenges on corporate governance is the neo-institutional theory (DiMaggio & Powell, 1983; Scott, 1994). This approach provides a sound explanation of why and how New Public Management reforms may stimulate a university’s interest in public visibility. The key concept of neo-institutionalism—that of legitimation—was originally introduced by Max Weber and refers to the ability to convince others that something is “right,” that is, that the activities of a university are justified. To ensure legitimacy, organizations anticipate the expectations of their institutional environment. Therefore, they observe how competing organizations promote their eligibility for public support. Major shifts in the “organizational field” are supposed to have fundamental effects on the legitimization efforts of an organization and its members (DiMaggio & Powell, 1983). Through this perspective, organizational processes relate directly to organization members and their assumptions of how to meet such institutional expectations. Beyond doubt, the new governance of science entails a fundamental transformation of the institutional environment of universities and therefore compels them to intensify their legitimation efforts. These efforts may have an important impact on science, as some scholars even claim that the “distinguishing feature of contemporary universities is the extent of the (self-)marketing practices in which they have come immersed” (Wernick, 2006, p. 566).

Hence, the so-called crisis of legitimacy of science (see Weingart, 2001) is affecting primarily scientific organizations and not scientists or disciplines in the first place (Holmström, 2005, p. 498). This change that scientific organizations are undergoing probably has an effect on the media efforts of their members. In order to confront the challenge of displaying their scholarly competitiveness, as well as their service to the public, universities have become increasingly interested in scientists’ public communication. This is linked with the fact that the legitimacy of science depends far more on research and scientific results than on teaching (Bishop, 2006, p. 571). The most effective way to achieve media visibility and public attention is therefore to communicate new research results and scientific expertise (Engwall, 2008, p. 38).

From Visible Scientists to Visible Universities

Since favorable publicity constitutes a powerful resource in the developing science market, media resonance has become a major concern of universities. The significance of the media is related to the principle of everyone assuming that important others will also take notice, referred to in the research literature as “the influence of presumed influence” (Gunther & Storey, 2003; inspired by Davison, 1983). This approach emphasizes the importance of self-reinforcing effects: That is, the stronger the presumed media influence on others, the more one’s own media motivation and effort are supposed to increase (Tsfati et al., 2011). As noted in the literature on university public relations (PR), scientific organizations perceive an increased urgency to attract positive media coverage and to minimize negative publicity (Borchelt, 2008; Rowe & Brass, 2011). For this purpose, several universities now foster relationships with journalists and instruct their academic staff actively to communicate their research and expertise (Brass & Rowe, 2009; Engwall, 2008; Kallfass, 2008; Rowe & Brass, 2008).

Important policy initiatives promote the notion of a moral obligation toward media efforts: “Scientists must learn to communicate with the public, be willing to do so, and indeed consider it their duty to do so” (Bodmer, 1985, p. 6; see also Ehlers & Committee on Science, 1998). The effects of science policy initiatives on media-related issues are noted by many authors (e.g., Martín-Sempere et al., 2008; Miller, 2001; Pearson, 2001; see also the German “Science in Dialogue,” Winter, 2004). As yet, however, there is no consensus on how to deal with this entanglement with higher education policy. Several studies have pinpointed that this norm is an important predictor of scientist media motivation and effort (Tsfati et al., 2011; see also Bentley & Kyvik, 2011; Gregory & Miller, 2000). At the same time, it is reasonable to assume that the perceived obligation to disseminate scientific results beyond the scientific community arises from within the university, as well as from the discipline. There are many indications that a sense of a “public duty” has spread throughout the academic organizations. Within the scope of other factors relating to scientific culture and professional motivations, one study found that the “sense of duty” correlated most with the motif “make my center better known or more visible” (Martín-Sempere et al., 2008, p. 360). Kyvik (2005) found that the vast majority of faculty members perceive an increased pressure to disseminate their research.

As such, scientists’ perceptions of a “public duty” contain little information on what makes him or her think that way. Nonetheless, the timing of change in these perceptions may provide some useful indications. There is a mounting perception of a “new commandment” (Gregory & Miller, 2000, p. 1) to scientists, and it coincides perfectly with new public management reforms by universities. Recently “renewed and growing attention” (Dunwoody et al., 2009, p. 299) to the media efforts of scientists might be attributed to a renewed need for organizational legitimization as a result of a new governance of science. In order to shed more light on this background, this article investigates two channels through which such expectations are conveyed and manifest themselves in observable behavior. The two channels are the disciplinary communities and the academic organizations.

Discipline

Scientists willing to engage in raising the public visibility of their university may find themselves in conflict with scientific reputation (Rödder, 2012). The latter derives from academic disciplines (Clark, 1983), so we should assume that the disciplinary background strongly affects scientists’ media efforts. Some disciplines are affected by a particularly strong sense of duty to go public (e.g., Smith, Singer, & Kromm, 2010). By contrast, disciplines maintaining traditional scientific norms may nourish a “touch-me-not” attitude toward journalists (Dunwoody, 1986, p. 8). These disciplines may even actively impede the public prominence of their members to some degree, for instance, by routinely challenging the academic reputation of those who give priority to media appearance. By doing so, they inhibit the use of media prominence in order to avoid skepticism from within the profession (see also Gregory & Miller, 2000).

Although scientific norms are a potential barrier to media contact, there is some evidence that this is less significant than previously assumed. A recent five-country survey indicates that almost half of scientists expect “possible critical reactions from peers,” while the other half refer to an “enhanced personal reputation among peers” as a result of media coverage (Peters, Brossard, et al., 2008, p. 204; see also Peters, Heinrichs, et al., 2008). Against this background, we test the predictive power of the perceived reputational consequences of a scientist’s discipline on his or her media efforts. As discussed above, supporting as well as impeding effects of the discipline on media efforts have to be taken into account. This leads to our first hypothesis:

Hypothesis 1: Scientists’ media efforts vary according to perceived reputational consequences (loss or gain): The stronger the positive recognition of public prominence by the discipline, the greater the extent of scientists’ media contact and vice versa.

