EVALUATING PUBLIC DISCOURSE
IN NEWSPAPER OPINION ARTICLES:
VALUES-FRAMING AND INTEGRATIVE
COMPLEXITY IN SUBSTANCE
AND HEALTH POLICY ISSUES
By Lindsay H. Hoffman and Michael D. Slater
The exchange of opinions is an important component of participatory
democracies, and newspaper forum pages have been hailed as a conduit
for such discussion. This research explored Schwartz's value framework
to characterize lay and journalist frames in a national sample of local
newspapers. A content analysis was conducted on newspaper forum
articles on health policy issues—notably, alcohol, tobacco, illegal drugs,
and crime—for the presence and complexity of value frames. Significant
differences in values were found by article type and topic. Values typically associated with liberalism were also predictive of greater integrative complexity.
Normative questions about the quality, content, and form of public discourse have been at the center of political and social theories for
centuries. However, research on such discourse typically focuses on
archival studies of political speeches or statements, or the professional
reportage of news.' The forum pages of newspapers provide a unique
opportunity to study opinion expression of the citizenry, as well as of
professional journalists across the United States. That is, such pages
contain content—opinions that have arguably been reasonably
attained—that is not ordinarily archived. As communication scholars,
we have a unique obligation to assess the nature and quality of that discourse.
In this study, we seek to characterize and compare lay and journalist opinion expression regarding substance use and health policy issues,
using a unique nationally representative sample of local newspapers.
In particular, we propose that research on human values^ provides a
viable theoretical framework for characterizing both lay and journalist
opinion frames. We also argue that policy-relevant discourse can be
assessed with respect not only to value stance, but to the complexity of
opinions expressed.'
]&MC Quarterly
Vol. 84, No. 1
Spring 2007
58-74
©2007AE/MC
58
Lindsay H. Hoffman will be an assistant professor at the University of Delaware beginning fall 2007. Michael D. Slater is a professor at Ohio State University. This research
was supported by grant AA10377 to the second author from the National Institute on
Alcohol Abuse and Alcoholism. The authors would like to acknowledge graduate students Wendy Maxian, Texas Tech University, and jason B. Reineke, Ohio State
University, for their contributions to coding and analysis.
louRNALisM & MASS
COMMUNICATION
QUARTERLY
Public Discourse within Newspaper Forum Pages. Forum
pages—which include letters to the editor, editorials, and opinion
columns—provide a historically important place in public discourse.
Some scholars have argued that the opinions expressed in these pages
serve as "a place where democracy blossoms because regular citizens are
allowed a voice of their own."'* Although there are other forums for this
type of expression—such as television talk shows, radio call-in programs, or Internet group discussions—newspaper forum pages are generally not limited by a specialized topic and reach large numbers of community members.^
Most literature on forum pages focuses on the professional
norms and practices of journalists.^ Little research has actually examined the content of those opinions. In fact, Perrin claimed that no studies to date have examined the general tone of letters to the editor over
a period of time.'' Those who have analyzed such content generally
have focused on purely descriptive accounts." Yet this content gives
communication scholars a window through which to explore a vast
array of theoretical possibilities. Moreover, the availability of newspaper
opinion pieces makes it feasible to sample them in such a way as to provide a reasonable national representation of public discourse in these
pages.
Framing and Opinion Content in Newspapers. Much of the literature on framing focuses not on lay discourse, but on news or elite discourse.' But joumalists and elites are arguably not the only ones competent enough to strategically frame issues. Citizens organize their
thoughts about issues through relevant discourse and—if they choose to
write a letter to the editor, for example—may reconfigure applicable
information (or "reframe") information themselves.'" The scholarly
emphasis on journalistic and elite perspectives has neglected the ways in
which individuals frame arguments in public discourse. Moreover, framing is a central concept in evaluating the quality of citizens' performance
in such public processes."
Framing and Values. In newspapers, writers often need to get their
ideas across succinctly and clearly in limited space. Underlying the framing of such writings is a heuristic that encourages writers to refer to—or
"frame"—their argument based upon values. Framing capitalizes on the
fact that audiences hold distinct values and strive toward coherent
understandings of an issue.'^
Values serve as the underlying mechanisms of attitudes and can
dictate which attitudes people consider to be personally important."
Values also transcend specific situations and differ from attitudes in their
generality," can be ranked hierarchically by relative importance,'^ and do
not decay.'* People often use values to organize their positions on
issues," and values have been found to determine virtually all kinds of
social behavior.'* The longevity, stability, and influence values have on
attitudes and behavior make them important objects of consideration, as
they are likely to appear in such behavior as the expression of one's opinion in a public forum.
EVALUATING PUBLIC DISCOURSE IN NEWSPAPER OPINION ARTICLES
Literature
Review
59
One challenge in framing research is to develop coherent frameworks for understanding and analyzing frames that can be applied
across issues. We argue that frames, insofar as they characterize schemata in the minds of writers," are likely to derive at least in part from summary attitudes about appropriate social and personal priorities.
