|
Newspapers and Public Opinion.
An examination of The Des Moines Register and its Iowa Poll.
A paper submitted to the Newspaper division of AEJMC.
April 1, 1995
Craig Trumbo
Department of Agricultural Journalism
440 Henry Mall, University of Wisconsin
Madison, WI 53706
608-262-4902
[log in to unmask]
Newspapers and Public Opinion.
Craig Trumbo, Dept. of Agricultural Journalism, University of Wisconsin -
Madison
ABSTRACT
Literature about newspaper representation of public opinion presents a
negative view of the relationship, predicting a newspaper's news
judgment
might not be independent of the polls published in the paper.
This
hypothesis is evaluated by looking at The Des Moines Register's
coverage of
issues featured in its Iowa Poll. Only a very weak relationship
was found,
suggesting that either The Register is somehow unique or that
the concerns
stated in the literature are over stated.
Newspapers and Public Opinion.
An examination of The Des Moines Register and its Iowa Poll.
INTRODUCTION
When it comes to the question of the news media and opinion polling critics
generally hold that the union is somewhere between unsavory and
unholy. The
argument goes something like this. Working under the guise of
"precision
journalism," impatient reporters unschooled in the fine points of the
scientific method create news to the detriment of both democracy and
society:
social conditions are distorted, the electoral process is subverted,
and the
effectiveness of valid opinion research is diluted in a wash of
mediocrity.
Yet many good things can also be said about media polling and the media's use
of opinion polls in general. The media's use of public opinion
polls provides
an important forum in which one variety of public sentiment can gain
equal
status against elite discourse. Opinion polls conducted by the news
media can
have a strong impact on public policy, prompting government action
against
social problems (Paletz et al., 1980). Media use of opinion polls
also
provides society with a mirror D perhaps a somewhat wavy mirror at times
D
but nonetheless a mirror of itself. And perhaps most importantly, at
least in
a political context as Traugott argues (1992, p. 130), opinion polls
in the
media can serve to enhance the public's sometimes flagging interest
in things
political.
From an effects perspective, the consequences of the media's use of polls can
generally be divided between those involving effects on the
public and those
involving effects on politicians and the democratic process.
Research
presents a mixed picture in terms of effects on the public. Ideas like
bandwagon, underdog, and others, have not been strongly supported by
consistent empirical evidence (Gollin, 1980; Traugott, 1992). But ideas
like
the third person effect and the spiral of silence do suggest viable
me
chanisms through which media representation of public opinion can have an
effect on the public (Noelle-Neumann, 1974; Davidson, 1983). In any
case, the
impact of media polling on the public is more elusive and indirect
than
common wisdom, and the critics, may sometimes hold.
The effect of media polling on politicians is perhaps more tenable. Media
polls can have a strong impact on the initial viability of a
campaign; either
enhancing or thwarting critical early financial support and
volunteerism. For
example, media polling has been cited as strongly influencing the
selection
of vice-presidential candidates (Traugott, 1992). Poll results can
place
positive or negative constraints on the actions of public officials.
And
media focus on poll results can completely overwhelm the message that a
politician wishes to transmit to the electorate (Dionne, 1992).
The question of the impact of poll reporting on the public and politicians
has generated an abundance of material in academic, professional,
and popular
publications. But largely ignored, at least in a relative sense, is
the
question of the impact of media opinion polling on the media.
POLLING AND JOURNALISM: AN OVERVIEW
Of specific interest to this article is the question of what impact
newspaper-conducted opinion polls might have on the newspapers that
conduct
such polls. A brief overview of the history of newspaper opinion
polling and
a look at some of the arguments for and against the practice will
preface an
investigation of the relationship between The Des Moines Register
and its
Iowa Poll.
Ismach (1984) cites a series of benchmark dates in the relationship between
newspapers and opinion polling. Newspaper opinion polling D as
straw polling
D has been traced back as far as 1824 by Tankard (1972). But the
real
foundation of the relationship was laid in the 1920s as newspapers began
to
publish the findings of Roper, Gallup, and Crossley.
It was in the mid-1940s that the Minneapolis Tribune and The Des Moines
Register became the first newspapers to regularly conduct and
publish their
own public opinion research. The collaboration of Sid Goldish, then
Minnesota
Poll director, and Ralph Nafzinger at the University of Minnesota
did much to
establish research-quality polling at the Minneapolis Tribune.
Nafzinger
brought valuable experience from the Office of War Information that
influenced not only early newspaper opinion polling but also the study of
public opinion in journalism programs, first at Minnesota and later at
the
University of Wisconsin where he was director of the School of
Journalism
from 1949 to 1965 (Rogers and Chaffee, 1994).
But it wasn't until the late 1960s and early 1970s that Philip Meyer's work
for Knight-Ridder, at the Detroit Free Press and other
publications, began to
more widely integrate the disciplines of social science into
newspaper
opinion polling. Meyer's training at Harvard and experience with
Knight-Ridder led him to publish his influential work Precision Journalism in
1973. Thereafter, both academic curriculum and professional
symposiums
continued to focus on integrating the investigative skills of the social
scientist into the journalist's "kit bag."
Weaver and McCombs (1980) make the argument that the relationship between
journalism and social science predates the coining of "precision
journalism."
They see the roots of the relationship going back to the original
humanistic
philosophies of journalism. What they see recently is a
strengthening of the
ties between social science and journalism as journalism has moved
away from
the simple reporting of facts and toward placing news in a social
context. As
the social sciences became increasingly empirical so did
journalism's
application of social science. Noelle-Neumann (1980) prefers the phrase
"data
precision journalism" because it better captures the nature of the
new
relationship.