Organization

Traditionally, scientists primarily identify themselves with their disciplinary community—and far less with their organization (Clark, 1983). The preeminence of this disciplinary identification is challenged, since the university is regarded increasingly as a most essential entity in which to thrive in the face of competition with many others (Krücken & Meyer, 2006). Thus, it is necessary to examine the extent to which researchers anticipate the need for media visibility, in order to legitimate the academic organization as a whole. In line with the neo-institutional approach, we account for (Hypothesis 2) scientists’ perceptions of their university’s need to legitimize via the media and the acceptance of this challenge by scientists. Hypotheses 3 to 6 deal with structural conditions of universities that may be connected with their legitimizational needs, and Hypotheses 7 and 8 relate to organized activities in order to foster a university’s public recognition.

A scientist’s adoption of organizational concerns for media visibility can develop in three stages. At first, he or she may perceive a university’s need to strive toward esteem in the media, potentially without feeling obliged to support the associated efforts actively. Second, a scientist may observe that his or her colleagues are concerned about a university’s media visibility. This equals to observing of a social relevance of organizational media visibility in his or her field. Third, a scientist may have adopted this concern and individually demonstrates an interest in the media visibility of his or her university. Our hypotheses are driven by the assumption that organizational influence can be related to the perception of a university’s media orientation and affects researchers’ efforts to engage in public communication:

Hypothesis 2: A strong concern to make the university visible in the media is associated with a higher frequency of media efforts.

Beside individual perceptions, as explained above, structural conditions may affect a university’s need to strive for visibility in the media. First, we differentiate between two types of universities, a distinction that has strong structural implications in Germany. Research universities (Universitäten) generally undertake substantial research. Their staff have relatively low teaching loads, whereas professors at “university of applied sciences” (Fachhochschulen) have teaching loads that are roughly twice as high. German higher education laws treat teaching performance as the main objective of this second type of universities. Therefore, the legitimization of universities of applied sciences presumably depends mainly on teaching and only to a lesser extent on research exposure and prominence in the media. Thus, there may be greater pressure on scientists to promote media visibility in research universities. Second, the effects of university funding are relevant. Whereas German public sector universities (state and ecclesiastical) receive more than 80% of their funding from the state (Statistisches Bundesamt, 2012), private universities receive only partial or temporary start-up funding. Therefore, they are in particular need of the favor of private fund givers, potential tuition fee payers, and alumni. Third, German universities are characterized by considerable disparity in size. The number of academic personnel ranges from 5 to 8,285. It is generally assumed that small universities have limited PR resources and are less recognized by the public, so there is a certain probability of critical size effects. In the light of several studies (Engwall, 2008; Rowe & Brass, 2008; Weingart & Pansegrau, 1999), we assume an effect of the extent to which universities rely on third-party funding. In order to attract the attention of private and public funding partners, media visibility is presumably of greater importance if the share of third-party funding is high (Dunwoody, 1986; Dunwoody & Ryan, 1985; Gascoigne & Metcalfe, 1997; Peters, Heinrichs, et al., 2008).

These considerations result in four further hypotheses:

Hypothesis 3: Scientists at research universities conduct more media efforts than those at universities of applied science.

Hypothesis 4: Private universities are associated with higher frequencies of media contact than public sector universities.

Hypothesis 5: Large universities are associated with higher frequencies of media contact than small universities.

Hypothesis 6: Universities with a high proportion of third-party funding are associated with higher frequencies of media contact than universities with a low proportion.

The Role of Public Relations

Organizational support of scientists’ media efforts is concentrated in the media offices of universities. In the past few decades, a global trend toward the establishment of media offices in higher education institutions has been evident. Almost all universities are investing in PR in order to provide media visibility and raise stakeholder acceptance (Borchelt, 2008; Engwall, 2008; Weingart, 2001). Scientists can seek the support of a PR person with respect to drafting press releases and establishing contact with journalists (Gregory & Miller, 2000). PR departments cultivate traditional media relations, thus focusing more on press releases and press clippings than on strategic issues and decisions (Borchelt, 2008; Gregory & Miller, 2000; Kohring, Marcinkowski, Lindner, & Karis, 2013). The typical press officer is “the practitioner in the middle” (Rogers, 1986) of science and the media. However, as systematic research on PR of universities is generally rare (Borchelt, 2008), we know little about the interactions of PR professionals and scientists. Even less is known about the influence that PR efforts exert on scientists’ media contact.

As a considerable influence of PR offices is widely assumed in the literature, but less so investigated, we aim at providing evidence supporting this assumption. It can be argued that this influence depends on the degree of PR activity and expertise. Considering the differing resources and specific task profiles of PR offices, PR professionals have varying opportunities for maintaining regular contact with researchers. The more they are able to request news items, the more they can facilitate press releases about research results. With growing enquiries from PR persons, researchers anticipate the expectations and needs of their organization. As a consequence, they are likely to enlarge media contact. Given the traditionally strained relationships between science and the media (Peters, 2008), such influence is presumably also dependent on PR’s ability to create an impression of competence. If scientists perceive expertise and feel well supported, they will probably be more willing to cooperate. Moreover, PR offices may shape scientists’ media affinity to a certain degree, by compiling and providing press clippings. By reading the latter, scientists not only receive feedback on their own communication efforts but also recognize that their peers are becoming publicly visible. Overall, university compilations of press clippings demonstrate visible activities of peers and university expectations to attract public attention. It can be assumed that a more frequent usage of press clippings is related to a more pronounced ambition to participate actively in producing news. Therefore, we propose the following:

Hypothesis 7: The more scientists are influenced by professional PR departments and personnel, the greater the extent of scientist media efforts.

