Skip Navigation

Federal Communications Commission

English Display Options

Commission Document

Media Ownership Study 6-Submitted Study

Download Options

Released: June 15, 2011
















Less of the Same:

The Lack of Local News on the Internet




Matthew Hindman
The George Washington University




Author Information: Matthew Hindman, School of Media and Public Affairs, The George
Washington University, 805 21st St NW, Washington, DC 20052. Email:
Hindman@GWU.edu. Thanks are due to Jonathan Levy, Tracy Waldon, and Jessica
Almond, who oversaw this report for the FCC and who provided numerous helpful
insights and suggestions that improved the final version. Any remaining errors or
infelicities are my own.

HINDMAN
LESS OF THE SAME: LOCAL NEWS ON THE INTERNET
2

Introduction


Perhaps no part of the American media environment is as little understood as Web
based local news. The relative importance of Internet news has grown steadily;
recent Pew studies have found that more Americans now get their news from the
Web than get it from print newspapers (Purcell et al. 2010). But even as the Web is a
larger slice of the American news diet, systematic data on local news has been
scarce. Even very basic questions have remained unanswered. How many online
local news outlets are there in a typical media market? Are successful local news
sites new, or just online versions of traditional media? How much competition is
there in online local news? Just how much attention do local news Websites receive?

These questions have acquired particular urgency since 2007, as the newspaper
industry has faced a financial crisis. Local newspaper reporters produce most
journalism in the United States, but that economic model is now imperiled.
Newspaper advertising dropped by an inflationadjusted 54 percent between 2005
and 2009--and though the decline has recently slowed, there are no signs of
recovery (Newspaper Association of America 2010). Eight major newspaper chains
entered bankruptcy between 2008 and 2010, and papers such as the Seattle Post
Intelligencer and the Rocky Mountain News shut down their presses (Kirchoff 2010).
Newsroom staffing at U.S. papers is now less than 70 percent of what it was in 2001
(PEJ 2011).

The Internet's potential to expand local news voices has been of interest to
policymakers and regulators for more than a decade. In 2003, online media diversity
featured prominently in FCC and congressional debates about broadcast ownership
regulation (see, for example, the discussion in Noam 2009). The federal courts have
highlighted the same questions, focusing on the (disputed) ability of the Internet to
provide greater viewpoint diversity for local news (e.g. Prometheus v FCC, 2004).
More recently policymakers at the FCC, in congress, and at the Federal Trade
Commission have studied the Internet's ability to sustain local journalism even as
newspapers struggle financially.

But for all of the discussion of local news online, there has been little systematic
evidence about the local news environment on the Web. Arguments have been
waged mostly with anecdotes and assumptions instead of comprehensive data.

This study aims to change that. Using comScore panel data that tracks a quarter of a
million Internet users across more than a million Web domains, this paper examines
online local news within the top 100 US television markets. I identify and analyze
1074 local news and information sources across these 100 markets, studying their
audience reach, traffic, and affiliation (or lack thereof) to traditional media. I also
look at concentration in local online news markets, and conduct a census of
Internetonly local news sites that reach more than a minimum threshold of traffic.

FCC PUR11000027

DRAFT SUBMITTED 4.6.2011

HINDMAN
LESS OF THE SAME: LOCAL NEWS ON THE INTERNET
3
The breadth and the marketlevel granularity of the comScore data makes this study
the first comprehensive look at of Internetbased local news. The portrait that
emerges is surprising. New, Webnative news organizations are nearly absent from
this traffic data. Local news on the Web is fundamentally about consuming less news
from the same old sources. Understanding the local news landscape online has
profound implications for policymakers, journalists, and local selfgovernance in the
21st century.


Data and Methodology

As noted above, this study relies centrally on data provided by comScore, a large U.S.
Web measurement firm. comScore tracks the browsing behavior of a large panel of
Internet users with userinstalled software, and its coverage is exceptionally broad:
as of July 2010, the firm reported tracking traffic to 1,049,453 Web domains.
Previous research has used aggregate U.S. Web usage data, in which only the largest
local sites in the largest local markets could be studied. For this study, full data from
February, March, and April 2010 was purchased for the 100 largest US broadcast
market areas, and provided to the author as governmentfurnished information.
These 100 markets contained a monthly average of 253,953 comScore panelists.
Because comScore strives for a nationally representative sample, the number of
panelists varies according to a market's size, from 19,998 in the greater New York
City market (the largest in the sample) to 647 in the BurlingtonPlattsburgh market
(the smallest by panel size, and one of the smallest by population). The median
market by panel size, Little RockPine Bluff, had 1,606 panelists. In most cases sites
are tracked at the domain level: all pages on example.com would be considered
together as traffic to a single site. With particularly popular sites, however,
comScore tracks different parts of the domain both individually (for example,
images.google.com and maps.google.com) and collectively (all Googleowned
properties together).

Thanks to the breadth and detail of the comScore data, this study can provide the
first comprehensive, nationwide look at the state of local news on the Web. In our
case, the comScore data provides several key traffic metrics within each broadcast
market. These include: monthly audience reach, which is the portion of panelists that
visits a Website at least once in the calendar month; the number of monthly page
views that a site receives; monthly minutes, which measure the time spent on a site;
and the number of monthly panelist sessions that a site accumulates, measuring the
number of times that a person accessed one or more of a site's pages with no more
than a 30 minutes between clicks. For each market, comScore's listings include all
sites visited by at least six of that market's panelists.