In any case, it is clear is that newspapers dramatically increased their use
of opinion poll data by first publishing the results of polling
organizations
and later by either establishing in-house polling efforts or hiring
the
services of professional polling organizations. By the 1970s the opinion
poll
had become a highly visible fixture of the print journalism scene.
Demers (1987) replicated a 1978 study by Rippey in order to see if the
established nature of newspaper opinion polling had changed in recent
times.
Both Rippey and Demers surveyed about 400 daily newspapers selected
from the
Editor & Publisher Yearbook. Overall, Demers found that newspaper
opinion
polling had not changed a great deal over the decade. He also found
those
papers that are polling intend to continue to do so and those that are
not
have no intention of starting. This shows that opinion polling has
achieved
stability in the profession.
Demers reports that in 1986 about 35% of daily newspapers were conducting
their own public opinion polls (down insignificantly from 37%). He
did find
some change in how the polls were being conducted, though. There was
a shift
toward professionalization as newspapers moved away from conducting
all
aspects of polling in-house and toward using the services of outside
polling
firms and consultants for both data gathering and analysis. The
strongest
shift involved a move toward using professional survey interviewers.
While
this suggests an improvement in the quality of newspaper-conducted
opinion
polls, about half of the polling was still done on an exclusively
in-house
basis.
The Rippey-Demers work did not address the question of newspaper use of
opinion polls, only the prevalence of polls conducted by newspapers.
Even in
the absence of empirical support, it seems very safe to say that all
newspapers routinely report public opinion polls of one variety or another.
In fact, it is probably impossible to go through any issue of a
medium sized
paper without encountering an opinion poll in some form.
Miller and Hurd (1982) did examine conformity to AAPOR poll reporting
standards at three major newspapers: Los Angeles Times, Chicago
Tribune,
Atlanta Constitution. AAPOR recommends reporting size, sponsor,
wording,
error, population, method, and timing. Reporting poll sample size and
sponsorship were the most stringently followed standards. It was found
that
conformity to the standards was highest for election-related polls
and for
polls conducted in-house. On average, the standards were employed in
58 per
cent of polls reported.
Atkin and Gaudino (1984) ask why journalists have become so enamored with
opinion polling. They find an answer by looking at how the opinion
poll meets
the general tests for newsworthiness. They paint a broad picture of
"newsworthiness," admitting that there are no rules, only conventions. They
cite the commonly referred to conventions of interest, consequence,
proximity, timeliness, prominence, exclusivity, and the investment of a
publication's resources as factors that make something newsworthy.
By these standards, Atkin and Gaudino argue, opinion polls are very
newsworthy. Polls often deal with political material, containing
information
of interest and consequence to the audience. They can also serve to
sustain
audience involvement in issues. If it is allowed that public opinion
is
stable in the short term, then opinion polls also have a very useful
shelf-life. Further, polling represents a sizable expenditure of resources
so
editors will feel inclined to use the results in some way. These
factors are
only intensified if the poll material is either exclusive or if
there is a
competing publication flaunting its own poll findings.
With the popularity and rationale for newspaper opinion polling given, the
more important questions remain: What is good and bad about
newspaper opinion
polling and how does it affect news coverage?
The academic literature presents a litany of complaints against newspaper
opinion polling and media opinion polling in general D and very
little in its
support. Ismach (1984) presents a list of eight general categories
of charges
against media polling. Media polls: cause bandwagon or underdog
effects, fail
to predict elections and thus threaten polling credibility,
emphasize the
trivial, influence the news judgments of editors, have low validity
when they
are done on the cheap, are often inappropriately analyzed by
editors, and
usually lack conceptual and methodological rigor.
Noelle-Neumann (1980), while allowing that journalism can benefit through the
use of social scientific methods, nonetheless cautions against
journalism
becoming "social research in a hurry." She sees five acute areas of
difficulty that must be overcome if journalists are to harness the power of
the social scientific method. Journalists do not understand the
concepts
underlying measurement, do not see how measurement and prediction
relate, and
misunderstand the idea of the interview as a way to gather data. Her
other
cautions involve journalist's general lack of training in statistics
and the
specific problem with correlation and causality.
Ladd (1980) approaches the relationship between polling and journalism by
considering the two as unique institutions that each have distinct
characteristics that tend to defy a clean fit between the two. The news media
must often work quickly, but polling is a task D like most all tasks
in the
social sciences D that often requires slow progress. There is also
an almost
unavoidable disjuncture caused by the media's need to use "tight
editing."
Ladd points out that tightly edited poll results are usually
oversimplified
poll results.
Another point of disjuncture between journalism and polling is that "good
news reporting has focus and arrives at relatively clear and
unambiguous
conclusions. In contrast, good opinion research typically reveals such
characteristics of popular thinking as tentativeness, ambivalence,
uncertainty, and lack of information or awareness" (1980 p. 577). The
analytical techniques applied to newspaper-conducted poll data are
typically
restrained to crosstabulation with limited demographic variables:
techniques
often inadequate to the task of discovering subtle relationships.
Syndicated columnist Nicholas Von Hoffman (1979) minces no words as he
criticizes newspaper polls, charging them with one of the mortal sins
of
journalism: the making of news. Von Hoffman argues that it would be
indefensible for journalists to use the checkbook to create a news event.
Yet, according to Von Hoffman, that is exactly what journalists are
doing
each time they conduct an opinion poll. "The big news organizations,
therefore, are making their own news and flacking it as though it were an
event they were reporting on over which they had no control, like an
earthquake or a traffic accident" (p. 573).