To date, there has been little research on the question of whether scientists approach their university’s PR office or whether they prefer to establish their own personal contact with journalists. Almost 30 years ago, Dunwoody and Ryan (1983) found the status of PR in scientific institutions to be marginal as scientists “overwhelmingly preferred direct contact with journalists to other mechanisms, such as press releases, that would channel their work through the public information office” (p. 655). By contrast, there is the stereotype of scientists that “might speak to a press officer but not to a journalist” (Gregory & Miller, 2000, p. 82). A recent study by Peters (2008) reported a substantial influence of PR offices that might also lead to self-reinforcing effects. In all likelihood, scientists receive more journalistic inquiries through regular collaboration with PR (Peters, 2008). Furthermore, it is reasonable to assume that reservations and inhibitions about science journalism decrease with the first successful press releases. Being supported by the PR department in drafting press releases, scientists become better known and can gain both competence and self-confidence in managing further journalistic inquiries. Accordingly, we assume a strong correlation between both dependent variables and derive our final hypothesis:

Hypothesis 8: The more scientists seek the support of the PR department to draft press releases, the more contact they seek with journalists, and vice versa.

Method and Data

The study is part of a major research project analyzing PR issues in a context of shifting university governance in Germany and has been funded by the Federal Ministry of Education and Research. We conducted a whole-population survey of all state, private, and ecclesiastical universities in Germany. The investigation includes survey data from members of the administrative board (rectorate), PR managers, university council chairs, and scientists involved in the collegial board of self-government (academic senate). The latter are of interest in the report on hand.

Survey Data

The most favorable sample for an investigation of factors that promote media efforts comprises senior scientists, as they are likely to reveal a minimum level of such efforts. A high proportion of scientists with no observable media efforts at all would hinder the statistical analysis of factors promoting these efforts. Several studies have registered a considerable proportion of young scientist who do not undertake any of such efforts, which renders them unfit for our purposes, whereas seniority has a positive effect on media efforts (Dunwoody et al., 2009; Martín-Sempere et al., 2008). For each university, we contacted a sample of senior researchers who represent the disciplinary affiliation of professors at a given university. This sample was derived by questioning all professorial members of the senate, which is a central body of academic self-government. At German universities, members of the senate are elected by their peers in a combined group of academic departments. The duration of their membership at the university is above average. In Germany, there is no tenure track, and the age of professors tends to be higher than in several other countries. Accordingly, our sample consists of established senior researchers who have a good overview of the research proceedings and associated activities at their university. The questionnaire was written in German. In sum, the invitation to participate was sent to a total of 265 higher education institutions and to nearly 2,000 scientists. The web-based survey was programmed by means of the Electronic Fields Survey Software EFS (Décieux, Heinz, & Jacob, 2011). The respondents received a postcard announcing the survey, which was followed by an initial mailing of the survey instrument. E-mail reminders were sent twice to those individuals who had not yet responded. As a result, the questionnaire was completed by 1,029 scientists during September and November 2010, yielding a response rate of 52%. A missing data analysis revealed that 87 respondents (8.5%) had missing data on 20% or more of our variables. They were excluded, so the analysis was conducted on a final sample of 942 scientists.

Operationalization

The statistical investigation is based on 2 dependent and 14 independent variables, explained further below. Eight independent variables are single-item measurements that resulted from our survey, and the other 6 are based on statistical data from official sources.

Dependent Variable: Self-Initiated Professional Press Releases

The first dependent variable aims at capturing the extent of media efforts that are supported by the press office of the university. Scientists responded to the open-ended question: “How often do you seek the support of the press office to draft press releases?” Due to the fact that this is a self-initiated activity, respondents were asked to state a definite number for this activity in a typical half-year (semester, customary academic time scale in Germany). Based on this most important time unit of academic life, the test measured the extent to which the drafting of press releases can be considered a regular activity (M = 1.3, SD = 1.8).

Dependent Variable: Direct Contact With Journalists

In order to include media efforts that are based on direct contact with journalists, respondents were asked, “How often do you get in touch with journalists?” This second dependent variable contains scientists’ initiatives as well as journalists’ initiatives, including different communications such as short statements and telephone calls. Because of this variety, direct contacts with journalists could rarely be remembered exactly (Dunwoody et al., 2009). Therefore, respondents replied on a scale with five options: (almost) no contact (coded 1), at least once a half-year, at least once a month, at least once a week, (almost) daily (coded 5; M = 1.7, SD = 0.8).

Independent Variables: Disciplinary Influence

This block contains four variables, two related to the area of research, another two on supporting and impeding effects of the discipline.

Areas of research: We aggregated three clusters of academic disciplines: (a) medical science, (b) humanities and social sciences, and (c) natural and engineering sciences. In the following regression analysis (c) serves as a reference category.

Scientific reputation: We considered two dimensions of perceived effects and used an 11-point Likert-type scale ranging from strongly disagree (coded 0) to strongly agree (coded 10).

On one hand, a discipline’s preventing scientists from giving priority to the media was questioned, “In my scientific community, one’s reputation would suffer, if research findings appeared in the media first” (M = 2.8, SD = 3.1). On the other, the perceived recognition of public prominence by the discipline was measured: “Colleagues with high media prominence are frequently invited to high-level conferences” (M = 5.9, SD = 3.0).

Independent Variables: Organizational Structures

This block is based solely on structural data that were obtainable from the German Federal Statistical Office.

Research university: The data set contains information about the type of institution, including the dichotomy between universities of applied science (coded 0) and research universities (coded 1).

Private university: We coded 0 for public sector universities and 1 for private universities.

Small university: Given that German universities vary considerably in size, we constructed one variable on the basis of the number of scientific staff members during 2010. Small organizations with less than 200 scientists were coded 1, while those with at least 200 staff members were coded 0.

Third-party funding: In this case, it is important to consider not only absolute figures but also third-party funding related to the budget of the institution as a whole. Accordingly, the extent to which total expenses are covered by third-party funding was calculated. As in most cases medical facilities have extremely high nonacademic expenditures compared to other faculties, medical expenses were likely to distort the findings on a university and were therefore excluded from this specific calculation.

