You are on page 1of 27

Structural Strain in Science: Organizational Context,

Career Stage, Discipline, and Role Composition*

David R. Johnson, University of Nevada Reno


Brandon Vaidyanathan, University of Notre Dame
Elaine Howard Ecklund, Rice University

This article examines the relationship between structural strain (the imbalance
between actual and preferred conditions of work) and anomie in science (the absence of
opportunities to achieve recognition). Using data from a nationally representative survey
of physicists and biologists in the United Kingdom (N = 1,604), we test competing
hypotheses about the occupational factors that produce structural strain. We find that
structural strain is influenced by organizational context and career stage, but not in the
manner existing theory suggests. We elaborate existing theoretical frameworks by show-
ing that role composition mediates the effects of organizational context and career stage.

Introduction
Scholars argue that professional occupations are associated with highly
autonomous conditions of work that produce job satisfaction. The autonomy of
professionals in law, medicine, science, and engineering, for example, is predi-
cated on abstract expertise that only members of a profession are able to evalu-
ate, meaning these expert laborers wield a great deal of control over the
definition, execution, and conditions of their work (Freidson 2001). Neverthe-
less, situated as they are in organizations (Randle 1996), professional workers
are subject to many sources of structural strain, an imbalance between actual
and preferred conditions of work.
Consider academic science. Structural strain can alter how scientists under-
stand their roles, including tensions between: access to funding and excess of
competition (Stephan 2012); the pursuit of traditional scientific inquiry and
demands for commercially relevant research (Johnson 2017); or work and fam-
ily (Ecklund and Lincoln 2016; Thompson and Bunderson 2001). Although
academic science has historically experienced structural changes corresponding
to economic prosperity in a given national context (Geiger 1999), the core
changes characterizing contemporary science involve adaptation to greater scar-
city of resources and opportunity (Hermanowicz 2009).
Little research, however, has directly examined how the distribution of
structural strain may vary across organizational contexts, career stages, or

Sociological Inquiry, Vol. xx, No. x, 2017, 1–27


© 2017 Alpha Kappa Delta: The International Sociology Honor Society
DOI: 10.1111/soin.12176
2 DAVID R. JOHNSON ET AL.

professional specialties.1 There is research on scientists, however, which exami-


nes the concept of anomie, a cultural concept having to do with perceived
normlessness within social systems. Anomie is the result of a divide between
the achievement aspirations of scientists and the academic profession’s capacity
to recognize scientists for their contributions (Hagstrom 1964; Hermanowicz
2009). This suggests a sense of breakdown in the collective order of science, in
part brought about by changes in structure that fail to provide meaningful
bases by which to understand roles, expectations, and aspirations. Key to this
argument is the notion that structural conditions of work related to the research
role enable and constrain scientific achievement. In contrast to anomie theory,
which focuses on the normative order of science, strain theory examines the
structural distribution of legitimate opportunities to achieve institutional goals.
This article builds on existing research by examining the distribution of
structural strain in science across organizational contexts, career stage, and dis-
ciplines. We use data from a study of UK physicists and biologists that exami-
nes the social dimensions of scientific work. Here, we examine mismatches in
actual and preferred allocation of effort. Such an analysis is important for a
number of reasons. First, existing research on anomie in science is based on
qualitative studies that focus on single disciplines. Using a nationally represen-
tative sample, we provide new insights into anomie and strain theories by
assessing the segments of science in which strain, a source of anomie, is found.
Second, examination of strain in science provides an empirical assessment of
the conditions of work in higher education. The application of strain theory to
science itself illuminates inequality by identifying groups of scientists who are
deprived of the opportunity to achieve their preferred conditions of work.
Third, our analysis is important because we elaborate existing theory by
explaining how role composition influences structural strain.
In contrast to what existing theory predicts, we find that scientists in non-
elite contexts—where expectations are thought to be lower and career adapta-
tion is thought to be highest—experience more strain than their peers in more
elite contexts. Also inconsistent with theory, we find that independent of orga-
nizational context, the gap between actual and preferred conditions of work
grows wide for scientists by mid-career and is never resolved. The findings
point to the value of a multidimensional approach to the study of control and
autonomy in professional occupations.
Theoretical Background
Durkheim uses the concept of anomie to describe a condition of deregula-
tion in society, where the norms that usually govern expectations for behavior
are not present. Anomie results in a mismatch between the actual circumstances
in which individuals find themselves and their preferences for their future.
STRUCTURAL STRAIN IN SCIENCE 3

According to Durkheim, “No living being can be happy or even exist unless
his needs are sufficiently proportioned to his means” (Durkheim [1897] 1951,
246). He first introduces the term in The Division of Labor in Society to
describe how industrialization created a division of labor that threatened social
cohesion. In Suicide, Durkheim reveals the relationship between a society’s
ability to regulate the aspirations and behaviors of its members and its suicide
rate, showing that anomic suicide is the result of a lack of social solidarity.
Merton (1938) draws upon the concept of anomie in his theory of social
order, arguing that anomie is the result of an imbalance between cultural struc-
ture and social structure. According to Merton, the dominant cultural goal in
the United States is material wealth, while education and hard work comprise
the legitimate means by which individuals may realize that goal. Merton argues
that the problem with this type of society is that the means for success are not
universally distributed across the social structure. Both Durkheim and Merton
conceptualize anomie as normlessness, but they differ in their views of what
causes anomie. Durkheim argues that rapid change in society leads to anomie,
whereas Merton argues that anomie is caused by excessive emphasis on the
goals of wealth and success in the United States.
Merton also distinguishes between anomie and strain, yet these distinct
theoretical components are not often distinguished in research because his
usage of the concept of anomie is inconsistent (Levine 1985). As Featherstone
and Deflem (2003) show, Merton presents two distinct theories: a theory of
anomie, the deinstitutionalization of norms that results from an imbalance
between emphasis on cultural goals and institutional means (Merton 1938; p.
673; Merton 1968, p. 189); and strain theory, which argues that people are
more likely to pursue illegitimate means to goal attainment when they are
blocked from access to the institutionalized means to these goals (Merton
1938, p. 679; 1968, p. 211). As Merton argues (1968, pp. 216–217), “social
structure strains the cultural values, making action in accord with them readily
possible for those occupying certain statuses within the society and difficult or
impossible for others.” Both anomie and strain explain social disorganization.
Whereas anomie theory argues that culturally induced pressure to achieve insti-
tutional goals leads to a sense of normlessness or a breakdown in the collective
order, strain theory is a structural argument that focuses on the distribution of
legitimate means to be successful.
Anomie and Strain in Science
The sociology of science has profited from anomie theory in particular.
Based on interviews with mathematicians, Hagstrom (1964) introduced the
notion of anomie in science, a contribution important for illustrating the con-
ceptual connections between Durkheim and Merton’s formulations and the
4 DAVID R. JOHNSON ET AL.