Local news, by definition, draws a larger audience within its home market than it
does nationally. A local news Website covering Seattle will have a larger audience
share in the greater Seattle area than it will in Tulsa or Toledo. This fact, along with
FCC PUR11000027

DRAFT SUBMITTED 4.6.2011

HINDMAN
LESS OF THE SAME: LOCAL NEWS ON THE INTERNET
4
the richness of the comScore data, allows us to distinguish local from national news
outlets. Local Websites are operationalized as sites that have higher levels of usage
within a given media market than they do in the rest of the nationwide sample
. How
much higher? The simplest rule is to look at any usage difference big enough that is
unlikely to have been produced by chance. The study uses a standard difference of
means test, comparing sites' mean audience reach within a market to its reach in the
rest of the national panel.

For our purposes, sites where the observed local vs. national gap in usage is at least
three times larger than the estimated standard error are examined as possible local
sites. Formally, this is equivalent to a tscore > 3. (The samples are large enough that
zscores or tscores are equivalent.) Qualitative assessments (detailed below)
suggest that this decision rule for discerning local from national content works
extremely well for the types of sites we hope to examine. A lower decision
threshold--such as 2.5--produces few additional genuinely local sites, with the
large majority of additional sites false positives. As we would expect, much lower
decision thresholds, such as t > 2, swamp the analysis with false positives. Given that
the data are roughly normally distributed, sampling error alone should produce a t
statistic > 2.0 about 2.5 percent of the time. In a data set of more than a million
observations, such a low threshold produces an unmanageably high false positive
rate.

If a differenceofmeans test provides a powerful heuristic to distinguish local from
national content, the other important task is to distinguish between sites that
provide news and sites that do not. The initial research design called for this study
to use comScore's proprietary "Web dictionary," with places tracked sites into one
of many of categories and subcategories, including a category for
"News/Information." However, comScore's coding scheme was discovered to have
significant limitations. Substantively identical news sites are often placed in
different categories and subcategories. Even within the same market, it is common
to find television station sites or newspaper sites spread across different categories
and groupings. Since it is essential that we know the affiliation (if any) between
online news sources and traditional media outlets, the comScore data needs to be
supplemented.

Because of these limits with the comScore categorization, the author himself coded
sites for news content, locality, and traditional media affiliation (more on coding
guidelines below). While the comScore data categories are imprecise and
inconsistently applied, they do provide some guidance. A newspaper site might end
up in "Regional/Local" or "News/Information: Newspapers" or "News/Information:
General News," but it is unlikely to end up in "Retail." First, 10 broadcast markets
were chosen using random numbers generated by random.org. For February,
March, and April, all sites with a tscore > 3 in these 10 markets were examined. The
comScore category was recorded for all sites that provided local news. News sites
were found in the three comScore "News/Information" subcategories, in the
"Regional/Local" category, and in the "Entertainment" category (particularly the
FCC PUR11000027

DRAFT SUBMITTED 4.6.2011

HINDMAN
LESS OF THE SAME: LOCAL NEWS ON THE INTERNET
5
"Entertainment: TV" and "Entertainment: Radio" subcategories). There was no
discernible difference between local TV stations that ended up in the
"Entertainment" category and those that ended up in "News/Information" or
"Regional/Local." Even with radio, a number of hard news stations were placed in
the "Entertainment" category.

The study also requires setting a consistent, crossmarket audience share standard
for inclusion in the analysis. Without such a standard far more local news sites will
be found in bigger markets than in smaller ones. For example, a site that got 5
panelist visits in Burlington would be omitted from the analysis, while a site that got
8 visits in New York would be included--even though the market reach of the
Burlington site is 18 times higher.

Since the study aims to provide the broadest possible survey of online local news,
this base threshold is set as low as the data allows. First, minimum standards for
inclusion in the analysis are based on monthly audience reach rather than other
traffic metrics. As will be seen below, lesstrafficked sites usually score far better on
audience reach than they do on page views or time spent on the site. Second, these
audience reach metrics should be as small as consistent crossmarket comparison
permits. Most of the smallest sites in comScore's data are visited by six of a market's
panelists in a given month, and the smallest markets have between 600 and 700
panelists. This means that 1 percent audience reach is the smallest consistent
threshold possible. Sites that reach this minimal 1 percent threshold in at least one
of the three months are included in the analysis.

Putting these requirements together means that local news site candidates are all
sites in the sample with the following characteristics:

Sites in the News/Information category, Local/Regional category, or the
Entertainment category.
Sites where the difference in audience reach within a market vs. nationally
produces a tstatistic >3
Sites that achieve 1 percent audience reach in at least one of the three
months examined.