It is another working journalist, Bill Kovach (formerly New York Times editor
now curator of the Nieman Foundation at Harvard) who offers
somewhat lonely
praise in the academic literature for journalists' use of opinion
polling
(1980). In his address to the 35th Annual Conference of the American
Association for Public Opinion Research, Kovach admits to the difficulty he
had in the task of recalling how things were before journalists
embraced
opinion polling:
But it didn't turn out to be that simple because the polling tool has been so
completely factored into our decision-making process, especially
in political
reporting, that I had difficulty remembering how we worked
before we had this
tool. How did we report the issues? How did we determine their
rank and
importance in electoral matters? (p. 567)
He goes on to describe the beneficial attributes of the relationship between
the newspaper and opinion polling:
As a result of polling we have, I believe, been able to more adequately
examine political issues, to check our perception of the relative
importance
of the issues, to discover D through open-ended questions D
issues that
otherwise might have escaped our attention, and to test suspicions
or
tentative notions about public attitudes and behavior. All of these
analyses
have been invaluable in the allocation of limited resources.
Time and again,
poll results have helped us decide how and where and in what
numbers to
commit our reporters, and in most cases, these have been more
useful,
productive, and informative decisions. (p. 568)
Overall, Kovach argues that opinion poll data have helped New York Times
reporters and editors to understand the audience, their interests,
and the
social world in which they live. Yet Kovach recognizes many of the
criticisms
leveled against media opinion polling. He admits that the horse-race
has
become too heated. He also admits that editors feel compelled to use
their
expensive poll results and sometimes tend to give in-house generated
results
preference over the observations of outside pollsters.
Clearly, media opinion polling is not without its pros and cons. Most of the
criticism of the practice revolves around matters of a technical
nature. The
critics are often academic and professional pollsters who have
studied public
opinion and polling in great depth and have developed it to a fine
and
complex art. From their perspective, the newspaper-conducted opinion poll
is
dangerously amateurish. Is social research really another thing that
journalists should be doing in a hurry? Their concern is certainly not
without basis. On technical grounds, mishandled data reported as the
"public's will" is a fact error of some magnitude: a misquote of societal
proportions. On conceptual grounds, even the possibility that poll
respondents are reacting to the poll question rather than reporting an
actual opinion that was held before the question was asked can turn the
validity of "public opinion" on its head.
But these arguments are usually restricted to the halls of the academy. The
fact of the matter is that in the "real world" public opinion
polls are a
popular aspect of today's media environment and, despite what problems
they
may possess or cause, at the very least represent another voice in
the rancor
of public debate D another offering in the marketplace of ideas.
THEORETICAL BASIS, METHODS, AND HYPOTHESES
As Noelle-Neumann (1980) suggests, an important consequence of journalists
using poll results to shape news judgment would be a partial
surrender of the
agenda-setting power of the media. This observation suggests a
viable
approach to the empirical question.
If any empirical evidence has emerged showing that media opinion polling is
shaping later media coverage it has come from the area of
agenda-setting.
Rogers, Dearing, and Chang (1991) suggest that there was a
relationship
between media polling and subsequent media coverage of AIDS. And Trumbo
(1995) comes to a similar conclusion through an examination of the issue
of
global warming.
It must be pointed out that both of these studies involve national issues
with highly unique aspects D a far cry from the variety of
community or
state-level issue that often occupies local opinion polling.
Nonetheless,
agenda-setting may be a useful point of departure for the question at
hand.
In recent years the agenda-setting literature has been moving toward
the
perspective that agenda-setting operates in accordance with a general
systems
model. One may no longer stop with the idea that the media may be
good at
telling the public what to think about; the public may also be
credited with
the ability to tell the media what to think D or report D about (see
Smith,
1987; Rogers, Dearing, and Bregman 1993; Kosicki, 1993; McCombs,
1992; Rogers
and Dearing, 1988; Trumbo, 1995).
The present investigation of the relationship between newspapers and opinion
polling will borrow from the systems-oriented agenda-setting
tradition and
allow that the typical agenda-setting hypothesis may be reversed. The
generalized research question asks if the newspaper's agenda remains
unchanged following an opinion poll or if it tends to move into synch with
the poll results. But this investigation's operationalization of the
public
agenda (discussed below) falls somewhat short of the idealized
concept of
"salience" that is central to agenda-setting. It must therefore be
emphasized
that the newspaper's agenda is not being placed as a dependent
variable to
the public agenda D as agenda-setting would be prone to do D but the
two are
being examined together in an effort to look for a relationship. The
causal
assumption of agenda-setting is not embraced.
Returning to the focus of this investigation, we ask specifically about the
relationship between newspaper-conducted polls and the
newspaper. Of all the
possible impacts that media-conducted polls could have on the media
the most
significant might be a subtle influence on the news judgment of
journalists.
This begs the empirical question: is newspaper opinion polling measurably
linked to subsequent newspaper news judgment? As Atkin and Gaudino
(1980)
observe, the literature offers only anecdotal evidence on this
question. To
date, there have apparently been no attempts to empirically
investigate the
hypothesis that the way a newspaper covers the issues is independent
of the
results of its own opinion polling. Judging from the nature of the
commentary
found in the academic literature concerning the relationship between
newspapers and opinion polling, one would hypothesize that a strong and
fairly easily detected empirical relationship should exist.
Reprints of news articles based on Iowa Poll results are republished annually
by The Register. This collection was browsed for potentially
interesting
configurations of questions. In early 1985 The Register executed a
round of
polling to see how Iowans would vote on a set of 12 upcoming
legislative
items. Following the legislative action, The Register republished the
results
in a table showing how the poll results compared with the
legislative results
(as it turned out, passage by the Iowa Poll predicted passage by the
legislature for less than half of the items). In late 1988 and early 1989
The
Iowa Poll again asked a set of questions asking how Iowans would
vote on 10
upcoming legislative items (they were never presented as a group
with a
comparison to law-maker's actions). The results of these two polls are
reported in Tables 1 and 2.