system of science. Some scholars have argued that anomie is a cause of


research misconduct in science (Hackett 1994; Zuckerman 1988), although little
research has verified this claim (Braxton 1993). Empirical assessments of
anomie in science as a condition languished until Hermanowicz’s (2009) longi-
tudinal qualitative study of U.S. physicists.
Merton argues that excessive emphasis on the goals of wealth and success
in the United States is a source of general anomie in society, while science
places an equally powerful emphasis on the institutional goal of advancing
knowledge. In the United Kingdom, for example, universally high institutional
expectations for contributions to knowledge are communicated to scientists
through accountability regimes such as the Research Assessment Exercise,
which ties funding for academic departments to contributions to knowledge.2
Globally, scientists undergo significant professional socialization during gradu-
ate school where they are exposed to core institutional expectations of science
such as intensive hours of work and success through original contributions to
knowledge. Because recognition for original research is institutionally expected
of scientists, anomie is likely to emerge in structural conditions associated with
an absence of opportunities to achieve recognition, and where cultural condi-
tions that foster commitment to professional norms are low (Hermanowicz
2009).
Here, we focus on the structural conditions of the workplace that constrain
opportunities to achieve recognition. Specifically, our analytic focus is on the
disjuncture between the actual and preferred conditions of work that logically
enable or constrain scientists’ ability to fulfill their expectations for research
output. As Schuster and Finkelstein (2006, p. 87) argue, “when organizational
expectations diverge from those activities that individual faculty members value
most, they are likely to experience considerable strain. . ..” This disjuncture
between actual and preferred conditions of work is indicative of Hackett’s
(1990) emphasis on tensions between autonomy and accountability and
between producing research and educating students. In the United Kingdom,
such tensions in the structural arrangements of work that cause strain are often
associated with New Public Management (Chandler, Barry, and Clark 2002),
which stresses greater accountability of faculty through measurable standards of
performance. In what follows, we draw on existing work to formulate different
conditions under which strain in scientists’ ability to fulfill expectations for
research output may arise.
Discipline
Consensus, an aspect of codification (Merton and Zuckerman 1973), refers
to the extent to which scientists agree on what problems are most important for
research, which theories and methods are appropriate, and what constitutes a
STRUCTURAL STRAIN IN SCIENCE 5

successful career (Hargens and Kelly-Wilson 1994). High consensus disciplines


exhibit clear and rigid definitions of success, while low consensus disciplines
possess ambiguous and varied definitions of achievement (Cole 1983). It is
more difficult to satisfy expectations for success in high consensus disciplines,
such as physics, because success is so rigidly defined around specific types of
contributions to knowledge, whereas fulfilling expectations for success in low
consensus fields such as sociology or psychology is easier because there are
multiple definitions of success to which scholars can tailor their self-definitions.
Anomie and strain should therefore vary by discipline because expecta-
tions for success in research are more rigidly defined in some disciplines rela-
tive to others. Anomie should be highest in highly codified disciplines such as
physics because definitions of success are so absolute and difficult to fulfill at
all points of a career. Biology, by comparison, is generally associated with
intermediate codification. Relative to physics, there is lower consensus about
problem choice, suitable theory, and methods of research, meaning there are
more opaque and permissive definitions of success to which scientists can
attach themselves (Braxton and Hargens 1996). The lower degree of consensus
is observable by the structural differentiation of biology departments, which are
much more fragmented than physics. We therefore hypothesize that:
H1: Incidence of structural strain will be higher among physicists than among biologists.

Organizational Context
The failure of organizations to establish an environment where a minimum
set of worker preferences can be met is critical to how employees experience
their careers (Hodson 1999). Organizational context is particularly important in
science because where scientists work makes a significant difference in how
much they publish (Allison and Long 1990), meaning anomie and strain should
vary by organizational context because of varying emphases on research and
the structural properties of universities that enable and constrain scientific
achievement. A widespread embrace of research now characterizes university
systems (Morphew 2009), with universities seeking funding, prestige, and bet-
ter faculty and students. In the context of scientific careers, this means a greater
stress on research productivity, even at universities where teaching is more
common than research.
While anomie is found throughout science (Hackett 1990; Hermanowicz
2011), scholars disagree about where it may be more pronounced. Hermanow-
icz (2011), for example, argues that the presence of resources and pressure
encourages anomie less than the lack of it. In his study, he finds that the “pov-
erty” of non-elite contexts protects scientists from anomie because scientists
perceive limited opportunity for recognition and advancement and consequently
6 DAVID R. JOHNSON ET AL.

develop limited aspirations (Hermanowicz 2011). Kohn and Schooler (1973)


work on the reciprocal effects of job conditions and personality suggests that
habituation to work of high complexity (demanding and challenging profes-
sions like science) facilitates a preference for intellectual flexibility and the
capacity for self-direction, meaning elite scientists’ desires to be highly produc-
tive will always outweigh their circumstances, no matter how abundant their
resources (see also Mortimer and Lorence 1994). Accordingly, we hypothesize
that:
H2a: Incidence of structural strain will be higher among elite scientists than among non-elite
scientists.

Hagstrom (1964), by contrast, posits anomie as a condition of the marginal


academic. Hagstrom based this claim on interviews with mathematicians who
work in arcane areas of knowledge and thus are more likely to have their work
go unnoticed. While Hagstrom’s argument was about areas of knowledge, the
general principles that motivate his hypothesis are marginality and integration.
Because scientific visibility is predicated in part on publication productivity,
scientists in non-elite contexts may be less integrated than their peers in elite
contexts because there is a lower emphasis on research and fewer resources
with which to succeed as a researcher. A competing hypothesis about the role
of organizational context is therefore:
H2b: Incidence of structural strain will be higher among non-elite scientists than among elite
scientists.

Career Stage
Anomie and strain vary by career stage because the expectations for
careers vary over time. A possibility that one might naturally assume is that
strain is higher in earlier career stages as scientists experience the pressure that
accompanies the pursuit of tenure, while scientists in later career stages conduct
their work in the relative security of a permanent position. Hermanowicz’s
(2009) study suggests the effects of career stage are context specific. Among
elite scientists, the disjuncture between present and future expectations is high-
est in late career stages (Hermanowicz 2009). During early- and mid-career
stages, elite scientists experience cumulative advantage and opportunities, mani-
fest in accelerated publication productivity and heightened expectations. As
elites approached the end of their careers, however, they experience anomie
because greater achievement is circumvented by a lack of time, lower capaci-
ties, and a sense that their careers did not progress as they expected.
Anomie emerges earlier among non-elite scientists, primarily because the
structural conditions of their work contexts offer fewer professional
STRUCTURAL STRAIN IN SCIENCE 7

opportunities to sustain research achievement and dampen motivation to


achieve (Hermanowicz 2009). Some scientists may begin to reject the emphasis
on research achievement during graduate or postdoctoral training, but these
career stages are characterized by strong socialization processes and ritual
ordeals that foster commitment to research as the preeminent role of the scien-
tist (Corcoran and Clark 1984). Elite scientists end their careers experiencing a
“reversal of fortunes,” while “communitarian scientists”—those at institutions
that place a higher premium on teaching than research—approach the end of
their careers at peace because they altered their achievement expectations as
they progressed through prior career stages (Hermanowicz 2009).
Hermanowicz, focusing on anomie, emphasizes the dynamics of scientists’
expectations for their careers over time. Much of what contributes to altered
expectations, his argument implies, is structural strain; he argues that the orga-
nizational contexts or resources for research are central to heightening and
dampening scientists’ expectations. We therefore hypothesize that:
H3a: Incidence of structural strain among non-elite scientists will be higher among those in
early to mid-career stages relative to non-elites in late career stages.

H3b: Incidence of structural strain among elite scientists will be higher among scientists in
late career stages relative to elites in early or mid-career stages.