Numerous studies have examined the effects of value frames in news,
applying various operationalizations of values.™ In some cases, values
framing overrides attitudes about policy costs and benefits, indicating
that such framing has a strong impact on decision making.^'
These studies emphasize values framing from the perspective of
journalists and professionals. In the present study, we examine whether
lay discourse can be characterized in terms of values and compare these
value frames with more "elite" perspectives expressed in the editorials
and columns of the same newspapers. Moreover, we aim to develop a
theory-based methodology for examining values framing by building
on the work of Schwartz.^
Prior work examining values in message content has utilized
Inglehart's dichotomy of economic values,^•' or similar schemes.^''
Others have used value surveys such as Rokeach's Value Survey.^ But
Schwartz^*^ has argued that there is no theoretical connectivity among
the various values, making it difficult to interpret beyond simple
descriptive assessments. Schwartz^'' offered an alternative model that
includes two dimensions with opposite ends: a self-transcendence and
self-enhancement dimension, as well as an opermess-to-change and conservation dimension. That is, the openness to change dimension (hedonism, stimulation, and self-direction) stands contrary to conservation
(security, conformity, and tradition). Similarly, the self-enhancement
dimension (achievement, power) opposes self-transcendence (universalism, benevolence). This model places values in a more theoretical context than Rokeach's, suggesting that people's priorities on adjacent values will be similar, while those opposite will be maximally different.^* If
such patterns exist in experimental and survey data, we would expect to
find these patterns in newspaper opinion content as well:
HI: If more than one value is endorsed in a unit, these
values will likely be nearer to each other on Schwartz's location of values.
However, it is likely that a writer may endorse one or more values while
denouncing only one. In this case:
H2: If one value is denounced in a unit, this value will
likely be further apart (or opposite) values that are
endorsed.
The outer edges of Schwartz's structure of values represent the
dimensions—or "motivational goals"—that are likely to be oppositely
endorsed, because they cannot be easily pursued at the same time. It follows, then, that:
60
JOURNALISM & MASS COMMUNICATION
QUARTERLY
H3: Motivational goals (or dimensions), when endorsed, will be negatively associated with those motivational goals on the opposite side of the value structure.
A primary concern of this study is an assessment of how opinion
pieces articulate and frame their arguments. The ideals of a vibrant public discourse suggest that intellectual richness and complexity can and
should be found in such public expressions of opinion. One way of
assessing such complexity is the extent to which opinion pieces and letters diverge from the hypothesized patterns of value endorsement and
incorporate more complex arguments that assimilate other value perspectives. In fact, one might argue that framing as a persuasive technique
at its most sophisticated involves co-opting opponents' value stances and
using them to support one's own position. Therefore, we ask:
RQl: To what extent do (a) letters to the editor compared to (b) editorials and columns incorporate value
endorsements that do not represent clusters of adjacent values, as in Schwartz's value scheme?
Integrative Complexity. Integrative complexity provides yet
another approach to assessing opinion complexity. This measure assesses how successful writers are at evaluating an issue with multiple perspectives and making connections among them.^' Research in this area
has generally examined archival content,^" although some have applied
this to controversies that appear among groups in newspaper articles.^'
Integrative complexity consists of two components. Differentiation
refers to the awareness of different dimensions or perspective-taking
when considering an issue, while integration refers to conceptual connectivity among differentiated dimensions of or perspectives within the
issue.'^ In this way, some amount of differentiation is required to achieve
integration. Research on integrative complexity has proliferated in social
psychology in both experimental and archival research, but has not regularly been applied in content analyses of news media.
We propose two major hypotheses. The first hypothesis is rooted in
Tetlock and colleagues' work on the "value pluralism model."-'' Simply,
this model suggests that value conflict on an issue positively predicts
integrative complexity. Tetlock's seminal findings revealed that individuals think about policy issues in more integratively complex ways the
more they express conflict on related values. Importantly, research on the
value pluralism model has relied upon Rokeach's Value Survey.
Although it is relatively simple to conceptualize value conflict between
two values using this survey (e.g., "a world of beauty" conflicts with
"personal prosperity,"),** when more than two values are introduced,
measurement becomes less straightforward. We have employed the
Schwartz value structure because of its strong theoretical support, as well
as its intuitive approach to measuring value "conflict" as values that
oppose each other on the value circle. We operationalized this conflict as
the mean difference among values. It follows that such conflict would be
EVALUATING
PUBLIC DISCOURSE
IN NEWSPAPER
OPINION
ARTICLES
related to greater integrative complexity according to the value pluralism model. Thus, we hypothesize:
H4: Mean difference among values will positively predict integrative complexity, controlling for length and type
of article.
Other work suggests there is a relationship between both values
and integrative complexity and political ideology.'' Specifically,
American liberalism tends to score higher on integrative complexity,
while values associated with American conservatism score lower. In
addition, Schwartz and others found that endorsement of certain values
is strongly associated with political ideology.^^ Those values associated
with liberalism are housed under the self-transcendence dimension, while
those associated with conservatism can be found under the conservation
dimension. We hypothesize that these values will be associated with
integrative complexity, such that:
H5: Integrative complexity scores will (a) positively
predict the endorsement of self-transcendence (or "liberal")
values, and (b) negatively predict the endorsement of conservation (or "conservative") values, controlling for length
and type of article.
The Context: Substance-Policy Issues in Public Discourse.
Although much framing literature centers on political issues, little
research has conducted such analyses of the opinions expressed in
newspaper stories about health issues. Frames do emerge in such discourse, even though it is presumably more "neutral" than overtly political issues. Because many political issues are polarized by ideological
stances,^' our sample provides a fairly tough test of values framing
because it is less ideologically laden. That is, arguments are less likely to
rely upon knee-jerk values endorsed by political ideology
Examining value frames is of particular interest with respect to
health policy issues for other reasons as well. Presumably, few people
would say they are "against" health. Yet there is substantial debate over
health policy initiatives that are based primarily on competing values.
For example, smoking bans and alcohol-control initiatives protect
health, but also limit personal freedoms. In order to understand the
nature of these debates, and the role of media in influencing public support, understanding value framing of these health issues is essential.
The present sample provides potential answers as to which values are
associated with topics surrounding health-policy issues, enabling scholars to better understand the predominant values associated with them.
We ask:
RQ2: For each subtopic (i.e., alcohol, smoking, accidents, crime, drugs, cancer, and general health), which values will be appealed to most frequently?