Independent Variables: Concern for Organizational Media Visibility

This block consists of three items that were measured with an 11-point Likert-type scale ranging from strongly disagree (coded 0) to strongly agree (coded 10).

Perceived need for organizational legitimization via media: This item focused on the extent to which media visibility is perceived as a need the university has to meet: “The university has to strive for a better acceptance by mass media” (M = 5.6, SD = 2.5).

Observed relevance to others: This item is constructed to capture the extent to which a scientist observes that colleagues care about the media visibility of their university: “At my university, there are more and more talks about how we are represented in the media” (M = 6.3, SD = 2.8).

Adoption of concern for organizational media visibility: This item measures a scientist’s personal identification with organizational needs to be visible in the media: “It’s important to me that my university attracts positive media coverage” (M = 7.2, SD = 2.2).

Independent Variables: PR Influence

PR effort and expertise were covered by three items.

Press clippings usage: We assumed an influence from the process of observing university peers becoming visible. “How often do you read your university’s compilation of press clippings?” This was measured with an 11-point Likert-type scale ranging from never (coded 0) to always (coded 10; M = 4.6, SD = 3.9).

PR expertise: Respondents were asked to evaluate the core skill of a PR manager, thus expressing their confidence in receiving good support from their university PR manager. “The PR manager has expert knowledge when it comes to creating a positive public image for my university.” To validate this judgment, respondents used an 11-point Likert-type scale, ranging from strongly disagree (coded 0) to strongly agree (coded 10; M = 4.9, SD = 3.1).

PR activity: The extent of this activity was measured with the open-ended question: “How often does the press office ask for pieces of news or propose press releases?” Respondents were asked to state a specific number for this activity in a typical half-year (semester; M = 1.3, SD = 2.1).

Results

After demonstrating the extent of scientists’ media efforts and further descriptive findings, we analyze organizational influence by using regression analysis.

Descriptive Findings

As noted above, we investigated scientists’ media efforts regarding self-initiated professional press releases, as well as direct contact with journalists. Similar to the findings of Jensen (2011, p. 28), we found three types of scientists: the “inactive” (no contact at all), “open” (engaging regularly), and “always active” (monthly contact) scientist. Table 1 shows the results of the different ways to disseminate research results.

Table 1.

Media Efforts in a Typical Half-Year.

Professional press releases		Contacts with journalists
0:	36%	(Almost) no contact: 50%
1:	36%	At least once a half-year: 34%
2:	15%	At least once a month: 13%
3:	5%	At least once a week: 2%
4-5:	6%	(Almost) daily: 1%
6-20:	2%
N = 876		N = 938

In our survey, one in two scientists has no contact at all with journalists in a typical half-year. By comparison, one in three scientists does not issue press releases with the help of PR. However, taken together, a relatively small group of scientists are regarded as completely “inactive.” Only 22% of all respondents neither get in touch with journalists nor draft press releases. In accordance with existing studies (Jensen, 2011; Kyvik, 2005), a considerable amount of all media efforts is undertaken by only a minority of scientists. These highly media-oriented scientists are usually less dependent on support by PR professionals, compared to others. However, a large proportion of scientists engage in public communication at least once a half-year.

Hypothesis Tests

Regression analysis was used to explain the differing levels of scientist media contact. In the case of the count variable (number of press releases), we conducted a stepwise linear regression analysis in order to identify the importance of the levels of possible influence discussed above. The results obtained from this analysis are presented in Table 2. Each block’s contribution to the explained variance is indicated by values for R² change.

Table 2.

Stepwise Linear Regression Analysis of Self-Initiated Professional Press Releases (N = 788).

	Model (R² change)
Independent variable	Model 1 (.02**)	Model 2 (.02***)	Model 3 (.02***)	Model 4 (.27***)
Disciplinary influence
Medical science	.12***	.13***	.12***	.09***
Humanities and social sciences	.04	.02	.04	.02
Reputational damage	.00	.01	.01	−.03
Reputational gains	.02	−.01	−.02	−.01
Organizational influence
Organizational structures
Research university		−.10**	−.08*	−.05
Private university		.07*	.05	−.03
Small university		.08*	.07*	.00
Third-party funding		.08*	.07	.06
Concern for organizational visibility
Perceived need			.00	.02
Observed relevance to others			.04	−.01
Adoption of concern			.13***	.07**
PR influence
Press clippings usage				.09***
PR expertise				.07**
PR activity				.52***
Adjusted R²	.01	.03	.05	.32

Note: *p ≤ .10. **p ≤ .05. ***p ≤ .01.

According to the ordinal measurement of contact with journalists, stepwise binary logistic regression analysis was chosen (Table 3). For this purpose, the dependent variable was dichotomized into one that distinguishes between no contact (0) and any contact (1).

Table 3.

Stepwise Binary Logistic Regression Analysis of Direct Contact With Journalists (N = 808).

	Model (R² change)
	Model 1 (.03)			Model 2 (.04)			Model 3 (.04)			Model 4 (.03)
Independent variable	b	eβ	p	b	eβ	p	b	eβ	p	b	eβ	p
Constant	0.13	1.14		−0.41	0.66	*	−1.82	0.16	***	−2.05	0.13	***
Disciplinary influence
Medical science	0.93	2.53	**	0.76	2.14	**	0.74	2.10	**	0.68	1.98	*
Humanities and social sciences	0.40	1.49	**	0.38	1.47	**	0.49	1.63	***	0.48	1.61	***
Reputational damage	0.01	1.01		0.00	1.00		0.00	1.00		0.00	0.99
Reputational gains	−0.06	0.95	**	−0.03	0.97		−0.06	0.95	**	−0.06	0.94	**
Organizational influence
Organizational structures
Research university				0.25	1.28	*	0.41	1.50	**	0.47	1.61	***
Private university				0.50	1.65		0.39	1.48		0.32	1.38
Small university				−0.31	0.73		−0.34	0.72		−0.44	0.64	*
Third-party funding				2.18	8.86	***	1.95	7.00	***	2.01	7.43	***
Organizational media orientation
Perceived need							0.05	1.05		0.05	1.05
Observed relevance to others							0.00	1.00		0.00	0.99
Adoption of concern							0.16	1.17	***	0.14	1.15	***
PR influence
Press clippings usage										0.05	1.05	**
PR expertise										0.01	1.01
PR activity										0.15	1.16	***
Nagelkerke’s R²	.03***			.07***			.11***			.14***

Note: Coding: no contact = 0, any contact = 1. b = unstandardized regression coefficient, eβ = exponential coefficient.