Role Composition and Research Funding


Discipline, organizational context, and career stage are group-level condi-
tions that are thought to influence the incidence of strain and anomie, but the
scientists that comprise such groups such as “mid-career biologists” or “elite
physicists” are not homogenous. In the non-elite sphere of science, there are
scientists whose careers resemble those of the elite; similarly in the elite
spheres of science, there are individuals whose careers gravitate toward those
of the non-elite (Hermanowicz 1998). Tenured or permanent elite scientists
whose research performance no longer meets group expectations turn to, or are
forced into, teaching and service roles. And while non-elite scientists spend
more time in teaching and service roles than their elite counterparts, research
funding enables high-performing scientists in non-elite contexts to conform
more closely to institutional expectations for research.
These types of patterns point to the influence of role composition and
funding on structural strain. By role composition, we refer to the amount of
time scientists spend on research, teaching undergraduates, graduate training,
and other tasks. The allocation of time to these tasks matters because it condi-
tions scientists’ ability to live up to institutional expectations for achievement
through commitment to the research role. Even at the “post-1992” universities
in the UK—institutions that acquired university status in 1992 and which
8 DAVID R. JOHNSON ET AL.

traditionally have emphasized teaching over research—promotion criteria


emphasize both teaching and research (Parker 2008).3 Simply put, no matter
what context, scientists who are unable to spend enough time on, or who lack
money for, research are likely to experience anomie because of structural barri-
ers to achievement. In contrast to discipline, organizational context, and career
stage, these are individual-level attributes that vary within groups and over
time. Because the amount of money and time a scientist can spend on research
is so critical to achievement, we expect that:
H4: Role composition and research funding will explain the effects of discipline, organiza-
tional context, and career stage on the incidence of structural strain.

Data, Measures, and Analytic Strategy


We test these hypotheses using data from the British subset of an eight-
nation study of physicists and biologists working in universities and institutes.4
The broader study examines the social context of scientific work, including
gender, family, ethics, and religion, and other themes relevant to the careers of
scientists around the globe. Although they cannot be generalized to all of
science, physics and biology are ideal disciplines for the study of these social
features of science due to their differing levels of women and their intersections
with religion and ethics. As core science disciplines, they are also ideal for the
study of structural strain. For complete sampling details of the broader study,
see Ecklund et al. (2016).
The present study focuses on the United Kingdom. We created a nation-
ally representative sampling frame of physicists and biologists in the UK
through a two-stage stratified sampling process. In the first stage, we identified
academic departments through three strategies: (1) author affiliations on publi-
cations in physics and biology journals found on Thomson Reuters Web of
Science database, an online database with citation information for more than
12,000 scientific journals; (2) published university ranking reports (e.g., The
Guardian’s physics department rankings for 2011 and 2012, the 2008 Research
Assessment Exercise, and The Complete University Guide’s ranking of biology
departments); and (3) guidance from in-country experts. This sampling frame
was then stratified by discipline (biology and physics), UK region (England,
Scotland, Wales, Northern Ireland), and the university’s academic reputation
(elite versus non-elite [see below for details on this distinction]). Through this
process, we identified a total of of 320 biology and physics departments in the
UK, and selected a stratified random sample of 74 departments (24 elite Phy-
sics departments, 15 non-elite Physics, 18 elite Biology, and 17 non-elite Biol-
ogy). We then visited the Web site of each department to confirm eligibility of
each department.5 For second stage strata, we included all scientists from each
STRUCTURAL STRAIN IN SCIENCE 9

of the eligible sampled departments, stratified by gender and career stage, total-
ing 3,372 physicists and biologists. We oversampled women because they are
underrepresented in science.
The survey was administered from September 19, 2013, to October 15,
2013, by GfK, an international survey research firm based in London, England.
The scientists were contacted in advance with an invitation to participate in the
study, which included a description of the study and a £5 pre-incentive. Fol-
lowing this letter, scientists received an e-mail containing a link to the online
survey, another description of the study, and a letter of support encouraging
participation from the study’s International Natural Science Advisory Board. E-
mail and telephone reminders followed, and scientists were given the option to
complete the survey over the phone.6
Of the 3,372 scientists in our initial sample, 1,604 completed the survey.
Using the American Association for Public Opinion Research’s definition num-
ber 3 for response rates, our UK survey obtained a 50 percent response rate.7
We account for departmental clustering and selection probabilities in our analy-
sis through sampling weights and data stratification. This minimizes any bias
caused by the sampling design and the deficiencies of the sampling frame. t-
Tests conducted in the non-response analysis confirm that our sample matches
the sampling frame on known characteristics. The results of our analysis thus
reflect a representative cross section of the population of academic physicists
and biologists in the UK.8 Although in the remainder of the article we use the
term “scientists” to refer to the population we studied, this is merely for gram-
matical simplicity; we cannot generalize to scientists from other disciplines
such as chemistry and social sciences.
For our analytic sample in this article, we only consider those scientists
who possess a PhD (N = 1,084). We exclude postgraduate students (the equiva-
lent of master’s and doctoral students in the United States), because our mea-
sure of anomie is based on the discrepancy between actual and preferred tasks
that scientists perform. Postgraduate students do not engage in graduate and
undergraduate instruction, and because these activities are key components of
the scientific role (and may contribute to strain vis-a-vis research activities), we
exclude these scientists.9 Our sample thus comprises only scientists in the fol-
lowing positions: Postdoctoral Fellows, Research Fellows, Lecturers, Senior
Lecturers, Readers, and Professors.10 Note that in the UK, the lecturer position
is quite different from the position of the same title in the United States. In the
UK, lecturers are engaged in both research and teaching, much like any profes-
sor in the United States, whereas in the United States lecturers are exclusively
engaged in instructional tasks and not research. Reader is an academic rank
above senior lecturer and below professor. Like these other positions, readers
are engaged in both research and teaching. In our final models, 114 cases were
10 DAVID R. JOHNSON ET AL.

dropped due to missing data across variables, giving us a sample size of


N = 970.
Dependent Variable
Because research output is a central expectation and criterion of success
for scientists, we measure structural strain using two survey items that capture
the difference between scientists’ preferred and actual allocation of time to
research.11 These measures were part of a module in which scientists were
asked a series of questions related to how much they “usually work.” The sur-
vey question on actual time allocation asked:
We would like to know the percentage of your work time devoted to undergraduate instruc-
tional activities, postgraduate instructional activities, research activities and all other activities.
If you are not sure, give your best estimate.

The survey was programmed to exclude invalid answers (e.g., fractions,


negative numbers, and numbers exceeding 100). Responses to the four response
categories—”Undergraduate instructional activities,” “Postgraduate instructional
activities,” “Research activities,” and “All other activities”—were stored as sep-
arate variables. For the preferred time allocation question, we asked:
We are interested in what would be your ideal work arrangement as a scientist. Thinking
about the four categories just discussed, please report the percentage of your work time you
would prefer to devote to the following activities.