62
JOURNALISM & MASS COMMUNICATION
QUARTERLY
RQ3: On which value dimensions will they cluster?
RQ4: How does this vary by article type (letter to the
editor, editorial, opinion column)?
This study employed a content analysis to assess values framed in
opinion columns, editorials, and letters to the editor (N = 591 articles).
Sample. To obtain a representative national sample of daily newspapers, media outlets were stratified into six strata across the country
based on their designated market areas or DMA®'* with each stratum
consisting of approximately one-sixth of all U.S. households." One constructed month was sampled each from 2002 and 2003 to avoid problems
associated with using random samples of media content.''" Each year's
constructed month also consisted of one constructed week from each season to account for seasonal change in content. Normally, eighteen local
daily newspapers and one daily national newspaper {USA Today), were
sampled per sampling date.
Article Selection.
Stage 1—Health Issue Topics. Coders in an earlier phase of this
research were trained to identify stories that contained information about
alcohol, illegal drugs, violent crimes, and accidents.*' This phase of coding took place for the parent study of the present analysis, and Cohen's
kappa was used to measure story type reliability (.64 to .86).
Stage 2—Opinion Content. Two coders, both journalism instructors,
were trained to identify letters to the editor, opinion columns, and editorials. Excluded from this analysis were sports commentaries, food
columns, entertainment columns, garden columns, feature or profile stories, travel columns, or children's commentary. The coders were trained
to identify stories by looking at the headline, byline, and dateline; checking for advice columns (which were excluded); and examining the content of the story for opinion expression. Krippendorf's alpha (which controls for chance agreement) was utilized in this stage of reliability testing,
and was .90.
Measurement and Reliability.
Values. Schwartz's value structure was used to code values. A coding sheet obtained from Suedfeld and Weiszbeck^^ was modified slightly
to fit the present analysis (see Table 1). If one or more of the code words
for a value was referenced in an article, that article was coded as 1 for
that value; if not, it was coded as 0.
Also included on the values coding sheet was whether the value
was endorsed (coded as 1) or not (-1). An article was defined as endorsing a value if the author either implied or stated outright that he or she
felt positively toward that value as a goal in one's life. If the author either
implied or stated that he or she felt negatively about the value and would
not recommend it as a goal in one's life, the value was coded as not
endorsed. An excerpt from one article in the sample demonstrates how
values were coded: "Early parental intervention is a way to detour a teen
from what might become a lifelong struggle with alcohol or drugs....
EVALUATING PUBUC DISCOURSE IN NEWSPAPER OPINION ARTICLES
63
TABLE 1
Values and Their Code Words
Value
Power
Achievement
Hedonism
Stimulation
Self-direction
Universalism
Benevolence
Tradition
Conformity
Security
Code Words
Social status, prestige, control, dominance, authority, wealth
Successful, capable, ambitious, influential, intelligent, hard work, perseverance,
competitiveness, competence
Pleasure, enjoying life, sensuality
Daring, a varied life, novelty, an exciting life
Curious, creativity, freedom, choose own goals, independent, explore, autonomy, agency
Protecting the environment, a world of beauty, social justice, wisdom, equality,
a world at peace, welfare of others, tolerance
Helpful, honest, forgiving, friendship
Humble, respect for tradition, accepting my portion in life, customs, religious
Obedient, obey norms, respect for others, honors parents and elders, politeness,
self-discipline
Clean, national security, patriotism, social order, family security, harmony,
healthy
The greatest gift that parents can give a child is to care enough to stop
him or her from starting on this road to self-destruction." The article was
coded as endorsing universalism (welfare of others, specifically the
teenagers), achievement (hard work, influential, and competence on the
part of the parents), and security (family security, as in protecting one's
children from harm).
Two coders were trained on a sample of 75 letters to the editor,
opinion columns, and editorials not included in the sample. Half of the
total sample was then randomly assigned to each coder. A subset (28%
or 170 articles) was randomly selected and coded for reliability.
Krippendorf's alpha ranged from .74 to 1.0 for inclusion of values and
.72 to 1.0 for endorsement of values.
Measurement of Integrative Complexity. The two coders were
trained on a sample of approximately 200 articles not included in the
sample over a three-month period using the standard codebook for integrative complexity." After training, coders rated a subset of articles in
the sample (28%), and Krippendorf's alpha = .79. The integrative complexity scale ranges from 1 to 7, although the maximum score in our
sample was a 6 and the mean was 2.3 (sd = 1.3). This implies that on
average, articles were relatively simple in their integrative complexity,
with some acknowledgment of differentiation of concepts and an increased tolerance for ambiguity, but no integration of these concepts
(which begins at a score of 4). Tetlock acknowledged that people generally prefer integratively simple styles of reasoning, so this is typical of
integrative complexity scores.""* But because this variable was non-normally distributed, we conducted a square-root transformation in order
to reduce the positive skewness of the variable for regression analyses.
This resulted in a variable ranging from 1 to 2.45, with a mean of 1.47 (sd
= .42).
In order to create a variable for value conflict we created SPSS syntax"*^ to calculate the distance between values when more than one value
JOURNALISM & MASS
COMMUNICATION
QUARTERLY
•
TABLE 2
Mean Occurrences of Values and Value Endorsements
Topic
Power
Power Endorsed
Achievement
Achievement Endorsed
Hedonism
Hedonism Endorsed
Stimulation
Stimulation Endorsed
Self-Direction
Self-Direction Endorsed
Universalism
Universalism Endorsed
Benevolence
Benevolence Endorsed
Tradition
Tradition Endorsed
Conformity
Conformity Endorsed
Security
Security Endorsed
n
M
sd
591
214
591
217
591
11
591
6
591
89
591
288
591
28
591
46
591
80
591
182
-.06
.40
.36
.98
.02
.64
.01
.83
.15
.97
.48
.98
.05
1.0
.08
.74
.14
.98
.31
.98
.592
.491
.480
.150
.135
.505
.100
.408
.356
.181
.500
.131
.212
.000
.265
.444
.342
.157
.461
.147
was endorsed.""^ Four iterations of each of the 10 values were performed. If the value of interest was not endorsed, then the value difference score for that article was coded as missing. If only the value of
interest was endorsed, the value difference score for that story was
coded as zero. If only the value of interest and a single adjacent value
were endorsed, the value difference score for the story was 1, and so
forth.