*p ≤ .10. **p ≤ .05. ***p ≤ .01.

All in all, our regression models fit the data well, but they do so to different extents. The explanatory factors explain 32% of the variance of respondents’ self-initiated professional press releases. By contrast, direct contact with journalists is less predictable, as indicated by Nagelkerke’s R² (.14). The lower explanatory power is certainly due to the fact that the variance is constrained by dichotomizing.

Single-item measures as used in our study commonly fail to show a statistically significant vicinity to normal distribution. Therefore, not all of our data adhere strictly to all standard statistical assumptions underlying regression modeling. In order to control for potential biases in our results, we retested them by using resampling methods (Wu, 1986). From the original data set, 1,000 bootstrap samples were taken, calculating 95% confidence intervals with bias-corrected bootstrapping. Bootstrapping yielded an equivalent result pattern to all models as shown in Tables 1 and 2.

The most striking result to emerge from the data is that the number of contacts with journalists is explained by various levels to almost equal extents, while self-initiated press releases are predicted mainly by the degree of PR influence. The explanatory power of structural conditions associated with a university’s legitimizational need is similar to its concern for organizational visibility in the media. In these blocks, many variables have no significant effect, if they are controlled by all other variables. Furthermore, the data yield disciplinary influences. We found a negative effect of reputational gains on contact with journalists, although we had expected a positive one (see Hypothesis 1). Furthermore, we found no effect of perceived reputational damage, the more so as this perception was relatively weak (M = 2.8, SD = 3.1).

Hypothesis 2 predicted an effect of the concern for organizational media visibility. This hypothesis is supported by the finding that this block explains a further 2% of the variance in press releases and 4% of the variance in contact with journalists. Controlled by all other variables, only the item measuring the adoption of this concern predicts both dependent variables. Some evidence emerges in support of our hypotheses regarding structural conditions (Hypotheses 3-6). Organizational structures account for 2% of variance in press releases and 4% of the variance in contact with journalists. From Table 2, we can see that the explanatory power of organizational structures (in Model 2) disappears in the presence of predictors of Models 3 and 4. This indicates that media orientation and PR influence tend to be tied to specific structural conditions. We accounted for structural conditions that might be connected to the need of a university to legitimize via mass media, and our results are consistent with most of the expected directions of influence. The proportion of third-party funding has a positive effect on media efforts (Hypothesis 6). Looking at Model 2, academics at private universities have more media contact than those at state and ecclesiastical universities (Hypothesis 4). As expected, scientists at small colleges have fewer direct contacts with journalists (Hypothesis 5), and scientists at research universities have more (Hypothesis 3). However, regarding the activity of press releases, our expectations are not supported by the data.

Our data strongly confirm Hypothesis 7, which proposed an influence of professional PR on scientists’ media efforts. The lowest influence derives from the perceived expertise of PR professionals, which was not to be expected anyway in the case of contact with journalists. The strongest effect evolves from PR activity in managing the ongoing demand for press releases. Understandably, this particularly affected scientists’ self-initiated press releases and, additionally, their direct contact with journalists. Furthermore, the extent of reading press clippings explained significant further variance of both dependent variables. Finally, our data confirm Hypothesis 8, that there is a reinforcing interplay between self-initiated professional press releases and direct contact with journalists (Pearson’s r = .287, p < .000).

Discussion and Conclusion

This study set out with the aim of assessing the importance of university organization in strengthening scientists’ media efforts. In contrast to more piecemeal previous research proposing organizational influence (Engwall, 2008; Jacobson et al., 2004; Kyvik, 2005; Peters, Heinrichs, et al., 2008; Rowe & Brass, 2008, 2011), we based our research on scientists at all universities of one country and equally considered scientists’ perceptions, PR department activities, and organizational structures. In our view, an up-to-date analysis of broad scope was called for by previous studies that reported general attitudinal changes on the part of both university managers and researchers (Engwall, 2008; Kallfass, 2008). As discussed, there are strong indications that ongoing new public management reforms have altered contextual conditions and even causal relationships of scientists’ media efforts. As a new governance of what are now entrepreneurial universities may have an effect, organizational influence on scientists’ media efforts needs reconsideration. The measurements in our study cover only a limited part of the story but provide suggestions and insights for further research on systematic grounds.

The most prominent influence evident in our models emanates from the PR departments, in support of Hypotheses 7 and 8. Pure PR activity is the most powerful predictor of media efforts. The more often PR professionals ask for news items, the stronger the effect of scientists complying with such demands, especially when it comes to researchers’ self-initiated and PR-assisted press releases. Through well-equipped and active press offices, universities are able to stabilize and professionalize the media contact of researchers. These processes are accompanied by the internal circulation of press clippings, but they are influenced less by the perceived expertise of PR professionals, so there probably is an effect of mere exposure to PR efforts. This applies in particular to small universities, where the average physical distance between PR and scientists is not substantial, and to universities of applied sciences. Additionally, the data indicate that scientists in research universities are contacted directly by journalists more often, so that they have better chances of getting in contact with the media without PR support. However, our data display a firm entrenchment of PR in scientists’ public communication for the most part. Scientists generally have a closer relationship with their university’s press office than with journalists. Thus, scientists at universities receiving high third-party funding are evidently more motivated to maintain contact with journalists, not least to achieve legitimacy in the eyes of sponsors. Furthermore, there are indications that the PR departments’ influence is roughly in line with a university’s structural need for media visibility. Structural conditions that we accounted for in order to represent a universities dependency on public legitimization, for example, third-party funding and a private funding body, clearly show some predictive power, but their statistical significance diminishes in models that include PR influence (see Table 2, Model 4). This suggests a close association of the measured phenomena.