We then calculated the differences between the percentage of work hours


a scientist prefers to devote to research and the percentage actually devoted to
research, generating a continuous dependent variable ranging from 50 to 90.
Negative values indicate that the scientist would prefer to dedicate a smaller
percentage of time to research, while higher values indicate a preference to
allocate a higher percentage of work hours to research. Negative values suggest
something like structural conduciveness, where a work environment provides
an excess of hours for conducting research, which is analytically distinct from
structural strain as we conceptualize it. As our construct of structural strain is
focused on cases where scientists desire to dedicate more time to research than
they are actually able to, we recode all negative cases to zero.12 Using this
measure, we find that 63 percent of UK scientists experience some structural
strain.
This measure has a high degree of construct validity because the objective
realities of work in which scientists find themselves and the achievement
expectations they embrace are two of the core concepts of the strain and
anomie theoretical framework. Indeed, Durkheim (1951) emphasized the divide
between the realities of everyday situations (actual allocation of effort) and
STRUCTURAL STRAIN IN SCIENCE 11

expectations for the present and future (desired allocation of effort). Moreover,
the notion of time is only important in our measure to the extent that it
expresses a preference about what scientists would prefer to be doing, which is
presumably driven by the reward system of science.
A second reason this measure has a strong ability to capture the meaning
of strain is tied to the emphasis on deprivation from legitimate means of suc-
cess in Merton’s strain theory. The mismatch between actual and preferred con-
ditions of work captured in our measure is a structural measure that expresses a
scientist’s desire to change the condition of her work, meaning one may feel
deprived of what she perceives to be a legitimate opportunity to fulfill her
achievement expectations when it comes to research.
Our measure offers a methodological advance in the study of anomie and
strain relative to the two existing quantitative studies that model this concept.
Hargens and Kelly-Wilson (1994), who use anomie to predict “disciplinary dis-
content,” simply infer anomie from discipline, suggesting that math and sociol-
ogy are anomic fields. Braxton (1993) employs a theoretical framework that
incorporates anomie theory, but labels and operationalizes his anomie construct
as “alienation.” These approaches partially capture notions of anomie, but they
bear less logical connection to the theoretical components of anomie and strain
than the present study.
Explanatory Variables
Elite Status. To classify the organizational status of departments as elite
or non-elite, we triangulated three sources of information: research
productivity, published rankings, and regional insiders (i.e., prominent scientists
in each region). In terms of research productivity, we used Web of Science. As
part of our broader sampling procedure, we randomly sampled biology and
physics articles published between 2001 and 2011 that had departments based
in the UK. We then extracted department data to create a list of unique
organizations for each discipline, and created a measure summarizing the
number of publications per department. This number provides a proxy for the
relative level of engagement in research of a given department, with elite
departments being research-intensive and non-elite departments exhibiting
lower levels of publication productivity. Yet, even departments that publish
frequently may not be considered elite, for example, because the articles appear
in non-top-tier journals or because they are rarely cited. A continuous measure
could also be problematic because smaller departments that have a lower total
article output, but are nevertheless recognized as elite, would be misclassified.
For these reasons, we also relied on published rankings and in-country experts.
We assume that the stronger the reputation according to published rank-
ings, the more elite the organization. We utilized the 2008 Research
12 DAVID R. JOHNSON ET AL.

Assessment Exercise, which is conducted by four higher education funding


bodies in the UK who create “quality profiles” by which they rank depart-
ments.13 Because it is difficult to evaluate departmental quality from afar, and
because neither research productivity nor published rankings capture every
characteristic that might indicate an elite organization, we consulted with three
distinguished British scientists and a science policy expert, insiders who have
extensive knowledge of the UK science infrastructure.
Three principles underscored our classification process. First, if there was
a consensus as to the elite status of an organization among the insiders, we
generally assigned the status they recommended, regardless of the indication of
the other two criteria. Second, we were strongly inclined to classify organiza-
tions at the very top of the list ordered by publication count as elite, absent
compelling reason to doubt their elite status (e.g., ranked poorly by published
rankings or specified as non-elite by insiders). With organizations in descend-
ing order of publication count, we used published rankings and insider perspec-
tives to determine a rough threshold above which most organizations were elite
and below which most organizations were non-elite.14 Above this threshold, we
were inclined to consider organizations elite if indicated as such by either pub-
lished rankings or insiders. Below this threshold, we considered organizations
as elite only if indicated as such by both published rankings and insiders.
Combining these sources, we classified departments as elite (1) or non-
elite (0).

Career Stage. We conceive of career stage as comprising early, middle,


and late career phases. Early-career-stage scientists, who comprise 46 percent
of our weighted sample, completed their PhDs between 2003 and 2013.
Scientists in mid-career comprise those who completed their PhD more than
10 years ago but less than twenty-five years ago (roughly the 1988–2003
period), and account for 38 percent of our sample. Scientists in late career
received their PhD more than 35 years ago (before 1988, comprising 16
percent of our sample). These definitions match empirical precedent
(Hermanowicz 2009).
Discipline. Discipline is simply determined by whether scientists were
sampled from physics or biology departments (Physics = 0, Biology = 1).
Role Composition. We include three variables that measure work hours
per week that scientists spent on undergraduate teaching, postgraduate
teaching, and other non-research activities.
Funding. We include a variable measuring a scientist’s self-report of
funding in the past 3 years relative to other researchers in the UK (no funding,
below average funding, average funding, above average funding).
STRUCTURAL STRAIN IN SCIENCE 13

Control Variables. We include controls in our model for gender


(Female = 1), income (a 15-category variable ranging from income <£10,000
to income >£200,001), and publications (an 8-category variable ranging from 0
to more than 200). We also control for work-to-family conflict (measuring how
frequently the respondent has not had enough time for family because of work
obligations), and family-to-work conflict (measuring how frequently the
respondent has not had enough time for work because of family obligations),
which allows us to maintain a focus on strain exclusively within science by
controlling for pressures related to other life-realms.
Analytic Strategy. As we measure structural strain as a continuous
variable, we use ordinary least squares regression to estimate its relationship to
the various factors hypothesized above. Our models test the above hypotheses
sequentially. To test each hypothesis, our strategy is to first estimate a bivariate
model (examining simply the relationship between the independent and
dependent variable) and subsequently to add controls that we expect to affect
the relationship. Such a strategy allows us to differentiate between direct and
indirect effects of the hypothesized predictors of structural strain. Because we
have a single dependent variable, and the key independent variables from
initial hypotheses are used as controls in later hypotheses, we estimate all
models progressively in a single table below.
In all our analyses, we incorporate capped sample weights to account for
the complex sampling frame and stratification. We ensured that there are no
prohibitively high correlations between variables (see Table 2 below). To test
for multicollinearity, we used Stata’s vif command. A mean variance inflation
factor of 2.04 in our final models (including interaction effects) indicates that
multicollinearity is not a problem in our models (Allison 1999).
Results
Table 1 presents descriptive statistics of all variables used in our analysis.
Because elite institutions and biology departments were disproportionately more
likely to be sampled, we present both weighted and unweighted descriptive
statistics. Table 2 presents a correlation matrix of the variables used in the anal-
ysis.
Table 3 shows how the preferred and actual proportion of work time allo-
cated to research varies by discipline, organizational context, and career stage.
It also shows the distribution of our dependent variable, structural strain, along
these variables. We find little difference between physicists and biologists in
our dependent variable. Scientists in non-elite departments exhibit greater struc-
tural strain, indicating a preference to allocate more of their work hours to
research than their counterparts in elite departments. This difference is statisti-
cally significant (p = .002). Further, compared to early-career scientists,
14 DAVID R. JOHNSON ET AL.

Table 1
Descriptive Statistics (Weighted and Unweighted) for Variables Used in
Analysis (N = 970)

Weighted Unweighted Std.