In each case it is possible to create a rank order of issues from the poll
results. Ranking is achieved by means of the percentage of
respondents
"voting" in favor of an issue, creating an index of popular support. For
the
1985 case, questions were asked in the month of February. For the
1988-1989
case, questions were asked in two rounds spanning December 1988 to
February
1989. This presents some non-comparability in the two cases.
However, no
other set of 10 questions were asked during the time frame in which an
index
is available for examining The Register's news coverage (10 are
required for
the statistical operations).
It must be noted that The Des Moines Register Index was compiled by the Parks
Library at Iowa State University until June of 1992 when a
full-text CD-ROM
product became available for The Register. Unfortunately, the
electronic
database cannot be searched by topic, only by key word. Since the two
methods
are by no means comparable, this investigation was thus constrained
from
using poll data more current than late 1991.
The Des Moines Register Index was then consulted to acquire a story count by
poll topic for the 9 months prior to and the 9 months following
both poll
periods. These data were broken down into cumulative 3, 6, and 9 month
segments so that the time span of any relationship might be evaluated. To
evaluate the relationship, Spearman's rank order correlation, rho
(rs), is
calculated between the newspaper's agenda before the poll and the
poll's
index of support. This is then compared with the rank order correlation
between the newspaper's agenda after the poll and the support ranking.
For
cases where more than one tie occurs in the rank ordering, Kendall's
tau-b is
used in place of Spearman's rho. The research question asks if there
is a
stronger relationship between the Register's agenda and the Iowa Poll
after
the poll. Restated, the correlation should be significantly stronger
after
the poll than before. Since the strengthening could be in either
direction,
this presents the two-sided hypothesis:
H: rs (newspaper before poll w/poll) _ rs (newspaper after poll w/poll)
RESULTS
It was in 1943 that Des Moines Register president Mike Cowles, a long-time
friend of Iowa native George Gallup, initiated the Iowa Poll.
Newspapers in
Dallas and Austin had started the Texas Poll in 1940, but suspended
their
operation from 1941-1946 while the founder, Joe Belden, served in the
war.
This gives the Iowa Poll the distinction of being the oldest
continuously-conducted state newspaper poll in the nation. The
Minneapolis
Star Journal and Tribune, another Cowles family holding, began the
Minnesota
Poll in 1944 (Fogarty, 1993).
The motives for starting the poll predominately revolved around marketing. In
a fiftieth year retrospective on the Iowa Poll, The Register's
Thomas
Fogarty writes:
But the chief motivation for the Iowa Poll was by no means philanthropic. It
was viewed as a feature that would set The Register apart from
other papers
and boost advertising and circulation.(p. 8)
Today, the Iowa Poll is solidly grounded on the editorial side of the
newspaper. The poll is currently directed by Sharon Pilmer, who came
to the
Iowa Poll with a background in both marketing and opinion research.
She is
officially a part of the marketing department at The Register, but
works
exclusively with the editorial side of the organization on matters
pertaining
to the poll. She works closely with Fogarty, who has been The
Register's poll
writer for the past five years. They develop and implement the
questions
together. A survey organization in Dubuque, Iowa, then executes the
phone
interviews (typically about n = 800) and returns the data to Pilmer
who
performs analysis and provides Fogarty with a technical report on which
to
base his stories. Statistical analysis generally involves cross
tabulation
with demographic variables and with congressional districts. Data is
deposited with The Roper Center (Pilmer, 1994).
After some conferring between Pilmer and Fogarty over details, the poll
stories hit the paper. There is no advance report prepared on the
poll
results for pre-publication newsroom circulation and the poll results are
not
used by any other department at the newspaper. In fact, the poll
does not ask
respondents if they are subscribers.
Over the past 50 years the Iowa Poll has asked Iowans a wide variety of
questions. Much of the material has been in the spirit of
entertainment,
covering topics such as what names Iowans give to their cars, whom
Iowans
trust the most (God number one, the Iowa State Highway Patrol number
two), or
reporting that 5% of Iowans personally grieved the death of
punk-rock star
Sid Vicious (Cutler, 1990).
But the majority of the opinions tapped by the Iowa Poll have been political
in nature. And clearly, this is the aspect of the Iowa Poll that
The Register
now enjoys most. A sizable portion of the fiftieth anniversary
article
published on December 19, 1993, was devoted to celebrating how the Iowa
Poll
was "ahead of the curve" in terms of predicting the actions of state
legislators. A score-card graphic shows how that, over the years, the Iowa
Poll has reported a majority of Iowans in favor of one issue or
another and
how some time later it was law. The plebiscitary flavor of the Iowa
Poll is
very real.
But what impact has the Iowa Poll had on the representation of the issues in
the pages of The Register? The Iowa Poll represents an excellent
opportunity
to look for such evidence. The Register circulates throughout the
state and,
despite circulation cutbacks in the 1980s, faces little credible
challenge to
its motto: "The Newspaper Iowa Depends Upon." The Register's sole
proprietorship of the Iowa Poll and its lack of a competitor eliminates a
host of uncontrolled variables in the relationship between The
Register and
the Iowa Poll.
Tables 1 and 2 detail the issues over which The Register polled Iowans on
early in 1985 and in the 1988-1989 period. This set of issues
seems fairly
typical of those that communities face on a regular basis: ranging
from
taxation, to the regulation of vice, to larger initiatives such as
establishing a World Trade Center or beginning a state-sponsored lottery, to
issues involving insurance, education, and government.