More than 1800 sites in the data possess all three of the above characteristics. The
coding guidelines specify an inclusive definition for news sites. Websites were
counted as news and information outlets if they provided regularlyupdated
information about local news, community affairs, local officials, or issues of regional

concern. This definition is not formatdependent, and in principle it would include
sites such as local blogs. Static content by itself was not enough to count as a news
source; sites needed to have frontpage content updated within the preceding two
weeks. This coding was laborintensive, but did provide for an extremely detailed,
firsthand look at what local news sites consist of. Fortunately, the data presented
few difficult coding decisions. The large majority of sites identified by the three
FCC PUR11000027

DRAFT SUBMITTED 4.6.2011

HINDMAN
LESS OF THE SAME: LOCAL NEWS ON THE INTERNET
6
pronged test above were traditional media outlets. Ultimately, 1074 of the candidate
sites were classified as local news sites.

Because of the mandate to examine Internetonly local news sources in particular,
special care was taken to accurately record a site's affiliation with broadcast or print
sources. Every television station was confirmed to have broadcast or cable
distribution, and every print outlet was confirmed to have a paper version.

In 95 out of the 1074 news sites, higher usage levels (t > 3) were recorded in more
than one media market. These cases overwhelmingly involve a large newspaper or
(less often) a regional television station with a statewide or regional audience.1
Since the focus is on local news, rather than state or regional news, these secondary
regional markets are excluded from the definition of local content. The Seattle Times
may have aboveaverage readership in Spokane, WA, but it does not consistently
cover Spokane's local politics.

There are two prominent exceptions to this rule, however:

AL.com

and

Michigan

Live

. Both are statewide sites that feature content from newspapers in several
different broadcast markets. Participating newspapers forgo their own home pages
to host content on these statewide platforms. These outlets are thus counted as local
in every market with a participating news organization.


Metrics for Web Traffic



Before delving more deeply into the analysis of local news traffic, we should
elaborate a bit more on traffic metrics and methodology. Much discussion of online
audience, and online news audience in particular, talks about "monthly audience
reach" or equivalently "monthly unique visitors," statistics that capture the number
of individual users that view at least one of the site's pages over the course of a
month. Newspaper organizations in particular are fond of discussing monthly
audience reach, perhaps because it is the statistic that most resembles estimates of
print circulation.

In fact, monthly audience reach is a much shallower statistic, and certainly not
comparable to audited circulation numbers. The number of sites that a typical user
will visit over the course of any 30day period is huge, and any individual visit
means little. Those who visit a site once, spend less than 30 seconds, and then
immediately click away still counts as visitors. Moreover, many news sites have a
high "bounce rate," in which users visit a single page and then leave. Despite the

1 The only significant exception is the New York Times. Longstanding jokes that the New
York Times is the local paper of San Francisco have a kernel of truth: the greater San
Francisco market does indeed consume the Times at a higher rate than any market except
New York itself. This is the sole example in the data of an outlet generating a tstatistic > 3
in a nonlocal and nonadjacent market.
FCC PUR11000027

DRAFT SUBMITTED 4.6.2011

HINDMAN
LESS OF THE SAME: LOCAL NEWS ON THE INTERNET
7
connotation that these users have been "reached," many or even most unique
visitors make no real connection with a site and spend almost no time there. With
print, we do not count those who glance at newspaper headlines at a newsstand as
readers. We do not count those who flip past CNN looking for something else as
news viewers. But in the online world, equivalent behavior gets rolled into inflated
monthly audience reach estimates.

Even with these definitional issues aside, the comScore data can be expected to
produce lower levels of audience reach than most news sites themselves report
based on internal traffic measures. Overcounting of unique visitors is a widespread
problem in the online news industry.

For those that publish on the Web, there are two primary ways to measure the
number of unique visitors. First, sites may require that users register and log in to
view content, either with payment or (more often) without. When users log in, Web
publishers can set an individuallyidentifying browser cookie that--at least in
theory--can be used to track a user across different computers, different browsers,
and different locations. However, even the relatively minimal barrier of a required
registration can reduce traffic and make users less likely to return to a news site.

The second option for news sites is to track users by placing cookies on their Web
browsers without requiring a login. Most news sites in our sample follow this
course. However, a large portion of the public consumes news on multiple
computers, multiple devices or even just multiple browsers, especially over the
course of a month. When cookies are not tied to a specific registered user, every
computer and every browser counts as a unique reader. Simple clearing of cookies,
or browsing in "private" or "incognito" mode, can create the same problem. While
exact estimates are difficult to come by, one recent industry report estimated that
the uniquevisitortoactualperson ratio was nearly four to one on the average local
news site.2

Questions become even more complicated when sites adopt a mixture of these
models, allowing a portion of the site, some smaller number of articles, or just the
front page to be viewed without registration. Even when sites insist on registration,
those who visit the front page without registering might still be counted as unique
visitors.

A third option for counting unique visitors is to look at the IP addresses of users.
Though less common, this method makes the risk of overcounting even worse. Over
the course of a month, an itinerant user with a laptop can count as dozens of unique
visitors under this standard. This method also allows for undercounting, as multiple
users in a coffee shop or a business may share an IP address. It is left as an exercise
for the reader to consider how many computers or Webenabled devices she has

2 http://www.borrellassociates.com/reports?product_id=832
FCC PUR11000027

DRAFT SUBMITTED 4.6.2011

HINDMAN
LESS OF THE SAME: LOCAL NEWS ON THE INTERNET
8
used in how many different locations in the previous calendar month: by IP address,
each could count as a unique visitor.

comScore, of course, measures audience reach by installing software on users
computers. While there are methodological challenges with recruiting and
maintaining a representative sample, comScore's data should not suffer from the
overcounting of audience reach endemic to other data sources.