Value difference scores for the same article could potentially
vary depending on the starting-point value and the relative positions
of endorsed values, and whether more than two values were
endorsed. To account for this, we created a "mean difference" that
averaged all of the value difference scores generated for each article.
For example, if universalism and achievement were both endorsed in
an article, the value difference score would be a 4, because that is the
physical distance between them on Schwartz's value structure. But if
universalism, achievement, and security were endorsed, we would see
not only a 4-point difference because achievement is endorsed, but an
additional 3-point difference because security is endorsed. In this way,
the universalism difference score would be a 7 (4 + 3). However, this
would not be the score for the same article if we started with either the
achievement or security values. The three scores that were generated
from starting at each of these respective points (in this example)
account for the overall difference represented in that particular article.
There were more letters (51.6%) than either editorials (23%) or
columns (24%) in this analysis. All but 18 of the 591 articles included a
EVALUATING PUBLIC DISCOURSE IN NEWSPAPER OPINION ARTICLES
Results
65
TABLE 3
Bivariate Correlation Matrix of Values Endorsed
Powr
Powr
Acht
Hed
Stim
Self
Univ
Ben
Trad
Conf
Sec
.141**
.011
.008
-.029
-.023
.036
.043
.003
.142**
Acht
Hed
Stim
.141**
—
.050
-.031
-.123**
-.199**
.006
-.001
.011
.050
.008
-.031
-.003
—
-.014
.056
-.010
-.006
-.016
-.026
-.111*
-.117*
—
-.003
-.019
-.050
-.013
.105*
-.091'
-.034
Self
-.029
-.123**
-.019
.056
—
-.152**
-.064
-.036
-.007
-.097*
Univ
Ben
Trad
Conf
-.023
-.199**
-.050
-.014
-.152**
—
.054
.001
-.191**
-.162**
.036
.006
-.013
-.010
-.064
.054
—
-.025
.002
-.114*
.043
-.001
.105*
-.006
-.036
.001
-.025
—
-.040
-.064
.003
-.111*
-.091*
-.016
-.007
-.191**
.002
-.040
-.044
Sec
.142*
-.117*
-.034
-.026
-.097*
-.162*
-.114*
-.064
-.044
—
Note: Powr = Power, Acht = Achievement, Hed = Hedonism, Stim = Stimulation, Self = Self-Direction,
Univ = Universalism, Ben = Benevolence, Trad = Tradition, Conf = Conformity, Sec = Security.
Patterns in bold are consistent with Schwartz's value structure.
**p<.01 * p < . 0 5 +<.1O.
reference to at least one value (see Table 2). Universalism was the
most frequently occurring value (M = .48, sd = .50). Most values
were endorsed more often than denounced, as reflected in the positive
mean for each endorsed value, except for one value: power (M = -.06, sd
= .59).
Values. First we address the research questions. RQl asked to
what extent letters versus editorials and columns endorsed values adjacent to each other in the value structure. In order to test this, we utilized
the "mean difference of values" variable. Opinion columns and editorials were collapsed into one category ("professional" articles) and compared to letters to the editor. Letters had a lower mean difference of values (M = 1.86, sd = 2.07) than professional articles (M = 2.42, sd = 2.29),
and this difference was significant, f(524) = -2.97, p < .01. However, once
length of article was controlled, article type did not significantly predict
mean difference of values, p = .002, p = .974, R' = .04. These results suggest that once length is taken into account, professional articles are no
more likely to include values further apart on the value structure than
letters to the editor.
RQ2 asked which values appeared most frequently by topic. An
analysis of values by topic of article (i.e., smoking, accidents, crime,
drugs, alcohol, cancer, general health, and other) revealed that crime
represented 42% of the articles. The distribution of topics in all other
articles was smoking (13%), drugs (11.3%), accidents (10.2%), alcohol
(5.6%), general health (4.6%), cancer (2.9%), and other (10.5%). In order
to conduct a comparison of value means by topic, these topics were collapsed into two categories: substance (smoking, alcohol, drugs), and
non-substance health-policy issues (accidents, crime, cancer, other).
Universalism was endorsed significantly less in substance articles than
in non-substance health-policy articles, t(589) = -2.917, p < .01 as was
benevolence, f(550.5) = -4.446, p < .001.
66
JOURNALISM & MASS COMMUNICATION QUARTERLY
RQ3 asked on which value dimensions the topics would cluster.
The results suggest that self-transcendence, which encourages concern
for others and social justice, was less frequently endorsed in substance
use articles. Power, which is in the self-enhancement dimension, was
endorsed more frequently in non-substance health-policy articles than in
substance use articles, t(589) = -1.918, p < .10.
Finally, RQ4 asked which values appeared most frequently by article type. Results revealed that three values differed significantly between
letters and professional articles. Achievement was present significantly
more frequently in professional articles than in letters, f(575.4) = -4.11,
p < .01, as was benevolence, ^(523.2) = -1.70, p < .10. Tradition was not
only more frequent in letters, t(561.7) = 2.138, p < .05, but it was also
endorsed more frequently, t(21.4) = 2.046, p < .10.