Table 3 presents the results of the analysis for the 1985 case. Regardless of
the time span examined (3, 6, or 9 months), there is no
significant
correlation between ranking in news coverage and earlier poll results.
During
the post-poll period only one time frame achieved any generally
accepted
degree of significance. This correlation occurs in the 9 month frame.
Taking
the difference between the before and after correlation shows that
only the 9
month time frame even approached statistical significance at any
generally
accepted level. Table 4 presents the similar analysis of the
1988-1989 case
were no statistically significant relationships are found.
Had the hypothesis been directional with an alpha relaxed to .1, the
difference found in the 9 month frame for the 1985 case would have
achieved
significance. This might have been justified considering the
difficulty of
achieving statistical significance with such a small n. For example,
to
achieve a .05 level of significance in the rank order correlations a
coefficient of about .66 is required. However, to achieve the same
significance for the difference between correlations the coefficient must be
as high as .77 (3 observations are surrendered in degrees of
freedom).
Thus, for the case of 1989, the null hypothesis of no relationship between
The Register and the Iowa Poll cannot be rejected. Similarly for
the case of
1985, the null hypothesis cannot be formally rejected. However, the
nearly
significant relationship observed in 1985 can be further investigated
for
information that might bear on the research question.
The one weak (or near) correlation is worth looking at in greater detail.
Table 5 ranks The Register's coverage of issues for the 9 months
prior to the
February 1985 poll, the issues in the poll, and The Register's
coverage for
the 9 months after the poll. The inverse correlation that developed
after the
poll is somewhat apparent. Table 6 provides further detail
concerning the
change in ranking for each issue as well as the amount of coverage
each issue
received in the news.
It can be seen in Table 6 that the news coverage for these issues generally
increased during the post poll period, as would be expected as
the issues
became topics in the legislative session. It is worth asking if there
is any
relationship between the amount of change an issue experienced in
ranking and
the amount of coverage it received in the news. To investigate for
any such
pattern, the amount of rank change in both absolute value and in
directional
value was correlated with the amount of coverage before, after, in
total, and
its difference between before and after. Two significant
correlations were
detected. The absolute degree of change and the number of stories
before the
poll correlated ( r = - .61, p = .04), and the directional degree of
change
correlated with the difference between before and after poll story
counts (r
= . 59, p = .04).
Overall, according to these correlations, issues that were dynamic in
affecting a change in rank order correlation (regardless of direction)
were
the ones that received relatively less news coverage during the 9
months
before the poll. Further, issues that rose in rank are associated with
issues
that had a large difference in before/after poll news coverage (and
vice
versa). This identifies three issues that played a large role in
shifting the
rank correlation: local option tax, wine sales, and sales tax. They
all rose
in ranking after the poll and they all went from little or no
pre-poll
coverage to more substantial levels of coverage after the poll.
From Table 5 it can also be seen that these issues were tied in the pre-poll
period. Therefore, a significant amount of the change in
correlation can be
attributed to tied and non-attended issues subsequently receiving
news
attention, breaking ties, and becoming more closely aligned to the poll
ranking. Does this suggest that the changing relationship that is
suggested
is a statistical artifact? Not necessarily, because the correlation
between
coverage ranking and poll ranking after the poll is weakly
significant in
itself (r = -.51, p = .09).
DISCUSSION
Considering the content of the literature on newspapers and polling and the
comments made by some journalists, the lack of a stronger
relationship is
somewhat surprising. This study may suggest that the problems
attributed to
the newspaper-poll relationship, at least with regard to news
judgment, are
somewhat overstated. Alternatively, The Register may represent an
atypical
case. Generalizing from the population to an individual is always
risky.
But some weak relationship may in fact exist between poll results and news
judgment. The sign of the correlation between the poll and the
1985 coverage
is of some additional interest. The negative correlation suggests
that for
issues on which there was strong public support there was relatively
little
coverage by The Register. This makes some sense if recast in terms
of
conflict. Those issues for which there is less popular support must be the
issues over which there is greater opposition and conflict. It should
follow
that these would then be the more salient issues for news coverage.
Further,
the legislature is certainly having a strong impact the news agenda.
The
legislative proposals for which there is less popular support should
also be
more contentious in the legislature and thus must also be the issues
that
receive greater legislative debate D which once again would garner
more news
coverage.
The question of how much impact the poll has in the newsroom can also be
approached in a more qualitative manner. Fogarty suggests that in
some
circumstances the Iowa Poll does figure prominently into the news
judgments
at The Register (Fogarty, 1994). He describes how the poll recently
asked
Iowans if they felt the media was over-covering the O.J. Simpson
affair.
Resoundingly, Iowans expressed displeasure over the amount of attention
being
given to the celebrity trial. The Iowa Poll is a high profile item
at The
Register, so it was with some surprise that Fogarty found that the
O.J. poll
was being run inside of the paper rather on the front as usual.
However, the
paper then reduced the prominence of the trial coverage. In this
case the
poll was used as a conscious and direct instrument to evaluate and
shape news
judgment.
According to Fogarty, the effects of the poll can be most dramatic in the
political arena. On two occasions Iowans found unusual candidates
announcing
a run for public office. In one case, a local businessman-celebrity
announced
for governor, in another case a plastic surgeon attempted to unseat
an
incumbent Congressman. In both cases The Register was initially unsure of
how
to address the viability of the atypical candidacies. In both cases
they
polled 800 random Iowans. As a consequence of Iowans not taking his bid
seriously, the businessman-celebrity found less coverage in The Register
,
but the plastic surgeon found his campaign getting more attention
because the
random Iowans found him appealing as a candidate. Fogarty sums it
up: "So,
yes, I think there is quite a strong impact on the news coverage,
depending
on what we come out with on the poll" (1994).