The number of unique visitors a site receives thus tells us little about a site's usage
pattern, audience, or market power. Audience reach numbers have the additional
complication that they are not additive: two sites with five percent audience reach
don't add up to 10 percent audience reach, because there is no way to tell how much
their respective audiences overlap. Still, when audience reach is properly measured
this metric can tell us about the number of visitors who ended up at the site least
once. Precisely because audience reach includes even the shallowest interactions
with a site, monthly reach measures let us cast the net as wide as possible in
searching for local news sources.

In measuring Web traffic, page views and time spent on a site tell us much more
about a site's contribution to the overall media landscape. These two metrics are the
main focus of the rest of the paper. However, both need to be understood in a
broader context.

A site with thousands of page views a month may at first look impressively popular.
But page views are plentiful: users viewed 2700 pages a month in our sample on
average, or roughly 90 pages a day. Page views are highly concentrated on the most
popular sites. Facebook alone--the most visited site by page views--accounted for
10 percent of page views (270) in the median market. By the same measure, Google
properties accounted for another 188 page views.

Most page views are short. comScore reports that a page view lasts 26 seconds on
average; 98 percent of page views last less then 2 minutes, and 99.8 percent last less
than 10 minutes. Pages views are most helpful when used comparatively, in
understanding the relative audience that two sites have. The page view numbers we
are most interested are fractional--the portion of the total online audience, or the
portion of news traffic, or the portion of just local news traffic. It should be
remembered that each of these fractions has a very large denominator.

One disadvantage of page views as a metric, however, is that they can be impacted
by the specific architecture of a site. Site layouts that spread the same content in
different member pages can increase--or decrease--the number of page views
recorded. A few news sites are notorious for spreading short articles over multiple
pages in an attempt to maximize page views. Ideally, studies of page view traffic
should be supplemented by metrics of time spent on a site. As we will see below,
page views and time spent on site tell a substantively similar story about the
FCC PUR11000027

DRAFT SUBMITTED 4.6.2011

HINDMAN
LESS OF THE SAME: LOCAL NEWS ON THE INTERNET
9
audience for local news, both overall and for the relative audience that sites in the
same market receive.

Page views and time spent on site are important for another reason as well:
advertising. Most online news revenue comes from selling online ads, and ad sales
by the impression or the click are closely tied to page views. Video advertising on
the Web is somewhat more complicated, but is often sold by the second. The page
views and minutes that accrue to news sites are a good, if imperfect, proxy for the
amount of advertising space that local news sites control.

A final metric useful in understanding local news traffic is the number of sessions
the site receives. Sessions count the number of uninterrupted periods of browsing
on a site. When paired with page views, the number of sessions can measure the
regularity of usage. Consider two sites: Site A, in which users visit once a day and
look at ten pages each; and Site B, in which users visit 10 times a day but see only a
single page each time. Both sites would have the same number of page views, but
Site B would have ten times the recorded number of sessions. News sites have a
relatively high ratio of sessions per page view compared to other types of Web
content, indicating more frequent but less intense usage.

comScore data offers significant advantages over other sources, particularly many
news sites' selfreported traffic numbers. If we want to study a broad crosssection
of Web usage, there are no real alternative to panel data from large Web
measurement firms--particularly comScore, Nielsen/Netratings, and Experian
Hitwise. However, this sort of data also has limitations that should be borne in mind.

A key issue in any panel survey data is how representative participants are of the
general population. comScore reports that they use "an array of online recruitment
techniques to acquire the members of [comScore's] panel." Calibration panels
recruited offline, census data, and monthly phone surveys are used to weight online
recruited panelists in proportion to their prevalence in the general population. In
several validation studies, the weighted comScore traffic estimates have differed by
less than 5 percent on average of from estimates compiled from other independent
sources (Cook &Pettit 2009).

Still, many details of comScore's approach remain proprietary and cannot be
evaluated independently. A key concern for some research is that onlinerecruited
panelists may overrepresent particularly avid Web users: the more pages someone
visits, the more likely she is to see a comScore recruiting ad (Cook &Pettit 2009).
Even if weighting the panel did not fully correct for this, however, audience reach
numbers would be biased upwards rather than downward. The more active users
are, the more likely they are to visit any given site at least once, all else being equal.
Raw numbers of page views (though not share of page views) would also be biased
higher. On this dimension at least, the comScore data likely represent a favorable
portrait of online news audiences.

FCC PUR11000027

DRAFT SUBMITTED 4.6.2011

HINDMAN
LESS OF THE SAME: LOCAL NEWS ON THE INTERNET
10
The comScore panel data examined do not measure usage from mobile devices, a
small but growing portion of traffic to news sites. Even more significant is the worry
that news consumption habits may differ between home and workplace users. Many
employers have policies that preclude installing comScore's software, and
comScore's work panel is much smaller than its home panel. Given these issues, is it
likely that comScore does a better job of measuring home usage than it does in
capturing workplace usage. An excellent discussion of the broader issues
surrounding Web metrics and digital journalism can be found in Graves, Kelly, and
Gluck (2010).