To address HI and H2, which predicted that values endorsed
would be nearer to each other and opposite to those denounced, only
those cases with more than one value referenced were included. This
resulted in a sample size of 449 (76% of the entire sample). The values
that were correlated with each other according to Schwartz's location of
values are in boldface type in Table 3."''
Nearly all significant correlations produced the predicted relationships, such that they are negatively correlated with each other when the
values are opposite each other in the structure and positively correlated
when they are adjacent or nearly adjacent. There were only two sets of
values that were endorsed but correlated significantly in the opposite
direction of the prediction. Tradition and hedonism, which are opposite
of each other in the structure, were positively correlated, r = .105, p < .05.
This was, in fact, the only significant correlation for the tradition value. It
is possible that the result is an artifact of the very small number of cases
that endorsed hedonism (n = 7 out of 11 total references to hedonism).
The second set of values that did not follow from the prediction was selfdirection and universalism. These values are adjacent to each other but
were negatively correlated, r = -.152, p < .001. Overall, though, the results
fit remarkably well with the predicted structure of the values, for the
most part supporting Schwartz's thesis and supporting HI and H2.
H3 predicted relationships among the motivational goals or dimensions of the value types. The values underlying these motivational goals
cannot easily be pursued concurrently, so negative relationships were
predicted. New variables were created that summed values under each
motivational goal for those cases that referenced more than one value.
Openness to change (hedonism, stimulation, and self-direction) was
significantly negatively correlated with conservation (security, conformity, and tradition), r = -.111, p < .05. Similarly, self-enhancement
(achievement, power) was significantly negatively correlated with selftranscendence (benevolence, universalism), r = -.240, p < .001. H3 was
supported.
Integrative Complexity and Values. Three values were significantly correlated with integrative complexity: achievement, r = .12, p < .05,
universalism, r = .11, p < .05, and tradition, r = -.12, p < .05. Thus, on average, those articles that endorsed achievement and universalism were
EVALUATING PUBUC DISCOURSE IN NEWSPAPER OPINION ARTICLES
67
TABLE 4
Regression Analysis Predicting Integrative Complexity Scores in Forum Articles
Length of Article
Endorsement of Institutional Policy
Column/Editorial
Mean Difference of Values
.467***
.131**
.130**
-.059
Note: Results reflect square-root transformation applied to the dependent variable in order to reduce
skewness.
** p < .05. *** p < .001
R^ = .31
more integratively complex, but those that endorsed tradition were significantly less complex.
We coded for length of article and found that the minimum length
was 35 words, while the maximum was 1,300 words (M = 341.60, sd =
243.20). This variable was positively correlated with the integrative complexity score, r = .49, p < .001, so we controlled for this variable in subsequent analyses. We also coded for whether the article supported institutional policy, i.e., if the author endorsed the role of an authority to complete some action. An example would be a letter to the editor requesting
a local political figure to bring charges against a criminal. This endorsement was significantly and positively correlated with integrative complexity, r = .20, p < .01. Thus, those articles that explicitly proposed or endorsed some policy action (M = 2.82, sd = 1.30) also tended to be more
integratively complex than those that did not (M = 2.28, sd = 1.30) and
policy endorsement was controlled for in subsequent analyses. Finally,
we controlled for whether the article was a letter to the editor (M = 1.84, sd
= 1.02) or a column/editorial (M = 2.88, sd = 1.35), because we found that
there was a positive correlation between these variables, r = .40, p < .001.
To test H4, we conducted a correlation between the mean difference of values and integrative complexity and found a significant correlation, r = .11, p < .01. However, a regression analysis controlling for length
of article, article type, and policy endorsement revealed that mean difference
of values was not a significant predictor of integrative complexity (see
Table 4). Thus, mean value difference represents a different kind of complexity than is captured by the integrative complexity measure.
Our final set of hypotheses concerned the relationship between
values, integrative complexity, and political ideology. H5a predicted that
values associated with liberal ideologies would be endorsed in integratively complex articles, while H5b predicted that conservative values
would be endorsed in less integratively complex articles. H5a was supported, r = .12, p < .01. Results for H5b were not significant, although the
correlation was in the predicted direction, r = -.04, p = .34.
Discussion
68
This analysis endeavored to apply a reliable and valid value structure to the framing of arguments in editorials, opinion columns, and letJOURNALISM & MASS COMMUNICATION
QUARTERLY
ters to the editor in newspapers. The overwhelming presence of values
in arguments (97% of the sample referenced at least one value) implies
that values are indeed a means by which both professional journalists
and citizens frame their arguments.
There were significant differences among the values referenced
and endorsed when tested against the type of article. While self-enhancement (i.e., achievement) and self-transcendence (i.e., benevolence) values were referenced more often in opinion columns and editorials, conservation (i.e., tradition) was both referenced and endorsed significantly
more in letters to the editor. Our findings also imply somewhat greater
complexity on the part of the professional articles, because these articles
include values that are in opposition to each other in the value structure.'*" The results suggest that journalists and citizens prioritize different values in their framing of substance and health policy issues in this
public forum, and that journalists, unsurprisingly, tend to approach
issues in a more complex way.
By topic, we also found significant differences among the values
referenced and endorsed. Those articles on accidents, crime, cancer, and
general health more frequently endorsed the self-transcendence values
of universalism and benevolence. These values promote support for
social justice, tolerance, welfare of others, caring, and forgiveness. These
were not endorsed in the substance use articles, which were more likely
to frame arguments through the endorsement of power, which prioritizes control, dominance, and authority. When more than one value was
referenced, those that were endorsed more frequently were consistent
with adjacent values in Schwartz's value structure. Those values that
were denounced were more frequently on opposite sides of endorsed
values in the structure, as predicted. This pattern, while unsurprising, is
important in that it underscores the conceptual validity of values coding
as a way of capturing important dimensions of public discourse in media
content, as well as in experimental and survey data.