The consequences of the poll results, especially in terms of the legislative
agenda, must be felt by Iowa's law-makers as well. When not
writing Iowa Poll
stories, Fogarty is a political reporter for The Register, and is
thus well
placed to comment on the relationship between the Iowa Poll and the
state's
legislators:
I think they do pay attention very intently. I don't think they always follow
it mindlessly. But one of the things I do here is cover the
legislature and I
see it time and again. When we do our legislative polling I'll
leave for a
couple of days to write the stories, but then I'm usually right
back up there
[at the Statehouse] and I watch it play out. I think [the poll]
does have an
impact. For better or worse, it has a great impact on public
policymaking.
(Fogarty, 1994)
Clearly, it must be concluded from the data and other evidence presented here
that a complex relationship exists between The Register and its
Iowa Poll and
that this relationship cannot be seen as being distinct from the
legislative
and social processes that are also tied to the relationship. A more
sophisticated study might untangle the relationships in which a legislative
agenda develops, a newspaper conducts a poll on that agenda, and
then
legislators and journalists alike take some account of the poll results as
the set of public issues are debated.
CONCLUSIONS
It is important to note two limitations of this study. First, this
investigation does not present a completely unclouded picture of the
relationship between The Des Moines Register and its Iowa Poll. While a
marginal relationship may exist, it cannot be separated from the effects
of
other unmeasured variables. In reality the news is a complex of
relationships not easily influenced by a single force. For example, the
amount of attention that these issues received in the legislature was
not
measured but must have a much stronger influence on news coverage than
the
results of a poll.
Second, this study does not present findings that are easily generalizable to
the universe of newspapers. The Iowa Poll is somewhat atypical
of all
newspaper polling in that it is a regular and high profile state-wide
polling
operation. Most newspaper polls are done on a more sporadic basis
and
probably with less of a professional orientation than is the case for the
Iowa Poll.
But despite these limitations, this investigation does provide some valuable
empirically-based insight into one case of the newspaper opinion
poll and
will serve to enhance the small body of literature on this topic D a
literature generally lacking in empirical investigation.
It is worth concluding this article by examining the more general and
forward-looking question that underlies the pro/con debate over media
opinion
polling presented earlier. If the media's interest in opinion
polling can be
taken as a given, and if it can be similarly taken that journalists
do pay
some attention to their own polls, then the question remains: Can
this
relationship produce good journalism? While polemicists argue the good
and
evil of media polling it may be more constructive to ask how such
polling can
be used to best advantage for journalism and the communities that
journalists
serve. Jay Rosen offers entry into this line of thought:
Journalists made a mistake years ago when they more or less accepted the
results of polls as their working definition of public opinion. The
mistake
had little to do with the familiar problems involving
volatility,
reliability, and margin of error in polling. These are technical matters.
By
getting into the business of measuring public opinion,
journalists abandoned
their duty to improve it. This, I believe, was their mistake.
(Rosen, 1992,
p. 25)
Rosen argues that the greatest problem with polls and journalism involves how
journalists use polls. In the name of objectivity, journalists
distance
themselves from the political debates within their communities by making
public opinion a quantified and therefore "objective" entity. Rosen
echoes
the sentiment expressed by Daniel Yankelovich in his Coming to Public
Judgment: Making Democracy Work in a Complex World. The critical point is
that there is an important distinction between public opinion and
political
judgment, and that endowing opinion poll returns with the status of
a
community's political judgment is inadequate. Rosen offers some advice:
Suppose, then, we understand public opinion not as a verdict, but as a
process by which a political community comes to understand and debate
its
choices. Under this definition the press has an important task: to
improve
the chances that public opinion will evolve into public judgment.
The real challenge in improving public opinion is not to add more
information, or to grab public attention for a neglected issue. It is
to
encourage people to state their concerns, clarify their values,
order their
priorities, and appreciate the consequences of the views they
hold. These
improvements can only come about through the process of
deliberation and
debate. (Rosen, 1992, pp. 26-27)
Within this frame, the task of the media polling unit shifts from that of
generating information that is presented as an objective measure
of opinion
to the task of generating material over which public discourse might
then
proceed. Rosen argues that it is important for journalists to work
actively
to facilitate that discourse. By taking the results of surveys back
to the
people for additional consideration and discussion, rather than
reporting
them as fact, journalism can take a step toward re-engaging itself in
the
life of the community and its politics.
Rosen calls this kind of involvement in community affairs public journalism
(1994). Considering the sagging political participation and
falling
newspaper readership that seems to characterize this period of our
democracy, public journalism may be an idea that journalism and especially
journalist-pollsters should consider, if not embrace.
REFERENCES CITED
Atkin, Charles K., & Gaudino, James (1984)."The impact of polling on the mass
media." Annals AAPSS: 472: 119-128.
Crespi, Irving (1980)."Polls as journalism." Public Opinion Quarterly 44:
462-476.
Cutler, Blayne (1990)."Flamingos of the heartland." American Demographics
12:50-51.
Demers, David P. (1987)."Use of polls in reporting changes little since 1978."
Journalism Quarterly 64:839-842.
Dionne, E. J., Jr. (1992)."The illusion of technique: The impact of polls on
reporters and democracy." In Thomas E. Mann & Gary R. Orren (Eds.)
Media Polls
in American Politics.Washington DC: The Brookings Institution.
Fogarty, Thomas. A. (1993, Dec. 19)."Half a century of listening to Iowa." The
Des Moines Register, A-1, A-8.
Fogarty, Thomas A. (1994, Dec. 12).Telephone interview.