Mean Mean Dev. Min. Max.

Research Strain 13.78 14.14 14.53 .00 90.00


Discipline = Biology .60 .55 .50 .00 1.00
Gender = Female .26 .41 .49 .00 1.00
Income 5.78 5.97 2.62 2.00 15.00
Publications 2.86 2.96 1.60 .00 7.00
Work-to-Family Conflict 3.38 3.40 1.08 1.00 5.00
Family-to-Work Conflict 2.35 2.38 1.10 1.00 5.00
Elite Department .61 .63 .48 .00 1.00
Early Career .40 .37 .48 .00 1.00
Mid-Career .40 .41 .49 .00 1.00
Late Career .20 .21 .41 .00 1.00
Funding 2.76 2.81 .96 1.00 4.00
Work hours spent on 9.66 9.56 9.87 .00 75.00
undergraduate teaching
Work hours spent on 7.22 7.55 5.77 .00 32.00
graduate teaching
Work hours spent on other 8.18 8.53 7.97 .00 50.40
non-research activities

middle- and late-career scientists exhibit greater structural strain. There is a sta-
tistically significant difference (p < .001) between early-career scientists and
either of the others, but not between middle- and late-career scientists. (For
tests of statistical significance, see bivariate coefficients of these variables in
Table 4.)
Table 4 presents results from a weighted ordinary least squares regression
of structural strain on discipline, organizational context, and career stage. Mod-
els 1 and 2 aim to assess support for Hypothesis 1, that is, that incidence of
strain is higher among physicists than among biologists. Model 1 examines the
differences between Physics and Biology departments in the distribution of
structural strain, and Model 2 introduces controls. While the magnitude of dif-
ference increases when we introduce controls in Model 2, this effect is not sta-
tistically significant. There is no statistically significant difference between
Table 2
Correlation Matrix
Work-to- Family- Career Career Career
Family to-Work Stage Stage Stage Undergraduate Graduate Other
Strain Biology Female Income Publication Conflict Conflict Elite 1 2 3 Funding Teaching Teaching Activities

Strain 1
Biology .04 1
Female .04 .18 1
Income .07 .02 .22 1
Publications .04 .21 .20 .37 1
Work- .16 .05 .03 .02 .01 1
to-Family
Conflict
Family- .07 .03 .01 .01 .12 .18 1
to-Work
Conflict
Elite .10 .08 .06 .08 .12 .04 .04 1
Career .23 .02 .19 .41 .29 .03 .05 .05 1
Stage 1
Career .15 .02 .02 .08 .15 .04 .12 .04 .65 1
Stage 2
Career .09 .05 .20 .39 .17 .01 .08 .02 .40 .44 1
Stage 3
Funding .11 .03 .06 .26 .36 .02 .07 .14 .07 .06 .01 1
Undergraduate .34 .05 .06 .01 .03 .12 .10 .18 .27 .16 .13 .25 1
Teaching
Graduate .26 .00 .03 .18 .26 .15 .12 .00 .23 .12 .12 .18 .10 1
Teaching
STRUCTURAL STRAIN IN SCIENCE

Other .37 .02 .14 .30 .13 .09 .06 .01 .27 .09 .22 .00 .05 .12 1
Activities
15
16

Table 3
Structural Strain by Discipline, Organizational Context, and Career Stage (N = 970)
Early Late
Biology Physics Elite Non-elite career Mid-Career career
DAVID R. JOHNSON ET AL.

Preferred Actual Preferred Actual Preferred Actual Preferred Actual Preferred Actual Preferred Actual Preferred Actual

% Work 58.6 47.9 59.6 47.8 55.0 41.5 61.7 52.0 68.2 63.2 54.6 39.6 49.6 34.0
time
spent
on
research
Structural 13.6 14.1 12.5 15.7 9.8 16.7 16.0
strain
(%)

Note: Because cases where structural strain is less than zero are recoded to zero, mean values for structural strain do
not amount to the simple difference of means between preferred and actual proportions.
Table 4
Weighted Ordinary Least Squares Regression of Structural Strain on Discipline, Organizational Context, and Career Stage

Model (1) Model (2) Model (3) Model (4) Model (5) Model (6) Model (7)

Biology .485 .612 (1.104) .804 (1.111) .973 .788


(1.087) (1.089) (.897)
Female .100 (1.162) .0976 (1.151) .306 .848
(1.144) (.924)
Income .193 (.213) .195 (.212) .232 .357*
(.203) (.169)
Publications .00279 (.382) .0833 (.381) .282 .277
(.376) (.352)
Work-to-Family 2.284*** 2.245*** 2.131*** .666
Conflict (.505) (.500) (.492) (.423)
Family-to-Work .638 (.550) .675 (.548) .741 1.084*
Conflict (.550) (.434)
Elite context 3.166** 3.193** 1.823 .273
(1.092) (1.075) (1.671) (1.228)
Career stage
Mid-career 6.870*** 7.862*** 2.417
(1.259) (1.867) (1.593)
Late career 6.186*** 6.786*** .130
(1.368) (2.019) (1.916)
Mid-career 1.501 1.427
STRUCTURAL STRAIN IN SCIENCE

X Elite (2.434) (1.940)


17
Table 4 18
(continued)

Model (1) Model (2) Model (3) Model (4) Model (5) Model (6) Model (7)

Late career .534 (2.605) .224


X Elite (2.296)
Funding .701
(.620)
DAVID R. JOHNSON ET AL.

Undergraduate .418***
teaching hours (.0607)
Postgraduate .577***
teaching hours (.114)
Other .652***
non-research (.0930)
work hours
Constant 14.07*** 3.824 15.71*** 5.687* 9.786*** 4.366 .0334
(.794) (2.410) (.816) (2.511) (.888) (2.542) (2.391)
N 970 970 970 970 970 970 970
R-square .000 .034 .011 .045 .051 .092 .339
Adj. .001 .028 .010 .038 .049 .081 .329
R-square

Notes: Taylor-linearized standard errors in parentheses.


*p < .05, **p < .01, ***p < .001
STRUCTURAL STRAIN IN SCIENCE 19

physicists and biologists in their discrepancies between preferred and actual


proportions of work time dedicated to research. The control variables suggest
that greater income, greater family-to-work conflict, and greater work-to-family
conflict are all associated with a preference to allocate more time to research.
When we add the effects of elite organizational contexts and career stages in
later models, the magnitude of these changes remains small and statistically
insignificant. While our models suggest that organizational context and career
stage might suppress the effect of discipline, there is no statistically significant
difference in structural strain between UK physicists and biologists, meaning
we find no support for Hypothesis 1.
Models 3 and 4 examine the relationship between structural strain and elite
context, to assess support for either Hypothesis 2a or 2b, that is, whether scien-
tists in elite contexts experience significantly more strain than their counterparts
in non-elite contexts, or vice versa. Model 3 shows a negative coefficient for
elite, which should be interpreted in light of our findings in Table 3: non-elite
scientists experience more strain, and prefer to allocate more of their time to
research than elite scientists do. Once we introduce controls, we see a very
slight increase in the negative coefficient for elite. Entering the control vari-
ables sequentially suggests that this change is due to the lower income and
higher work-to-family conflict of non-elites. Overall, we find support for
Hypothesis 2b: Net of other factors, non-elites experience greater structural
strain than elites.
In Models 5 and 6, we investigate the relationship between structural strain
and career stage, to assess support for Hypotheses 3a and 3b, that is, whether
incidence of structural strain among non-elite scientists is higher among those
in early to mid-career stages relative to non-elites in late career stages, or
higher among those in late career stages. Model 5, consistent with Table 3,
shows that mid-career and late-career scientists want to allocate significantly
more of their time to conducting research in comparison with early-career sci-
entists. The magnitude and significance of these coefficients increase somewhat
when we include gender, income, publications, funding, work–family conflict,
discipline, and elite status in Model 6. In ancillary analysis (not shown), we
find the coefficient for elite status to be initially significant at p < .05. Once we
introduce the interaction effects between elite status and career stage, however,
the independent effect of elite status disappears. The magnitude and signifi-
cance of career stage coefficients persist unchanged; however, the interaction
coefficients are not significant. In sum, while scientists in middle and late
career stages experience greater strain than early-career scientists, these effects
do not vary by elite status—they are identical in elite and non-elite depart-
ments. Thus, we cannot confirm either Hypothesis 3a or 3b.
20 DAVID R. JOHNSON ET AL.