The representativeness of the sample—in geographic location,
media market, season, and time period—also makes this study an
important contribution to the literature. This study answered the calls of
previous research that lamented the lack of research revealing the tone of
letters to the editor.'" In this case, we provided a thorough analysis of
values in public health forum articles in a nationally representative sample.
We also aimed to relate the Schwartz value structure to integrative
complexity, building on previous research that linked integrative complexity to the Rokeach Value Survey.™ We found that measuring values
with the Schwartz value structure does not serve as an indicator of integrative complexity, but certain values were related to integrative complexity. Universalism and achievement values were positively correlated
with integrative complexity, while tradition was negatively correlated
with integrative complexity. These results are consistent with previous
findings suggesting that integrative complexity is negatively related
with conservatism.^' Yet, we must acknowledge that our correlation
coefficients are relatively modest, albeit statistically significant.
EVALUATING
PUBLIC DISCOURSE
IN NEWSPAPER
OPINION
ARTICLES
This research also serves practical purposes. Since media content
has been shown to influence the public's attitudes about health policy
issues,^^ the frames that practitioners use are particularly important.
Our results suggest that certain public health topics may be more easily
aligned with some values over others. Public health campaigns might
utilize this descriptive information to frame their messages in a manner
consistent with existing public discourse.
Overall, these results suggest that, indeed, both journalist and lay
discourse on issues—even issues that are not inherently highly ideological—can be characterized in terms of their value structure. Such an
analysis provides insight into the underlying ideological (i.e., valuebased) perspectives of the lay public and journalists regarding important public policy issues and how these may vary by issue topic. Our
results provide journalism and mass communication scholars with confirmation that values serve to frame editorials, letters to the editor, and
columns in newspapers. Equipped with an understanding of this value
structure—and how it differs by topic—scholars can continue to assess
how these values are represented in other types of journalistic writing
and with topics other than health policy. The research presented here
provides a link between long-standing psychological research and
extant interest in media frames, thus enhancing our understanding of
mass communication processes, providing deeper insight into the
effects of value frames in the media. These results should also prove
valuable in understanding the role of journalist and lay public values in
public opinion processes, as well as guiding advocacy efforts to enlist
support for issues such as those in health and public safety.
NOTES
1. Rodney Benson and Abigail C. Saquy, "Constructing Social Problems in an Age of Globalization; A French-American Comparison,"
American Sociological Review 70 (April 2005): 233-59; Peter Suedfeld and
A. Dennis Rank, "Revolutionary Leaders: Long-Term Success as a
Function of Changes in Conceptual Complexity," Journal of Personality
and Social Psychology 34 (February 1976): 169-78; Phillip E. Tetlock,
"Cognitive Style and Political Belief Systems in the British House of
Commons," Journal of Personality and Social Psychology: Personality
Processes and Individual Differences 46 (February 1984): 365-75.
2. Shalom H. Schwartz, "Universals in the Content and Structure of
Values: Theoretical Advances and Empirical Tests in 20 Countries,"
Advances in Experimental Social Psychology 25 (1992): 1-65; Shalom H,
Schwartz, "Are There Universal Aspects in the Structure and Contents
of Human Values?" Journal of Social Issues 50 (winter 1994): 19-45;
Shalom H, Schwartz, "Value Priorities and Behavior: Applying a Theory
of Integrated Value Systems," in The Psychology of Values: The Ontario
Symposium, Volume 8, ed. Clive Seligman, James M. Olson, and Mark P,
Zanna (1996), 1-23.
3. Tetlock, "Cognitive Style and Political Belief Systems in the
70
JOURNALISM & MASS COMMUNICATION QUARTERLY
British House of Commons."
4. Karin Wahl-Jorgensen, "A 'Legitimate Beef or 'Raw Meat?'
Civility, Multiculturalism, and Letters to the Editor," The Communication
Review 70 (March-May 2004): 89-105.
5. Wahl-Jorgensen, "A 'Legitimate Beef or 'Raw Meat?'"
6. See, for example, James B. Lemert, Barry N. Mitzman, Michael A.
Seither, Roxana H. Cook, and Regina Hackett, "Journalists and
Mobilizing Information," Journalism Quarterly 54 (winter 1977): 721-26;
Pamela J. Shoemaker and Stephen D. Reese, Mediating the Message:
Theories of Influence on Mass Media Content, 2d ed. (White Plains, NY:
Longman, 1996); Karin Wahl-Jorgensen, "The Normative-Economic
Justification for Public Discourse: Letters to the Editor as a 'Wide Open'
Forum," Journalism & Mass Communication Quarterly 79 (spring 2002):
121-33.
7. Andrew J. Perrin, "Political Microcultures: Linking Civic Life and
Democratic Discourse," Social Forces 84 (December 2005): 1049-82.
8. See, for example, Anita G. Day and Guy Golan, "Source and
Content Diversity in Op-Ed Pages: Assessing Editorial Strategies in The
New York Times and the Washington Post," Journalism Studies 6
(February 2005): 61-71; Graham C. Kinloch, "Changing Racial Attitudes
in Zimbabwe: Colonial/Post-Colonial Dynamics," Journal of Black Studies 34 (November 2003): 250-71; Sheryl Magzamen, Annemarie
Charlesworth, and Stanton A. Glantz, "Print Media Coverage of
California's Smoke-Free Bar Law," Tobacco Control 10 (summer 2001): 15460.