Gollin, Albert E. (1980)."Exploring the liaison between polling and the press."
Public Opinion Quarterly 44: 445-461.
Ismach, Arnold H. (1984)."Polling as a news gathering tool." Annals AAPSS 472:
106-118.
Kovach, Bill (1980)."A user's view of the polls." Public Opinion Quarterly
44:567-571.
Ladd, Everett Carl (1980)."Polling and the press: The clash of institutional
imperatives." Public Opinion Quarterly 44:574-584
Meyer, Philip (1990)."Polling as political science and polling as journalism."
Public Opinion Quarterly 54: 451-459.
Miller, M. Mark, & Robert Hurd (1984). "Conformity to AAPOR standards in
newspaper reporting of public opinion polls." Public Opinion Quarterly
46:243-249.
Noelle-Neumann, Elizabeth (1980)."The public opinion research correspondent."
Public Opinion Quarterly 44:585-597.
Paletz, David L., Jonathan Short, Helen Baker, Barbara Campbell, Richard
Cooper, and Rochelle Oeslander (1980). "Polls in the media: Content,
credibility, and consequences. Public Opinion Quarterly 44:495-513.
Pilmer, Sharon (1994, Dec. 13).Telephone interview.
Rogers, Everett M., & Steven H. Chaffee (1994). "Communication and Journalism
from 'Daddy' Bleyer to Wilbur Schramm. A Palimpsest." Journalism
Monographs
148.
Rogers, Everett M., & James W. Dearing (1988)."Agenda-setting research: Where
has it been, where is it going?" In J. Anderson (Ed.), Communication
Yearbook
11. Newbury Park, CA: Sage.
Rogers, Everett M., James W. Dearing, & Soonbum Chang (1991)."AIDS in the 1980s:
The agenda-setting process for a public issue." Journalism
Monographs 126.
Rosen, Jay (1992)."Politics, vision, and the press: Toward a public agenda for
journalism." In Jay Rosen & Paul Taylor (Eds.) The New News v. The
Old News.
New York: The Twentieth Century Fund Press.
Rosen, Jay (1994)."Public Journalism: First principles." In J. Rosen (Ed.), The
Public: An Enigma for Journalists. New York: Project on Public Life
and the
Press.
Smith, Kim A. (1987)."Newspaper coverage and public concern about community
issues." Journalism Monographs 101.
Traugott, Michael (1992)."The impact of media polls on the public." In T. E.
Mann and G. R. Owen (Eds.), Media Polls in American Politics (pp.
125-149).
Washington, DC: The Brookings Institution.
Trumbo, Craig (1995)."Longitudinal Modeling of Public Issues with the
Agenda-Setting Process: The Case of Global Warming." Journalism Monographs
152
(in press).
Von Hoffman, Nicholas (1979)."Public opinion polls: Newspapers making their own
news?" Public Opinion Quarterly 44:572-573.
Weaver, David H., & Maxwell E. McCombs (1980). "Journalism and social science:
A new relationship?" Public Opinion Quarterly 44:477-494.
Table 1. February 1985 Iowa Poll.
Issues and percentage for and against in the Iowa Poll.
Legislative action is indicated: Passed or
Failed.
ISSUE
pro
con Leg.
30 day suspension drunk driving
90%
09% F
Emergency farm loans
84
11 F
School expense tax deduction
79
16 F
Restrict public smoking
79
19 F
Raise drinking age to 21
76
22 F
Begin a state lottery
74
18 P
Sell wine in grocery stores
56
40 P
Mandatory seat belts
53
45 F
Raise gas tax to 3 /gal.
52
46 P
Allow local option taxes
37
56 P
Increase sales tax to 5%
24
74 F
Establish a World Trade Center
23
70 P
The Iowa Poll, conducted Feb. 6-21, 1985, asked the
following: Following are several proposed laws that
the Iowa Legislature is considering. For each
proposal, tell me if you favor or oppose it.
Restrict smoking in restaurants and other public
buildings to certain areas. Raise the gasoline tax
by 3 cents a gallon to help fund highway repair and
construction. Require automobile drivers to fasten
their seat belts. Allow the sale of bottled wine in
private stores. Raise the drinking age from 19 to
21. Allow cities and counties to levy local sales
or income taxes. Increase the state sales tax from
4 to 5 percent. Require 30-day suspension of
driver's license upon arrest for drunken driving.
Allow a state income tax deduction of up to $1,000
for each child in kindergarten through 12th grade.
Should Iowa state government help farmers obtain
loans so they can afford to plant crops this
spring? Do you favor or oppose a state-operated
lottery in Iowa?
The Iowa Poll, directed by Glenn Roberts, is based
on 801 interviews with Iowans 18 years of age or
older. Professionally trained interviewers
contacted households with telephone numbers
randomly selected by a computer, eliminating
interviewer's choices in selecting persons to be
interviewed.
Percentages based on the full poll sample are
subject to a maximum margin of error of plus or
minus 3.5 percentage points (from the Des Moines
Register).
Table 2. Dec. 1988 -
Feb. 1989 Iowa Poll.
Issues and percentage
for and against in the
Iowa Poll.
Legislative action is
indicated: Passed or
Failed.
ISSUE
pro
con
Leg.
Require decomposable
trash bag
93%
05% F
Unisex insurance
86
12 P
Campaign spending
reform
84
12 F
Raise state's minimum
wage
78
19 P
Tax on reusable
containers
76
22 F
Renew the state's
lottery
72
19 P
Mandatory health
insurance
70
26 F
Open enrollment
59
34 P
Allow riverboat
gambling
47
49 P
Reorganize university
programs
41
45 F
The
Iowa
Poll,
condu
cted
Dec.