Table 5
Preferred and Actual Allocation of Work Hours by Career Stage (N = 970)

Early career Mid-Career Late career

Preferred Actual Preferred Actual Preferred Actual

Total work hours 39.19 46.97 39.63 47.85 39.96 48.47


Work hours spent on 5.04 6.16 8.10 11.61 9.89 12.75
undergraduate
teaching
Work hours spent 6.43 5.72 8.86 8.18 8.76 8.27
on postgraduate
teaching
Work hours 3.45 5.81 4.56 9.14 5.83 10.97
spent on other
work activities
Work hours 32.06 29.27 26.33 18.92 24.0 16.50
spent on
research

Finally in Model 7, we assess whether role composition might explain the


effects of discipline, organizational context, and career stage on the incidence
of structural strain. Here, we introduce measures of funding and actual work
hours spent on teaching and other non-research activities. We find that the
magnitude of elite status diminishes considerably, as do the effect sizes and sta-
tistical significance of the career stage variables. While one’s perceptions of
funding relative to other scientists has no net effect on structural strain, more
hours spent on teaching and non-research activities are significantly associated
with a preference for allocating greater proportions of time to research. These
variables mediate the relationship between career stage and structural strain.
Thus, we find partial support for Hypothesis 4.
Table 5 takes a closer look at how role composition mediates the effect of
career stage on research strain. We see that across career stages, scientists pre-
fer to spend fewer hours working in general than they currently do, and fewer
hours on undergraduate teaching and other activities in particular. (Early and
mid-career scientists, however, would prefer to spend more time teaching grad-
uate students than they currently do.) On average, mid-career and late-career
STRUCTURAL STRAIN IN SCIENCE 21

scientists spend more time on such activities than early-career scientists do—
the role composition for scientists in later career stages entails more time spent
teaching and on administrative duties. Consequently, mid- and late-career scien-
tists would prefer to dedicate more of their work hours to doing research than
they currently do.
Discussion
These findings contribute to knowledge of scientific work in the UK and
have implications for the study of professions more broadly. Research on Bri-
tish scientists tends to focus on outcomes of scientific work—such as commer-
cialization of research (Haeussler and Colyvas 2011), science communication
with the public (Pearson 2001; Johnson et al. 2016), and scientists’ involve-
ment in policy (Waterton 2005)—rather than conditions of scientific work. We
build on earlier studies of British scientists (Gaston 1975) and more recent
essays (Morris and Rip 2006; Pritchard 1998) that consider workplace condi-
tions. Specifically, the conditions of structural strain we identify illuminate
threats to job satisfaction and professional autonomy in UK academic science
careers.
Our findings advance the theory of anomie in science by systematically
measuring the incidence of structural strain using a nationally representative
sample of scientists. We found that the majority (63%) of UK scientists experi-
ence some structural strain; that is, they prefer to spend a greater proportion of
their work time on research than they currently do. We developed hypotheses
about conditions that give rise to strain from research on anomie, and found
support for some, but not all of the propositions. In the UK, organizational con-
text and career stage matter, but not in the manner proposed for U.S. scientists.
We also elaborate the existing theoretical frameworks by showing that role
composition mediates the effects of organizational context and career stage.
Below, we consider our main findings and their implications.
This study provides the first test of structural strain that compares disci-
plinary differences, finding that there are no significant differences between
physicists and biologists. Based on codification theory, we expected physicists
to experience higher levels of strain because definitions of success are so strin-
gently tied to transformative discoveries in physics, whereas biology—charac-
terized by medium consensus—offers more definitions of success to which
scientists can attach their sense of self. Physicists do report a desire to spend
more time on research than biologists, but the differences are negligible and
not significant. It could be that while biology offers more varied definitions of
success, these definitions do not spillover into tasks unrelated to research. It
could also be that disciplinary differences in strain only exist between low con-
sensus (such as social science disciplines) and high consensus disciplines.
22 DAVID R. JOHNSON ET AL.

There are significant career stage differences in structural strain, but it is


surprising that the patterns observed did not vary by elite and non-elite contexts
as previous work on anomie led us to anticipate (Hermanowicz 2009). Strain
among elite scientists is at its highest toward the end of their careers, but that
peak emerges in mid-career and sustains into late career stages. And elite scien-
tists are no different than non-elites in this respect. Indeed, mid-career and late-
career scientists experience the most structural strain in our sample. On one
hand, it is unsurprising that early-career-stage scientists experience the least
structural strain because, during the initial years following completion of the
PhD, scientists assume postdoctoral appointments in which the majority of
one’s time is allocated to research. Although these appointments hold the
uncertainty of subsequent attainment of a permanent position, they offer the
luxury of time for research unencumbered by requirements (teaching, service,
etc.) that await scientists as they become lecturers. On the other hand, just over
half (58%) of the scientists in early career stage are postdoctoral scientists.15
Thus, for those who attain positions as lecturers, the transition from doctoral
and postdoctoral training to independent positions in the UK is quite smooth in
terms of the attainment of preferred conditions of work. Yet, as British scien-
tists enter mid- and late-career, regardless of organizational context, the dis-
tance between actual and preferred allocation of time to research grows wide
and is never resolved, even as they scale back their preferences to conduct
research.
The broader labor market ecology in the UK—specifically the number of
research-intensive jobs available—is worth considering to interpret how these
results may relate to patterns Hermanowicz observed in the United States. In
labor markets where job opportunities are more abundant than in the UK, the
typical move from a more elite university to a less or non-elite university (Burris
2004) is a more difficult adjustment because expectations for a permanent posi-
tion in a research-intensive department may be perceived as reasonable. Hence,
attaining a position in a department beneath one’s research expectations makes
the incidence of anomie and strain highest in early career stage for non-elite sci-
entists. If opportunities for academic appointments are lower in the UK, it is
possible that the concern for attaining a permanent position takes priority over
the nature of the tasks one would ultimately perform (e.g., primarily research
versus primarily instruction). In these scenarios, structural strain may occur only
in later phases because scientists begin their careers content rather than beneath
their own expectations. Future researchers could test this proposition by compar-
ing variation in the incidence of structural strain across countries characterized
by distinctive labor market conditions for academic science careers.
Finally, our study elaborates the existing anomie framework by consider-
ing the influence of role composition and research funding. The findings show
STRUCTURAL STRAIN IN SCIENCE 23