9. See, for example, Paul R. Brewer, "Value Words and Lizard Brains:
Do Citizens Deliberate About Appeals to Their Core Values?" Political
Psychology 22 (March 2001): 45-64; Karen Callaghan and Frauke Schnell,
"Assessing the Democratic Debate: How the News Media Frame Elite
Policy Discourse," Political Communication 18 (April 2001): 183-212;
Thomas E. Nelson, "Policy Goals, Public Rhetoric, and Political
Attitudes," Journal of Politics 66 (May 2004): 581-605; Thomas E. Nelson,
Zoe M. Oxley, and Rosalee A. Clawson, "Toward a Psychology of
Framing Effects," Political Behavior 19 (September 1997): 221-46; Fuyuan
Shen and Heidi H. Edwards, "Economic Individualism, Humanitarianism, and Welfare Reform: A Value-Based Account of Framing
Effects," Journal of Communication 55 (December 2005): 795-809.
10. Donald R. Kinder and Lynne M. Sanders, Divided by Color: Racial
Politics and Democratic Ideals (Chicago: University of Chicago Press, 1996);
Zhongdang Pan and Gerald M. Kosicki, "Framing and the Understanding of Citizenship," in The Evolution of Key Mass Communication
Concepts: Honoring Jack M. McLeod, ed. Sharon Dunwoody, Lee Becker,
Doug McLeod, and Gerald Kosicki (Cresskill, NJ: Hampton Press, 2005),
167-204; Vincent Price and David Tewksbury, "News Values and Public
Opinion: A Theoretical Account of Media Priming and Framing" in
Progress in Communication Sciences, ed. George A. Barnett and Franklin J.
Bolster (Greenwich, CT: Ablex, 1997), 173-212.
11. Pan and Kosicki, "Framing and the Understanding of Citizenship."
EVALUATING PUBLIC DISCOURSE IN NEWSPAPER OPINION ARTICLES
71
12. Pan and Kosicki, "Framing and the Understanding of Citizenship."
13. Connie M. Kristiansen and Mark P. Zanna, "Justifying Attitudes
by Appealing to Values: A Functional Perspective," British Journal of
Social Psychology 27 (September 1988): 247-56.
14. Schwartz, "Universals in the Content and Structure of Values."
15. Daniel M. Mayton, Sandra J. Ball-Rokeach, and William E. Loges,
"Human Values and Social Issues: An Introduction," Journal of Social
Issues 50 (winter 1994): 1-8; Bas Verplanken and Rob W. Holland,
"Motivated Decision Making: Effects of Activation and Self-Centrality of
Values on Choices and Behavior," Journal of Personality and Social
Psychology 82 (March 2002): 434-47.
16. Mayton, Ball-Rokeach, and Loges, "Human Values and Social
Issues: An Introduction."
17. M. Kent Jennings, "Thinking About Social Injustice," Political
Psychology 12 Qune 1991): 187-204; Phillip E. Tetlock, "A Value-Pluralism
Model of Ideological Reasoning," Journal of Personality and Social
Psychology 50 (April 1986): 819-27.
18. Jennings, "Thinking About Social Injustice"; Milton Rokeach, The
Nature of Human Values (New York: Free Press, 1973); Tetlock, "A ValuePluralism Model of Ideological Reasoning."
19. Robert M. Entman, "Framing: Toward Clarification of a Fractured Paradigm," Journal of Communication 43 (December 1993): 293-300.
20. Paul R. Brewer, "Framing, Value Words, and Citizens' Explanations of Their Issue Opinions," Political Communication 19 (July 2002):
303-16; David Domke, Dhavan V. Shah, and Daniel B. Wackman,
"'Moral Referendums': Values, News Media, and the Process of
Candidate Choice," Political Communication 15 0une 1998): 301-21;
Nelson, "Policy Goals, Public Rhetoric, and Political Attitudes"; Thomas
E. Nelson and Elaine A. Willey, "Issue Frames That Strike a Value
Balance: A Political Psychology Perspective," in Framing Public Life:
Perspectives on Media and Our Understanding of the Social World, ed.
Stephen D. Reese, Jr., Oscar H. Gandy, and August E. Grant (Mahwah,
NJ: Lawrence Eribaum, 2001), 245-65; Dhavan V. Shah, David Domke,
and Daniel B. Wackman, "The Effects of Value-Framing on Political
Judgment and Reasoning," in Framing Public Life: Perspectives on Media
and Our Understanding ofthe Social World, ed. Stephen D. Reese, Jr., Oscar
H. Gandy, and August E. Grant (Mahwah, NJ: Lawrence Eribaum,
2001), 227-43; Shen and Edwards, "Economic Individualism, Humanitarianism, and Welfare Reform"; Mira Sotirovic and Jack M. McLeod,
"Values, Communication Behavior, and Political Participation," Political
Communication 18 (July 2001): 273-300.
21. Nelson, "Policy Goals, Public Rhetoric, and Political Attitudes";
Shen and Edwards, "Economic Individualism, Humanitarianism, and
Welfare Reform."
22. Schwartz, "Universals in the Content and Structure of Values";
Schwartz, "Are There Universal Aspects in the Structure and Contents
of Human Values?"; Schwartz, "Value Priorities and Behavior."
23. Ronald Inglehart, Culture Shift in Advanced Industrial Society
72
JOURNALISM & MASS COMMUNICATION
QUARTERLY
(Princeton, NJ: Princeton University Press, 1990).
24. Hong Cheng, "Toward an Understanding of Cultural Values
Manifest in Advertising: A Content Analysis of Chinese Television
Commercials in 1990 and 1995," Journalism & Mass Communication
Quarterly 74 (winter 1998): 773-96; Hye-Jin Paek, Michelle R. Nelson, and
Douglas M. McLeod, "Beyond Polarized Cultural Values: A New
Approach to the Study of South Korean and U.S. Newspaper Ads," in
Advances in Consumer Research, ed. Barbara E. Kahn and Mary F. Luce
(Provo, UT: Association for Consumer Research), 495-502.