5 -
13,
1988,
asked
the
follo
wing:
Curre
ntly
there
are
no
limit
s on
how
candi
dates
for
publi
c
offic
e
can
spend
campa
ign
funds
.
Shoul
d
the
state
set
limit
s?
Witho
ut
any
actio
n by
the
Legis
latur
e,
the
Iowa
lotte
ry
will
end
in
19
90.
Do
you
think
the
Legis
latur
e
shoul
d
exten
d
the
lotte
ry?
The
state
Legis
latur
e
will
consi
der
plans
to
reorg
anize
the
three
state
unive
rsiti
es
to
reduc
e
dupli
catio
n.
Criti
cs
say
this
will
hurt
the
acade
mic
quali
ty.
Would
you
favor
or
oppos
e
reorg
aniza
tio
n?
To
prote
ct
the
envir
onmen
t
and
to
incre
ase
corn
sales
,
the
state
Legis
latur
e
will
consi
der
a
propo
sal
to
outla
w
the
use
of
plast
ic
garba
ge
bags
that
don't
disso
lve
in
landf
ills.
Inste
ad,
the
Legis
latur
e
will
requi
re
the
use
of
bags
made
from
corns
tarch
that
will
disso
lve
in
the
envir
onmen
t.
Would
you
say
you
favor
or
oppos
e
this
propo
sal?
(from
the
Des
Moine
s
Regis
ter)
The
Iowa
Poll,
condu
cted
Feb.
27 -
March
8,
1989,
asked
the
follo
wing:
Do
you
favor
or
oppos
e
requi
ring
all
emplo
yers
in
Iowa
to
provi
de
healt
h
insur
ance
for
their
worke
rs?
Do
you
favor
or
oppos
e
addin
g a
1-cen
t
tax
to
every
conta
iner
of
pop
and
beer
to
cr
eate
a
fund
to
prote
ct
Iowa'
s
envir
onmen
t?
Do
you
favor
or
oppos
e
requi
ring
insur
ance
compa
nies
to
treat
men
and
women
equal
ly
in
the
rates
they
pay
for
insur
ance?
Do
you
favor
or
oppos
e
legal
ized
river
boat
gambl
ing
on
the
Missi
ssipp
i
River
along
Iowa'
s bo
rder?
Do
you
favor
or
oppos
e
raisi
ng
the
state
's
minim
um
wage?
Would
you
favor
or
oppos
e an
open
enrol
lment
plan
that
would
allow
paren
ts
to
send
their
child
ren
to
the
publi
c
schoo
l of
their
choic
e?
(from
the
Des
Moine
s
Regis
ter)
Table 3. Rank Correlations Before and After February 1985 Iowa Poll.
Utilizing 3, 6, or 9 month time frames (cumulative), rank correlations are
calculated
between amount of the Register's coverage and the poll ranking
(by percentage favoring
issue). Due to ties, Kendall's taub is used for all before
correlations, and Spearman's
rho for all after correlations. Significance for the difference
in the before and
after correlations are then calculated. All tests are two-tailed.
9 month frame / sig.
6 month frame / sig.
3 month frame / sig.
Before
.02
n.s.
.14
n.s.
- .08
n.s.
After
- .51
p = .09
- .48
p = .11
- .29
n.s.
Difference
.53
p = .13
.62
n.s.
.21
n.s.
Table 4. Rank Correlations Before and After Dec. 1988 - Feb. 1989 Iowa Poll.
Utilizing 3, 6, or 9 month time frames (cumulative), rank correlations are
calculated
between amount of the Register's coverage and the poll ranking
(by percentage favoring
issue). Due to ties, Kendall's taub is used for all before
correlations, and Spearman's
rho for all after correlations. Significance for the difference
in the before and
after correlations are then calculated. All tests are two-tailed.
9 month frame / sig.
6 month frame / sig.
3 month frame / sig.
Before
.12
n.s.
.02
n.s.
.07
n.s.
After
.13
n.s.
- .12
n.s.
- .16
n.s.
Difference
.01
n.s.
.14
n.s.
.23
n.s.
Table 5. Significant Rank Order Correlations: Issues
Ranking of issues for Register coverage in 9 months before and
9 months after February 1985 Iowa Poll (% favoring). Rank of 1
is greatest coverage or most in favor. * indicates sets of ties.
rank
Before
Poll: Favoring
After
1
State Lottery
Drunk driving
State lottery
2
Farm loans
Farm loans
Trade center
3
Trade center
School tax cr.
Local option tax
4
Seat belts
Public smoking
Farm loans
5
Drunk driving
Drinking age
Sales tax
6
Drinking age
State lottery
Wine sales
7
*School tax cr.
Wine salesSeat belts
Gas tax
8
*Gas tax
Seat belts
Seat belts
9
**Public smoking
Gas tax
Drunk driving
10
**Sales tax
Local option tax
Public smoking
11
***Local option tax
Sales tax
School tax cr.
12
***Wine sales
Trade center
Drinking age
Table 6. Change in rank, number of stories by issue.
For 9 months before and after February 1989 Iowa Poll, taking
ties into consideration
change in rank
number of stories
change
ISSUE
before
after
total
+8
Local option tax
0
43
43
- 6
Drinking age
7
2
9
+ 6
Wine sales
0
21
21
+ 5
Sales tax
1
22
23
- 4
Drunk driving
10
7
17
- 4
School tax credit
3
4
7
- 4
Seat belts
12
8
20
- 2
Farm loans
25
25
50
+ 1
Gas tax
3
18
21
+ 1
World Trade Center
21
78
99
-1
Public Smoking
1
6
7
0
State Lottery
41
148
189
124
382
506
|