that the effects of career stage and organizational context on structural strain
are mediated by the amount of time scientists allocate to other tasks. Even
though scientists’ preference to work more hours increases over the course of
the career, this increase is outpaced by the expansion of hours scientists actu-
ally work. What is more, the actual amount of time scientists spend on research
decreases as they enter mid- and late-career stages, while commitments to
teaching and other tasks increase during these periods. How these tasks are
allocated ultimately matters more than the organizational context or career stage
in which a scientist is situated. Yet it is more difficult to achieve one’s pre-
ferred working conditions at non-elite universities primarily because the teach-
ing expectations are so high. The anomie experienced by elite scientists in late
career stages may result in part from the difficulty of living up to such height-
ened group expectations for success, but they nevertheless remain structurally
inhibited. Even in an organizational context of accumulative advantage, achiev-
ing one’s preferred conditions of work gets more challenging over time.
The limitations of our study provide important avenues for future research.
Although our work is unable to operationalize constructs related to cultural
conditions such as meaning, meaninglessness, and identification with the insti-
tutional goals, it nevertheless provides a measure of strain in science that is
empirically more precise than previous scholarship, which simply infers anomie
from discipline (Hargens and Kelly-Wilson 1994). Future studies would benefit
from direct measures of institutional expectations and scientists’ perceptions of
them. Further, we cannot rule out the possibility that the incidence of structural
strain and anomie vary across national contexts of science as national science
infrastructures vary in terms of the distribution of resources and career opportu-
nities and the culture of scientific work. It is thus probable that conditions of
structural strain vary across nations as a result.
The theory of strain and anomie in science could fruitfully be applied
more broadly as a theory of strain and anomie in professions such as medicine,
law, and architecture. In particular, this work suggests that rather than focusing
on broad dynamics that erode control over conditions of professional work
(Ritzer and Walczak 1988), the study of threats to autonomy should assume a
multidimensional approach including factors like organizational context, career
stage, and specialty. Like science, all professions are characterized by strong
socialization processes that encourage achievement aspirations and commitment
to institutional goals. Here, we focused on the elite and non-elite characteristics
of organizational context. Professions more broadly are similarly characterized
by internal differentiation (Becker 1982), meaning the organizational contexts
in which professional careers are carried out—like science—are characterized
by structural differences that enable and constrain achievement. Like non-elite
scientists who aspire to achieve through research but find themselves
24 DAVID R. JOHNSON ET AL.

encumbered by obligations to teach, one can imagine lawyers who aspire to


argue before the House of Lords but find themselves in a large law firm filing
divorce claims, or physicians who aspire to heal the sick but find themselves
awash in paperwork that interferes with time with patients. With respect to
career stage, the presence of status passages and social differentiation in law
and medicine suggest that conditions of strain could be similarly explored by
career stage and across elite and non-elite organizational contexts. A corollary
of disciplinary differences could be examined by accounting for differences in
strain and anomie by areas of practice.
Scarcity of resources, pressure to publish, and accountability regimes in
higher education will continue to shape the actual and preferred conditions of
scientific work for the foreseeable future, making structural strain a critical
topic for sociological investigation. We have sought to bring new insights into
the study of structural strain in science through disciplinary comparison, analy-
sis of new concepts such as role composition, more direct measures of strain,
and by looking beyond the United States. Continued research in this area, we
hope, will point to ways to close the gap between the preferred and actual con-
ditions of scientific work.

ENDNOTES

*Please direct correspondence to David R. Johnson, University of Nevada Reno, 1664 N. Vir-
ginia St, Reno, NV 89557, USA; e-mail: drj@unr.edu
David R. Johnson (drj@unr.edu) is an Assistant Professor of Higher Education at the
University of Nevada, Reno, and author of the forthcoming book, Commercialism and Conflict in
Academic Science: A Fractured Profession (Johns Hopkins University Press).
Brandon Vaidyanathan is a Public Policy Fellow at the Center for Ethics and Culture at the
University of Notre Dame.
Elaine Howard Ecklund is the Herbert S. Autrey Chair in Social Sciences and Professor of
Sociology at Rice University.
The authors wish to thank Scott V. Savage and Richard Williams for comments on an earlier
version of this manuscript.
Data collection for this study was funded by the Templeton World Charity Foundation, Elaine
Howard Ecklund PI, and Kirstin R. W. Matthews, and Steven W. Lewis Co-PIs, grant #0033/
AB14.
1
To be clear, there are clearly literatures on role strain caused by interinstitutional conflict
between spheres such as work and family or science and capitalism.
2
The Research Assessment Exercise is now entitled the Research Excellence Framework.
3
Teaching colleges and polytechnics did not have university status in the UK until the Further
and Higher Education Act of 1992. These institutions are much more focused on teaching than
older universities, but some have departments that can compete in research terms. The non-elite
subset of our sample includes six departments located at post-1992 universities.
4
In some countries, academic research takes place in research institutes that are not necessar-
ily affiliated with universities. All of the scientists in our UK subset are located at universities.
STRUCTURAL STRAIN IN SCIENCE 25

5
A department was identified as ineligible if (1) there was a lack of contact information for
scientists; (2) the department could not be found online; and (3) the department was not a biology
or physics department. In the event a department was ineligible, a new one was randomly sampled.
6
Sixteen scientists completed the survey by phone.
7
See http://www.aapor.org/AAPOR_Main/media/publications/Standard-Definitions20169thed
itionfinal.pdf
8
In the absence of a census of British physicists and biologists—no such census currently
exists—we cannot accurately verify the extent to which our sample is representative of the popula-
tion of physicists and biologists in the UK. However, we took several steps to minimize the poten-
tial for differences between our sample and the population of academic physicists and biologists,
including multistage random selection and stratification by key variables such as gender, rank, and
institutional status as we mentioned earlier. It is possible that our sampling method may have
missed some marginal segments of the population of academic biologists and physicists in the UK
that would require oversampling for representativeness, such as scientists in the most esoteric sub-
fields within each discipline. There is no theoretical reason this should have any implications for
our findings.
9
Most postdoctoral scientists and research fellows in the UK do not teach undergraduates, but
they are very involved in the training of graduate students and other administrative tasks—key roles
of relevance to structural strain.
10
The research questions and sampling strategy of the broader study focus on scientists in
graduate, postdoctoral, and the equivalent of tenure-track and tenured positions. The professional
position measure on the survey included an “other” option that allowed respondents to specify their
position. In the event a respondent indicated a rank for which we did not sample, such as adjunct
or technician, their case was dropped from the sample.
11
These measures are derived from the 2004 National Study of Postsecondary Faculty.
12
We thank an anonymous reviewer for this recommendation.
13
For more information about RAE quality profiles, see http://www.rae.ac.uk/results/intro.aspx
14
In the case of physics, for example, the total number of randomly selected publications per
department ranged from 1 to 128. We employed a threshold of 50 randomly selected articles. Only
14 departments were situated above this level.
15
To determine whether the career stage effect is a function of the number of postdoctoral sci-
entists situated in career stage 1, we ran an ancillary analysis in which we moved all lecturers and
readers into career stage 2 and restricted career stage 1 to scientists in fellowships, and found simi-
lar results. Results of this analysis are available on request.