As proposed above, given that the identification with organizational concerns for media visibility may develop in three stages (Hypothesis 2), in our models statistical relevance is only evident for the third of these stages. A scientist’s mere perception of his or her university’s need for legitimization in the media reveals little impact. The same applies to his or her observation of colleagues talking about their university’s media visibility. By contrast, a scientist’s internalization of this need clearly has a strong influence. Therefore, what actually matters is the internalization. Scientists who truly internalized the idea that a university should be visible in the media effectively show a higher frequency of media efforts (therefore supporting Hypothesis 2). This indicates, not surprisingly, that personal attitude has a closer relationship to actual behavior than the mere perception of an organizational need or the observation of colleagues. Still, what is decisive here is not the effect size but information on the paths that shape a scientist’s willingness to go public. In this case, it is the effect of the organization, which we measured by identification with organizational visibility.

Taken together, organizational factors revealed a lower explanatory power for contact with journalists than for self-initiated professional press releases. It seems quite plausible that organizational influence can more accurately predict those activities that are initiated exclusively within the university. Journalism, on the other hand, has its own logic about what to include and what to exclude and, therefore, its own criteria as to which scientists to contact and which to omit from the process (Kohring, 2005; Peters, 2008). Public relations specialize in anticipating this logic, exercising indirect influence without having control. With respect to the comparatively low prediction of contact with journalists, future studies should also include journalistic selection processes for science news (Badenschier & Wormer, 2012), a factor that is not covered in the present study.

As discussed above, we set out to investigate the role of a university in shaping scientists’ media efforts and considered the role of the academic discipline as a control. Our results suggest that both disciplinary and organizational factors influence the frequency of media efforts. In the models used, organizational factors yield greater higher explanatory power. We provided for only three clusters of disciplines in order to ensure the applicability of regression analysis. Accordingly, we found that medical scientists have closer ties with PR, while human and social scientists have strong relationships with journalists. Unmistakably, there are various connections between a university organization and the disciplines it hosts. Because universities perceive medical research as more susceptible to scandals, they are likely to take steps to reduce the direct media contact of the scientists in question. We further found that natural and engineering scientists have strong connections neither with the media nor with their PR department. These findings conform to the results of existing studies that indicate natural scientists have significantly less media contact than social scientists (Dunwoody, 1986; Jensen, 2011; Kyvik, 2005; Rowe & Brass, 2008). Clearly, a more detailed breakdown would be valuable for differentiating with respect to a discipline’s typical orientation toward the news media (e.g., Rödder & Schäfer, 2010).

Even more worthy of discussion are our findings on reputational effects. First of all, we found a negative effect of reputational gains on contact with journalists, whereas we expected a positive one (see Hypothesis 1). Still, there are some plausible explanations: Scientists who have relatively few contacts with journalists may be more critical than others in their observation of colleagues with high media prominence who are frequently invited to high-level conferences. Hence, perceived reputational gains may have no effect on one’s contact with journalists, but this contact probably shapes the critical perspective on popular scientists. Taken together, the effects of the interplay of scientific reputation and media prominence reveal more complexity than expected and may call for a more detailed breakdown on the peculiarity of academic norms in the various disciplines.

Our analysis of media contact paints a picture of scientists who participate regularly in the dissemination of research. These findings further support the concept of “feedback loops of media attention” (Peters, 2008, p. 137). Indeed, we detect a strong correlation between both dependent variables, pointing to the effect of reciprocal intensification. Media contact is associated with self-reinforcing processes in which scientists become “attractive addresses” for both journalists and PR professionals in search for publishable statements.

Our findings further support the notion that regularly visible scientists fulfill expectations of them as organizational members. Scholars compete not only for scientific reputation but also for institutional recognition. Communicating to the public and creating media visibility is one way to enhance one’s position within the university. But as long as media visibility is less connected with the role of a member of the scientific community, the extent of media efforts will probably remain moderate, at least compared with other actors such as entrepreneurs and politicians. However, in the face of growing challenges to universities, science communication is not being left to a small group of particularly active researchers but is also being shaped by a broad base of academics reacting to the needs and expectations of their university.

Limitations

Although this study improves our understanding of the organizational embedding of science communication, it has some limitations. First, our findings cannot be extrapolated to all academics, as we use a specific sample. In Germany, individual scientists’ freedom in research is a constitutional right, so they enjoy more pronounced autonomy than in many other countries. Additionally, there is no formal tenure track as in the United States, and there is an informal rule that the career of a professor must extend beyond one university. Therefore, organizational influence on media efforts of scientists in other countries may potentially be higher. The contrary may apply to the frequency of media efforts measured in our study. Compared to other studies based on information provided by scientists about their activities during a 3-year period (e.g., Dunwoody et al., 2009; Kyvik, 2005; Peters, Heinrichs, et al., 2008), we found a relatively higher frequency of media efforts, but this might result from the nature of our sample. Previous research (Dunwoody et al., 2009; Dunwoody & Scott, 1982; Jensen, 2011) showed that established researchers have greater media visibility than young scientists. Our respondents are, without exception, established researchers. Furthermore, scientists in very large universities and medical scientists are underrepresented, due to our selection of respondents. Extrafaculty research institutes, art colleges, hospitals, and industrial research are not covered in our sample. Therefore, the characteristics of our sample exacerbate general problems in the quantitative comparison of university systems in different countries. Another limitation relates to the omission of specific governance instruments targeting the increase of scientists’ media contact, such as incentive systems and media-related training.