How Much Local News Online? From Which Sources?


The broad landscape of online local news is easy to summarize. Local news is a tiny
part of Web usage; collectively, local news outlets receive less than half of a percent
of all page views in a typical market. Newspapers and television stations dominate
what local news can be found online. Only a handful of local news Websites--17 out
of 1074, all detailed below--are unaffiliated with traditional print or broadcast
media. Across the 100 markets, our methodology finds:
395 television station Websites
590 daily newspaper Websites
41 weekly news publication Websites (nearly all altweekly newspapers)
31 radio station Websites
17 Webnative local news Websites unconnected to print, television, or radio
outlets
Much more detailed information on each of the 100 markets is laid out in

Table 5

, at
the end of the report.

The big picture is that there is little evidence in the comScore data that the Internet
has expanded the number of local news outlets. And while the Internet adds only a
pittance of new sources of local news, the surprisingly small audience for local news
traffic helps explain the financial straits local news organizations now face.

Let us start with the total audience for local news outlets, which is uniformly small.
Detailed summary statistics for our Web traffic variables can be found in

Table 1


(appended below). Online local news sites received only 11.4 monthly page views
per person in the median market. Even with a few highend outliers, the overall
average rises to just 13.8 monthly page views, or roughly 3 pages a week. These
numbers represent just 0.43 percent of the total monthly page views in the median
market (with the overall average slightly higher at 0.51).

Local news sites were between 0.30 and 0.62 percent of all monthly page views in
half of the observations, equivalent to between 8.3 and 17.0 page views per person.
The largest outlier by far is Salt Lake City, where local news--and especially the
television site KSL.com--gets more than 3 percent of all page views. At the other
FCC PUR11000027

DRAFT SUBMITTED 4.6.2011

HINDMAN
LESS OF THE SAME: LOCAL NEWS ON THE INTERNET
11
end of the spectrum are Colorado SpringsPueblo, Las Vegas, and Los Angeles, which
give less than 0.15 percent of their page views to local news sites on average. All
three average less than 4 monthly local news page views per capita.

An extremely similar story can be found in looking at time spent on news sites
rather than page views. In the median market only 9.1 monthly minutes per person
went to local news sites, equal to just 0.45 percent of time online. Half of markets
had local news at between 0.33 and 0.63 percent of online minutes, or between 6.3
and 12.4 minutes per capita. Measured by page views or minutes, local news outlets
get just a tiny portion of citizens' attention.

Are these grim numbers for the local news audience because of low news
consumption overall, or because local news outlets are losing out to national news
sources? The comScore data shows that the answer is "both." Looking at all news
sites--both those in the "News/Information" category and all the additional news
sites identified above--we find that the average market sends 74 monthly page
views to news sites of all stripes. This works out to roughly 60 page views for
nonlocal news sources, and 14 for local ones. The figures for minutes spent are even
smaller. The average market shows users spending 60 percapita minutes per
month on nonlocal news Websites, but just 11 on local ones.

These numbers are consistent with previous findings from other data sources (e.g.
Hindman 2009) that news sites receive just a few percent points of Web traffic. Still,
the small proportion of local news is surprising. Less than one in five news page
views goes to a local news source.


How Many Outlets? Of What Type?

Even if local news is a small part of Web content, knowing where that content comes
from matters. How many such sites are there in a typical broadcast market?

Markets in the sample average about ten and a half online local news sources. On a
typical month that breaks that breaks down into 6.1 newspaper sites, 3.8 local
television sites, .3 radio stations, and less than .2 Internetnative news outlets. The
markets with the largest numbers of outlets are Chicago (19), New York (20),
Minneapolis (20), Cleveland (21), and Boston (with a whopping 28). As these
examples suggest, larger markets produce larger numbers of outlets, even adjusting
for market share. As we shall see, however, larger markets do NOT show greater
consumption of local news as measured by time or page views. The markets with
the fewest outlets in our survey were Baton Rouge (4), Ft. Smith (5), and El Paso (5).

The biggest differences in the number of outlets come from varying numbers of
newspapers. 88 out of the 100 markets have three, four, or five television station
FCC PUR11000027

DRAFT SUBMITTED 4.6.2011

HINDMAN
LESS OF THE SAME: LOCAL NEWS ON THE INTERNET
12
Websites, including stations with just cable distribution.3 Most have no radio or
Internetonly outlets that meet our minimal audience threshold. That leaves
differences in the number of newspapers to account for the rest of the variance.
About half of the markets have either 3 or fewer newspapers or 11+ newspapers. As
we would expect, several of the markets with only two print news sources--
RichmondPetersburg, Baton Rouge, Tucson, El Paso, Colorado SpringsPueblo, Fort
Smith--compete for the fewest number of outlets overall. The two markets with the
highest number of newspapers--HartfordNew Haven (14) and Boston (21)--both
boast a profusion of papers tied to small New England towns. In those two cases
geography and history seem to play a major role in explaining the number of print
news outlets. The large majority of newspapers found are published daily. However,
41 newspapers in our sample our weekly publications--almost all alternative
weeklies published in one of the larger metro areas in our sample.