25. Rokeach, The Nature of Human Values.
26. Schwartz, "Are There Universal Aspects in the Structure and
Contents of Human Values?"
27. Schwartz, "Universals in the Content and Structure of Values";
Schwartz, "Are There Universal Aspects in the Structure and Contents of
Human Values?"; Schwartz, "Value Priorities and Behavior."
28. Meg J. Rohan, "A Rose by Any Name? The Values Construct,"
Personality and Social Psychology Review 4 (August 2000): 255-77.
29. Phillip E. Tetlock, "Cognitive Structural Analysis of Political
Rhetoric: Methodological and Theoretical Issues," in Explorations in
Political Psychology, ed. Shanto Iyengar and William J. McGuire (Durham,
NC: Duke University Press, 1993), 380-405.
30. See, for example, Suedfeld and Rank, "Revolutionary Leaders."
31. Peter Suedfeld, Loraine LaValle, and Jennifer Brough, "Political
Language in an Environmental Controversy: Integrative Complexity and
Motive Imagery in Advocacy Propaganda and the Press," in Politically
Speaking: A Worldwide Examination of Language Used in the Public Sphere,
ed. Ofer Feldman and Christ'l De Landtsheer (Westport, CT: Praeger,
1998), 170-83.
32. Gloria Baker-Brown et al.. Coding Manual for Conceptual/Integrative
Complexity, 1992, http://www.psych.ubc.ca/~psuedfeld/MANUAL.
pdf.
33. Tetlock, "A Value-Pluralism Model of Ideological Reasoning";
Phillip E. Tetlock, Randall S. Peterson, and Jennifer S. Lerner, "Revising
the Value Pluralism Model: Incorporating Social Content and Context
Postulates," in The Psychology of Values: The Ontario Symposium, Volume 8,
ed. Clive Seligman, James M. Olson, and Mark P. Zanna (Mahwah, NJ:
Lawrence Erlbaum, 1996), 25-51.
34. Tetlock, "A Value-Pluralism Model of Ideological Reasoning."
35. See Schwartz, "Are There Universal Aspects in the Structure
and Contents of Human Values?"; Tetlock, "A Value-Pluralism Model of
Ideological Reasoning"; Tetlock, "Cognitive Structural Analysis of
Political Rhetoric"; Tetlock, Peterson and Lerner, "Revising the Value
Pluralism Model."
36. See Schwartz, "Are There Universal Aspects in the Structure and
Contents of Human Values?"; Valerie A. Braithwaite, "Beyond Rokeach's
Equality-Freedom Model: Two-Dimensional Values in a One-Dimensional World," Journal of Social Issues 50 (winter 1994): 67-94.
37. Pan and Kosicki, "Framing and the Understanding of Citizenship."
EVALUATING PUBLIC DISCOURSE IN NEWSPAPER OPINION ARTICLES
73
38. Standard Rate and Data Service, SRDS TV& Cable Source (Des
Plaines, IL: SRDS, Inc, 2000).
39. For details on sampling method and data collection, see Marilee
Long, Michael D. Slater, Greg Boiarsky, Linda Stapel, and Thomas Keefe,
"Obtaining Nationally Representative Samples of Local News Media
Outlets," Mass Communication and Society 8 (fall 2005): 299-322.
40. Daniel Riffe, Charles Aust, and Stephen Lacy, "The Effectiveness of
Random, Consecutive Day and Constructed Week Sampling in Newspaper Content Analysis," Journalism Quarterly 70 (spring 1993): 133-39;
Daniel Riffe, Stephen Lacy, and Frederick Fico, Analyzing Media Messages:
Using Quantitative Content Analyses in Research (Mahwah, NJ: Lawrence
Erlbaum, 1998).
41. Stories were hard copies that were scanned and converted to
Portable Document Format (PDF) files.
42. Peter Suedfeld and Tara Weiszbeck, "The Impact of Outer Space
on Inner Space," Aviation Space and Environmental Medicine 75 Quly 2004):
C6-C9.
43. Baker-Brown et al.. Coding Manual for Conceptual/lntegrative
Complexity.
44. Tetlock, "Cognitive Structural Analysis of Political Rhetoric."
45. Syntax is available from the first author.
46. Tradition and conformity were considered to occupy the same
space on the circle in terms of calculating distance to other values and
were considered adjacent to one another, consistent with Schwartz's conceptual model (see Schwartz, "Universals in the Content and Structure of
Values"; Schwartz, "Are There Universal Aspects in the Structure and
Contents of Human Values?"; Schwartz, "Value Priorities and Behavior").
47. Other correlations not in boldface type, even those that are significant, do not fit with Schwartz's value structure.
48. Schwartz, "Value Priorities and Behavior"; Tetlock, "A ValuePluralism Model of Ideological Reasoning."
49. See, for example, Perrin, "Political Microcultures."
50. Tetlock, "A Value-Pluralism Model of Ideological Reasoning."
51. Phillip E. Tetlock, "Personality and Isolationism: Content Analysis
of Senatorial Speeches," Journal of Personality and Social Psychology 41
(April 1981): 737-43; Phillip E. Tetlock, "Cognitive Style and Political
Ideology," Journal of Personality and Social Psychology 45 Qanuary 1983):
118-26.
52. Joanne Stryker, "Media and Marijuana: A Longitudinal Analysis of
News Media Effects on Adolescents' Marijuana Use and Related Outcomes," Journal of Health Communication 8 Quly-August 2003): 305-28;
Itzhak Yanovitzky and Joanne Stryker, "Mass Media, Social Norms, and
Health Promotion Effects: A Longitudinal Study of Media Effects on
Youth Binge Drinking," Communication Research 28 (April 2001): 208-39.
74
JOURNALISM & MASS COMMUNICATION QUARTERLY