Second, the focus of our study is on organizational influence. There are other possible paths that may play a role in conveying the notion that scientists should go public or in impelling media efforts. For instance, scientists may be influenced by a pronounced exposure to the political field or by an affiliation to higher education policy institutions. The study at hand provides little data for checking on these and other possible influences. Therefore, this study neither negates nor evaluates the importance of individual differences and discipline-dependent effects, but it does provide systematic evidence of the importance of organizational affiliation.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was conducted as part of the research initiative “New Governance of Science. Research on the Relationship Between Science, Politics and Society” (2008-2012), funded by the Federal Ministry of Education and Research (Grant No. 01UZ0902).

Notes

Author Biographies

Frank Marcinkowski holds a chair in communications at the University of Muenster, Germany. He has published widely in the fields of political communication, media theory, and science communication.

Matthias Kohring is a professor of communication at the Department of Media and Communication Studies, University of Mannheim, Germany. His research focuses on trust in media, the public sphere, and science communication.

Silke Fürst is a doctoral candidate in communication sciences and research assistant at the University of Fribourg, Switzerland. Her research focuses on media effects and science communication.

Andres Friedrichsmeier, PhD, is a researcher at the Department of Communication, University of Muenster, Germany. His work focuses on the mediatization of higher education and science communication.

References

Badenschier

Wormer

(2012). Issue selection in science journalism: Towards a special theory of news values for science news? In Rödder

Franzen

Weingart

(Eds.), The sciences’ media connection—public communication and its repercussions. Sociology of the sciences yearbook (pp. 59-85). Dordrecht, Netherlands: Springer.

Bentley

Kyvik

(2011). Academic staff and public communication: A survey of popular science publishing across 13 countries. Public Understanding of Science, 20, 48-63.

Bishop

(2006). Research. Theory, Culture & Society, 23, 570-571.

Bodmer

W. F.

(1985). The public understanding of science: Report of a Royal Society ad hoc group. London:The Royal Society.

Borchelt

R. E.

(2008). Public relations in science: Managing the trust portfolio. In Bucchi

Trench

(Eds.), Handbook of public communication of science and technology (pp. 147-157). New York, NY: Routledge.

Brass

Rowe

(2009). Knowledge limited: Public communication, risk and university media policy. Continuum: Journal of Media & Cultural Studies, 23, 53-76.

Braun

Merrien

F.-X.

(Eds.). (1999). Towards a new model of governance for universities? A comparative view. London, England: Kingsley.

Clark

B. R.

(1983). The higher education system: Academic organization in cross-national perspective. Berkeley: University of California Press.

Clark

B. R.

(1998). Creating entrepreneurial universities: Organizational pathways of transformation. Oxford, England: Pergamon.

10.

Davison

P. W.

(1983). The third-person effect in communication. Public Opinion Quarterly, 47, 1-15.

11.

Décieux

J. P.

Heinz

Jacob

(2011). Online-Erhebungen mit EFS Survey. Ein Überblick über grundlegende Funktionen des Erhebungsprogramms [Online Surveys with EFS Survey. An Overview on Basic Functions of the Survey Plattform]. Munich, Germany: GRIN.

12.

Delanty

(2001). Challenging knowledge: The university in the knowledge society. Buckingham, England: Society for Research into Higher Education.

13.

DiMaggio

P. J.

Powell

W. W.

(1983). The iron cage revisited: Institutional isomorphism and collective rationality in organizational fields. American Sociological Review, 48, 147-160.

14.

Dunwoody

(1986). The scientist as source. In Friedman

S. M.

Dunwoody

Rogers

C. L.

(Eds.), Scientists and journalists: Reporting science as news (pp. 3-16). New York, NY: Free Press.

15.

Dunwoody

Brossard

Dudo

(2009). Socialization or rewards? Predicting U.S. scientist-media interactions. Journalism & Mass Communication Quarterly, 86, 299-314.

16.

Dunwoody

Ryan

(1983). Public information persons as mediators between scientists and journalists. Journalism Quarterly, 60, 647-656.

17.

Dunwoody

Ryan

(1985). Scientific barriers to the popularization of science in the mass media. Journal of Communication, 35, 26-42.

18.

Dunwoody

Scott

B. T.

(1982). Scientists as mass media sources. Journalism Quarterly, 59, 52-59.

19.

Ehlers

, & Committee on Science. (Eds.). (1998). Unlocking our future: Toward a new national science policy. Retrieved from http://www.gpo.gov/fdsys/pkg/GPO-CPRT-105hprt105-b/pdf/GPO-CPRT-105hprt105-b.pdf

20.

Engwall

(2008). Minerva and the media: Universities protecting and promoting themselves. In Mazza

Quattrone

Riccaboni

(Eds.), European universities in transition: Issues, models and cases (pp. 31-48). Cheltenham, England: Edward Elgar.

21.

Friedrichsmeier

(2012). Die unterstellten Wirkungen der universitären Steuerungsinstrumente. Zur hochschulischen Dauerreform und den Möglichkeiten ihrer Entschleunigung [Insinuated Modes of Action of University Controls. On Chances to Decelerate a Perpetual Management Reform]. Muenster, Germany: LIT.

22.

Gascoigne

Metcalfe

(1997). Incentives and impediments to scientists communicating through the media. Science Communication, 18, 265-282.

23.

Gregory

Miller

(2000). Science in public: Communication, culture, and credibility. Cambridge, England: Basic Books.

24.

Gunther

A. C.

Storey

D. J.

(2003). The influence of presumed influence. Journal of Communication, 53, 199-215.

25.

Holmström

(2005). Reframing public relations: The evolution of a reflective paradigm for organizational legitimization. Public Relations Review, 31, 497-504.

26.

Jacobson

Butterill

Goering

(2004). Organizational factors that influence university-based researchers’ engagement in knowledge transfer activities. Science Communication, 25, 246-259.

27.

Jansen

(Ed.). (2007). New forms of governance in research organizations: Disciplinary approaches, interfaces and integration. Dordrecht, Netherlands: Springer.

28.