If there are more print than television online news sources, their collective online
audiences are closer to parity. Put together, newspaper sites average .25 percent of
their markets' monthly page views, versus .20 for television sites. But television
sites do better in terms of minutes spent. By minutes spent, television and
newspaper sites have identical averages: .25 percent of minutes each. This modest
improvement is likely due to online video, which produces longer than average page
view times.


Local News Competition on the Web

As the audience numbers above show, local news is a small niche in the broader
Web. But within that niche, most local news markets are quite concentrated.
Consider the top newspaper and top TV station site in each market. The top paper
earns .15 percent of all page views on average, while the top television station adds
another .16 percent. These averages are slightly skewed upwards by a few outliers,
though the top newspaper site and TV station site together get 56 percent of local
news page views in the median market.

Evidence of concentration shows up in more systematic metrics as well. Perhaps the
most commonly used metric of market concentration is the HerfindahlHirschman
Index (HHI). The HHI is the sum of the squared market share (in percent) of all of
the firms in a given market; it has possible values between 0 and 10,000. Though
developed for a somewhat different application, the U.S. Department of Justice and
Federal Trade Commission's joint antitrust guidelines can provide some context in
interpreting these numbers. According to revised DOJ and FTC rules, markets with
an HHI between 1500 and 2500 are classified as moderately concentrated, while

3 AlbanySchenectady, Chicago, Memphis, SacramentoStocktonModesto, and San
Diego each had six television station Websites; New York City had a total of eight
station Websites.
FCC PUR11000027

DRAFT SUBMITTED 4.6.2011

HINDMAN
LESS OF THE SAME: LOCAL NEWS ON THE INTERNET
13
markets with an HHI greater that 2500 are classified as highly concentrated. The
HHI statistic serves as an initial screen for heightened scrutiny, while the full test
examines other factors--such as entry conditions--that might allow a firm to
produce a "significant and nontransitory increase in price."

In any discussion of market power or antitrust issues, delimiting the relevant
market space is a critical (and sometimes contested) first task. In the analysis that
follows, I calculate the HHI as if local online news were a separate market from local
broadcast and print news
. This is a useful analytic assumption, and indeed the only
assumption possible with the data. Still, as we shall see, there seem to be strong
links between online and offline news markets that may make that assumption
problematic in other contexts. Because HHI attempts to assess firms' market power,
I combine the market shares of multiple outlets in the same market owned by the
same firm. For example, page views on the Atlanta JournalConstitution site, the
WSBTV site, and the WSBradio site are all summed together.

If we do examine local online news markets separately from print and broadcast
markets, we find a surprising level of concentration. Whether we use minutes or
page views to measure market share, the HHI indicates that many online local news
markets are within the envelope of closer regulatory scrutiny. Averaged across our
three months, the median market has an HHI of 2479 with page views and 2593
with minutes spent online. 95 of the 100 markets have an HHI above 1500
measured by page views, and 96 reach that level with online minutes. These
findings should be assessed cautiously, in part because many markets show high
monthtomonth variance in HHI. The median monthly swing is 296 points with
page views and 340 points in minutes spent. But the overall picture is clear: most
online local news markets are dominated by just a few news organizations.


A Census of OnlineOnly Local News Outlets


A central goal of this study is to catalog online news sites that are not affiliated with
traditional media outlets--and that therefore have a strong prima facie claim to be
adding to media diversity. Perhaps the single most surprising finding in this study is
just how few such outlets there are.

Out of the 1074 online local news sources this study identifies, only 17 are
genuinely new media outlets rather than just online outposts of an established print
or broadcast media. The dearth of new Internet outlets allows us to list these sites in
their entirety. In descending order of local audience reach, here are all of the
Internetonly news outlets that show up in our survey:

The onlineonly local news site with the largest audience reach is

SeattlePI.com

.
Once the online home of The Seattle PostIntelligencer newspaper, the site
remained active even after the PI stopped its presses and laid off the vast
FCC PUR11000027

DRAFT SUBMITTED 4.6.2011

HINDMAN
LESS OF THE SAME: LOCAL NEWS ON THE INTERNET
14
majority of its staff in 2007. In the comScore data, the PI Website achieved
reasonably broad reach but only shallow usage. Audience reach was between 7.7
and 12.7 percent, while the share of monthly page views (across all types of
Website) was between 0.026 and 0.046 percent.

Chattanoogan.com

, an online newspaper based in Chattanooga, Tennessee, is
one of the earliest onlineonly local news projects in the country.
Chattanoogan.com was founded in the summer of 1999, in the wake of the sale of
the Chattanooga Times to the larger Chattanooga Free Press; the consolidation
made Chattanooga a onenewspaper town. Chattanooga.com garnered between
6.3 and 8.6 percent monthly reach, and between 0.06 and 0.08 percent of
monthly page views.

TucsonCitizen.com

is the site of the former Tucson Citizen daily newspaper. The
site survived the closure of the paper in May 2009, with the revamped site now
having a heavy focus on political opinion. The site received between 2.6 and 6.3
percent audience reach in Tucson during the period studied, but just 0.003 to
0.007 percent of page views.