REFERENCES

Allison, Paul. 1999. Multiple Regression: A Primer. Thousand Oaks, CA: Pine Forge Press.
Allison, Paul and J. Scott Long. 1990. “Departmental Effects on Scientific Productivity.” American
Sociological Review 55(4):469–78.
Becker, Howard. 1982. Art Worlds. Berkeley, CA: University of California Press.
Braxton, John. 1993. “Deviancy from the Norms of Science: The Effects of Anomie and Alienation
in the Academic Profession.” Research in Higher Education 34(2):213–28.
Braxton, John and Lowell Hargens. 1996. “Variation Among Academic Disciplines: Analytical
Frameworks and Research.” Pp. 1–46 in Higher Education: Handbook of Theory and
Research, Vol. 11., edited by John Smart. New York: Agathon.
26 DAVID R. JOHNSON ET AL.

Burris, Val. 2004. “The Academic Caste System: Prestige Hierarchies in PhD Exchange Networks.”
American Sociological Review 69(2):239–64.
Chandler, John, Jim Barry, and Heather Clark. 2002. “Stressing Academe: The Wear and Tear of
the New Public Management.” Human Relations 55(9):1051–69.
Cole, Jonathan. 1983. “The Hierarchy of the Sciences?” American Journal of Sociology 89:111–39.
Corcoran, Mary and Shirley Clark. 1984. “Professional Socialization and Contemporary Career
Attitudes of Three Faculty Generations.” Research in Higher Education 20(2):131–53.
Durkheim, Emile. [1897] 1951. Suicide. Translated by John A. Spalding and George Simpson. New
York: Free Press.
Ecklund, Elaine Howard and Anne Lincoln. 2016. Failing Families, Failing Science: Work-family
Conflict in Academic Science. New York: New York University Press.
Ecklund, Elaine Howard., David R. Johnson, Christopher P. Scheitle, Kirstin R.W. Matthews and
Steven W. Lewis. 2016. “Religion Among Scientists in International Context: A New Study
of Scientists in Eight Regions.” Socius 2:1–9.
Featherstone, Richard and Mathieu Deflem. 2003. “Anomie and Strain: Context and Consequences
of Merton’s Two Theories.” Sociological Inquiry 73(4):471–89.
Freidson, Eliot. 2001. Professionalism: The Third Logic. Hoboken, NJ: Wiley.
Gaston, Jerry. 1975. “Autonomy in the Research Role and Participation in Departmental Decision-
Making.” British Journal of Sociology 26(2):227–41.
Geiger, Roger. 1999. “The Ten Generations of American Higher Education.” Pp. 38–69 in Higher
Education in the Twenty-First Century, edited by Robert Berdahl, Philip G. Altbach and
Patricia J. Gumport. Baltimore, MD: Johns Hopkins University Press.
Hackett, Edward. 1990. “Science as a Vocation in the 1990s: The Changing Organizational Culture
of Academic Science.” Journal of Higher Education 60(3):241–79.
———. 1994. “A Social Control Perspective on Scientific Misconduct.” Journal of Higher
Education 65(3):242–60.
Haeussler, Carolin and Jeannette Colyvas. 2011. “Breaking the Ivory Tower: Academic
Entrepreneurship in the Life Sciences in UK and Germany.” Research Policy 40(1):41–54.
Hagstrom, Warren. 1964. “Anomy in Scientific Communities.” Social Problems 12:186–95.
Hargens, Lowell and Lisa Kelly-Wilson. 1994. “Determinants of Disciplinary Discontent.” Social
Forces 72(4):1177–95.
Hermanowicz, Joseph C. 1998. The Stars Are Not Enough: Scientists-Their Passions and
Professions. Chicago, IL: University of Chicago Press.
———. 2009. Lives in Science: How Institutions Affect Academic Careers. Chicago, IL: University
of Chicago Press.
———. 2011. The American Academic Profession: Transformation in Contemporary Higher
Education. Baltimorem, MD: Johns Hopkins University Press.
Hodson, Randy. 1999. “Organizational Anomie and Worker Consent.” Work and Occupations 26
(3):292–323.
Johnson, David R. 2017. Commercialism and Conflict in Academic Science: A Fractured
Profession. Baltimore, MD: Johns Hopkins University Press.
Johnson, David R., Elaine Howard Ecklund, Di Di and Kirstin R.W. Matthews. 2016. “Responding
to Richard: Celebrity and (Mis)representation of Science.” Public Understanding of Science.
10.1177/0963662516673501
Kohn, Melvin L. and Carmi Schooler. 1973. “Occupational Experience and Psychological
Functioning: An Assessment of Reciprocal Effects.” American Sociological Review 38:97–
118.
Levine, Donald. 1985. The Flight from Ambiguity: Essays in Social and Cultural Theory. Chicago,
IL: University of Chicago Press.
STRUCTURAL STRAIN IN SCIENCE 27

Merton, Robert K. 1938. “Social Structure and Anomie.” American Sociological Review 3:672–82.
———. 1968. Social Theory and Social Structure. New York: The Free Press.
Merton, Robert K. and Harriet Zuckerman. 1973. “Age, Aging, and Age Structure in Science.” Pp.
497–559 in The Sociology of Science: Theoretical and Empirical Investigations, edited by
Norman Storer. Chicago, IL: University of Chicago Press.
Morphew, Christopher. 2009. “Conceptualizing Change in the Institutional Diversity of US
Colleges and Universities.” Journal of Higher Education 80:243–69.
Morris, Norma and Arie Rip. 2006. “Scientists’ Coping Strategies in an Evolving Research System:
The Case of Life Scientists in the UK.” Science and Public Policy 33(4):253–63.
Mortimer, Jeylan T. and Jon Lorence. 1994. “Social Psychology of Work.” Pp 497–523 in
Sociological Perspectives on Social Psychology, edited by Karen S. Cook, Gary Alan Fine
and James S. House. Boston, MA: Allyn and Bacon.
Parker, Jonathan. 2008. “Comparing Research and Teaching in University Promotion Criteria.”
Higher Education Quarterly 62(3):237–51.
Pearson, Gillian. 2001. “The Participation of Scientists in Public Understanding of Science
Activities: The Policy and Practice of the UK Research Councils.” Public Understanding of
Science 10(1):121–37.
Pritchard, Rosalind. 1998. “Academic Freedom and Autonomy in the United Kingdom and
Germany.” Minerva 36(2):101–24.
Randle, Keith. 1996. The White-Coated Worker: Professional Autonomy in a Period of Change.
Work, Employment, and Society. 10(4):737–53.
Ritzer, George and David Walczak. 1988. “Rationalization and the Deprofessionalization of
Physicians.” Social Forces 67(1):1–22.
Schuster, Jack and Martin Finkelstein. 2006. The American Faculty. Baltimore, MD: Hopkins
University Press.
Stephan, Paula. 2012. How Economics Shapes Science. Cambridgem, MA: Harvard University
Press.
Thompson, Jeffrey and J. Stuart Bunderson. 2001. “Work-Nonwork Conflict and the Phenomenology
of Time: Beyond the Balance Metaphor.” Work and Occupations 28(1):17–39.
Waterton, Claire. 2005. “Scientists’ Conceptions of the Boundaries between Their Own Research
and Policy.” Science and Public Policy 32(6):435–44.
Zuckerman, Harriet. 1988. “The Sociology of Science.” Pp. 511–74 in Handbook of Sociology,
edited by Neil Smelser. Newbury Park, CA: Sage.

Related Interests