Jensen

(2011). A statistical picture of popularization activities and their evolutions in France. Public Understanding of Science, 20, 26-36.

29.

Kallfass

(2008). Public Relations von Wissenschaftseinrichtungen—explorative Studie in Deutschland, Frankreich und Großbritannien [Public Relations of Scientific Organizations - Explorative Study in Germany, France, and the U.K.]. In Peters

H. P.

(Ed.), Medienorientierung biomedizinischer Forscher im internationalen Vergleich [Media Orientation of Biomedical Researchers in International Comparison] (pp. 101-175). Retrieved from http://wwwzb1.fz-juelich.de/contentenrichment/onlinepublikationen/Gesundheit_18.pdf

30.

Kehm

B. M.

Lanzendorf

(2007). The impacts of university management on academic work: Reform experiences in Austria and Germany. Management Revue, 18, 153-173.

31.

Kohring

(2005). Wissenschaftsjournalismus: Forschungsüberblick und Theorieentwurf [Science journalism: Research overview and theory design]. Konstanz, Germany: UVK.

32.

Kohring

Marcinkowski

Lindner

Karis

(2013). Media orientation of German university decision makers and the executive influence of public relations. Public Relations Review. Retrieved from http://dx.doi.org/10.1016/j.pubrev.2013.01.002

33.

Krücken

Meyer

(2006). Turning the university into an organizational actor. In Drori

Meyer

Hwang

(Eds.), Globalization and organization: World society and organizational change (pp. 241-257). Oxford, England: Oxford University Press.

34.

Kyvik

(2005). Popular science publishing and contributions to public discourse among university faculty. Science Communication, 26, 288-311.

35.

Martín-Sempere

M. J.

Garzón-García

Rey-Rocha

(2008). Scientists’ motivation to communicate science and technology to the public: Surveying participants at the Madrid Science Fair. Public Understanding of Science, 17, 349-367.

36.

Miller

(2001). Public understanding of science at the crossroads. Public Understanding of Science, 10, 115-120.

37.

Pearson

(2001). The participation of scientists in public understanding of science activities: The policy and practice of the U.K. Research Councils. Public Understanding of Science, 10, 121-137.

38.

Peters

H. P.

(2008). Scientists as public experts. In Bucchi

Trench

(Eds.), Handbook of public communication of science and technology (pp. 131-146). New York, NY: Routledge.

39.

Peters

H. P.

Brossard

Cheveigné

S. d.

Dunwoody

Kallfass

Miller

Tsuchida

(2008). Science-media interface. It’s time to reconsider. Science Communication, 30, 266-276.

40.

Peters

H. P.

Heinrichs

Jung

Kallfass

Petersen

(2008). Medialization of science as a prerequisite of its legitimization and political relevance. In Cheng

Claessens

Gascoigne

Metcalfe

Schiele

Shunke

(Eds.), Communicating science in social contexts: New models, new practices (pp. 71-92). Dordrecht, Netherlands: Springer.

41.

Poliakoff

Webb

(2007). What factors predict scientists’ intentions to participate in public engagement of science activities? Science Communication, 29, 242-263.

42.

Rödder

(2012). The ambivalence of visible scientists. In Rödder

Franzen

Weingart

(Eds.), The sciences’ media connection—public communication and its repercussions. Sociology of the sciences yearbook (pp. 155-178). Dordrecht, Netherlands: Springer.

43.

Rödder

Schäfer

M. S.

(2010). Repercussion and resistance: An empirical study on the interrelation between science and mass media. Communications, 35, 249-267.

44.

Rogers

C. L.

(1986). The practitioner in the middle. In Friedman

S. M.

Dunwoody

Rogers

C. L.

(Eds.), Scientists and journalists: Reporting science as news (pp. 42-54). New York, NY: Free Press.

45.

Rowe

Brass

(2008). The uses of academic knowledge: The university in the media. Media, Culture & Society, 30, 677-698.

46.

Rowe

Brass

(2011). “We take academic freedom very seriously.” How university media offices manage academic public communication. International Journal of Media & Cultural Politics, 7, 3-20.

47.

Scott

W. R.

(1994). Institutions and organizations: Toward a theoretical synthesis. In Scott

W. R.

Meyer

J. W.

(Eds.), Institutional environments and organizations: Structural complexity and individualism (pp. 55-80). Thousand Oaks, CA: Sage.

48.

Smith

K. C.

Singer

R. F.

Kromm

E. E.

(2010). Getting cancer research into the news: A communication case study centered on one U.S. comprehensive cancer center. Science Communication, 32, 202-231.

49.

Statistisches Bundesamt. (Ed.). (2012). Fachserie 11/Reihe 4.3.2. Bildung und Kultur: Monetäre hochschulstatistische Kennzahlen: Einzelne Hochschulen 2010 (Sonderveröffentlichung) [Subject-Matter Series 11/Series 4.3.2. Education and Culture. Monetary Key Figures in Higher Education Statistics]. Wiesbaden, Germany: Statistisches Bundesamt.

50.

Tsfati

Cohen

Gunther

A. C.

(2011). The influence of presumed media influence on news about science and scientists. Science Communication, 33, 143-166.

51.

Weingart

(2001). Die Stunde der Wahrheit? Zum Verhältnis der Wissenschaft zu Politik, Wirtschaft und Medien in der Wissensgesellschaft [The moment of truth: The relationship of science with politics, economics, and media in the knowledge society] (1st ed.). Weilerswist, Germany: Velbrück.

52.

Weingart

Pansegrau

(1999). Reputation in science and prominence in the media: The Goldhagen debate. Public Understanding of Science, 8, 1-16.

53.

Wernick

(2006). Rebranding Harvard. Theory, Culture & Society, 23, 566-567.

54.

Winter

(2004). Public communication of science and technology: German and European perspectives. Science Communication, 25, 288-293.

55.

C. F. J.

(1986). Jackknife, bootstrap and other resampling plans in regression analysis. Annals of Statistics, 14, 1261-1350.