KYPost.com

is an online newspaper serving northern Kentucky. Formerly the
Website of the Kentucky Post daily newspaper--a regional variant of the
Cincinnati Post--the Website continued on after the print versions ceased
publication in December 2007. The Post achieved between 2.1 and 3.3 percent
audience reach in Cincinnati, and between 0.007 and 0.005 percent of the
market's page views.

OnMilwaukee.com

is an online publication based in Milwaukee, Wisconsin, that
marries local arts and events coverage with some local news. It received
between 2.5 and 3.1 percent reach, and between 0.002 and 0.003 percent of
page views.

GoWilkes.com

is a local news and information site focusing on Wilkes County,
North Carolina. It has only modest reach, getting between 2.1 and 2.8 percent
audience reach. Surprisingly, however, it has a heavy page view count,
accounting for between 0.26 and 0.60 percent of page views in the Greensboro
High PointWinston Salem market area.

FingerLakes1.com

serves the Finger Lakes region in upstate New York. The
relatively simple site contains both local events coverage and a listing of local
news stories. It earned between 1.0 and 3.1 percent reach and between .11 and
.22 percent of page views in the Rochester market.

LorainCounty.com

is a local news and directory site founded by two brothers;
the site's history as a news source stretches back to the mid1990s. It received
FCC PUR11000027

DRAFT SUBMITTED 4.6.2011

HINDMAN
LESS OF THE SAME: LOCAL NEWS ON THE INTERNET
15
between 1.0 and 2.2 percent audience reach and 0.004 and 0.008 percent of local
page views in the Cleveland market.

PegasusNews.com

is a local site that serves DallasFort Worth. Some local news
on the site comes from staff writers, though a larger portion is taken from
content partners. While previously owned by Fisher Communications, which
operates a number of print and broadcast media (though none in Dallas), the site
was sold in January 2010; its current ownership status and crossmedia
affiliation is unclear. Reach varied between 0.8 and 2.0 percent, with page views
between 0.001 and 0.003.

GWDToday.com

serves Greenwood, SC. The site's design is unpolished, but its
reporting staff averages several local stories a day. The site had a market reach
between .7 and 1.8 percent, and a page view share between .005 and .010
percent in the GreenvilleSpartanburgAsheville market.

SanDiego.com

declares that it has "evolved from a destinationfocused travelers
portal" into an "online community partner for locals and visitors alike." Travel
links and resources remain prominent site features, though it does provide some
local news. The site had an audience reach between 1.1 and 1.7 percent in San
Diego, along with a minimal page view share of between 0.001 and 0.003
percent.

SOMD.com

is a local site focused on southern Maryland. It has local events,
listings, user forums, and new. Most news comes from content partners or law
enforcement releases; it does little of its own reporting. The site's audience
reach in the Washington DC market was between 1.0 and 1.2 percent, with page
views between 0.005 and 0.008 percent of the market total.

iBerkshires.com is a fourperson news and local information site that serves
western Massachusetts. The site appears only in our February data, with a reach
of 1.2 percent and 0.008 percent of page views in the AlbanySchenectadyTroy
market.

SanJose.com

is a "city guide" that focuses on travel, dining, and events, but also
provides some local news. The site had a market reach between 0.8 and 1.3
percent, and a page view share between 0.0007 and 0.0015 percent.

MinnPost.com,

a nonprofit news site, describes its mission as providing "high
quality journalism for newsintense people who care about Minnesota." Though
it has been discussed (often with Voice of San Diego, below) as a potential
business model for local news, its traffic in our data is minimal: between 0.5 and
1.3 percent audience reach in MinneapolisSt. Paul, and from .0009 to .0012
percent of page views.

FCC PUR11000027

DRAFT SUBMITTED 4.6.2011

HINDMAN
LESS OF THE SAME: LOCAL NEWS ON THE INTERNET
16

VoiceofSanDiego.com

bills itself as a "publicservice, nonprofit news
organization that focuses on indepth and investigative reporting." The site is
elegant and contentrich, but traffic numbers are low: reach was .48 percent in
February (with 0.0005 percent of San Diego pages), 1 percent in April (with
0.0008 percent of pages), and too low to measure in March.

In addition to the above sites,

SDNN.com

, the San Diego News Network "community
hub," also counted as an online news site during the period studied. However, the
site ceased business and stopped updating its content in the summer of 2010.

Some patterns in this data are obvious. The Internetonly sites that average more
than 3 percent monthly reach are Websites of newspapers that recently ceased
publication, or in the case of Chattanoogan.com were founded in the aftermath of
a newspaper closure. While these sites may help maintain a bit of news diversity
that would otherwise be lost, their persistence can't be counted as evidence that the
Internet is expanding local news options.

The poor showing of MinnPost.com and Voice of San Diego may be especially
surprising to some. While MinnPost and VoSD are particularly celebrated examples
of a new breed of local and regional online news organizations, numerous other
local online news sites are missing altogether in the above listing--including many
other sites mentioned as promising experiments. If traffic to these "model" outlets is
minimal across the board, this has profound implications for media diversity, and
for the future of journalism.

Is there anything more that the data can tell us about the Webbased local news
sites that have garnered so much journalistic attention? To provide a better answer
to this question, I do a fuller search of the data for Internetonly news outlets that
have featured in the journalism trade press, and in broader discussions of the future
of journalism. Perhaps these outlets are present in the data, but miscategorized, or
slightly below the 1 percent traffic threshold set as a consistent crossmarket bar.

I thus assemble a larger list of Internetonly news organizations, checking if any are
included in the comScore data. This deeper search looks for specific site names
regardless of category or traffic level, so long as the 6visit minimum is met. Sites
based outside one of the top 100 broadcast markets were excluded, as were
journalism experiments focusing on national rather than local news. The listing was
assembled from both online directories and from recent journalism scholarship. In
particular, listings were taken from the Columbia Journalism Review's News
Frontier Database,4 Michelle McLellan's listing at the Reynolds Journalism Institute,5

4 http://www.cjr.org/the_news_frontier_database/
5 http://www.rjionline.org/projects/mcellan/stories/communitynews
sites/index.php
FCC PUR11000027

DRAFT SUBMITTED 4.6.2011

HINDMAN
LESS OF THE SAME: LOCAL NEWS ON THE INTERNET
17
and all sites included in the Nonprofit News Organizations listing at the Harvard
Kennedy School's Hauser Center.6 The final list of news organizations includes:

The Arizona Guardian (Phoenix)
Baristanet (New York)
The Bay Citizen (San FranciscoOaklandSan Jose)
Capital (New York)
California Watch (San FranciscoOaklandSan Jose)
Chicago News Cooperative (Chicago)
The Colorado Independent (Denver)
The Connecticut Mirror (HartfordNew Haven)
Florida Center for Investigative Reporting (Miami)
The Florida Independent (Miami)
The Gotham Gazette (New York)
InDenver Times (Denver)
Investigate West (Seattle)
The Iowa Independent (Des MoinesAmes)
The Lens (New Orleans)
Maine Center for Investigative Reporting / Pine Tree Watchdog (Portland
Auburn)
The Michigan Messenger (Detroit)
The Minnesota Independent (MinneapolisSt. Paul)
New England Center for Investigative Reporting (Boston)
The New Haven Independent (HartfordNew Haven)
New Jersey Newsroom (New York)
Oakland Local (San FranciscoOaklandSan Jose)
Open Media Boston (Boston)
Portland Afoot (Portland, OR)
The Rapidian (Grand RapidsKalamazooBattle Creek)
Rocky Mountain Investigative News Network (Denver)
The Sacramento Press (SacramentoStocktonModesto)
The San Francisco Appeal (San FranciscoOaklandSan Jose)
The SF Public Press (San FranciscoOaklandSan Jose)
The Seattle Post Globe (Seattle)
Spot.us (San FranciscoOaklandSan Jose)
The St. Louis Beacon (St. Louis)
The South Los Angeles Report (Los Angeles)
The Texas Tribune (Austin)
The Tucson Sentinel (Tuscon)
Twin Cities Daily Planet (MinneapolisSt. Paul)
VTDigger.com (BurlingtonPlattsburgh)

6 http://www.hks.harvard.edu/hauser/engage/artsculturemedia/nonprofitnews
organizations/
FCC PUR11000027

DRAFT SUBMITTED 4.6.2011

HINDMAN
LESS OF THE SAME: LOCAL NEWS ON THE INTERNET
18
Wisconsin Center for Investigative Journalism (Madison)

The results of this deeper survey are striking. The Minnesota Independent shows up
in just the April MinneapolisSt. Paul data, with 6 visitors (out of 3201 panelists). In
the same market, the Twin Cities Daily Planet also marked 9 visitors in April, though
none were recorded in February or March. The San Francisco Appeal earned 8
visitors (out of 5540 panelists) in February, 6 visitors in April, and too few to
measure in March. The Gotham Gazette was in the New York sample for March and
April, though not February; it received 12 visitors both months, out of 19,998 NYC
market panelists. All of these numbers are far below our traffic threshold.

None of the other outlets appear even once in the comScore data.

Another site largely absent in our data is

Patch.com

, now part of AOL, the nation's
largest hyperlocal journalism project. Patch.com shows up in just four times in our
data: all three months in the New York market, where it receives between 37 and 50
visitors, and in the San Francisco market, where it receives 9 visitors in April. The
New York media market is where Patch.com started, as a project of Tim Armstrong
(who later joined AOL as CEO), and it has its densest listing of hyperlocal sites there.
Even in New York, however, the numerous Patch sites collectively did not achieve 1
percent audience reach. Recent published reports have suggested that the typical
Patch story gets only 100 or fewer page views (Kopytoff 2011). If these reports are
accurate, they put most Patch sites far below our expected detection threshold.

The broad comScore coverage also allows us to piggyback onto recent indepth
studies of local journalism in the digital age. First, the Institute for Interactive
Journalism authored a recent study of the online news ecosystem in Philadelphia.
They claim to have identified 260 local blogs, including "about 60 [with] some
journalistic DNA in that they report news, not just comment on it" (Shafer 2010).
While JLab does not provide a full listing of the sites, they single out several as
particularly successful examples. Metropolis is an online news outlet staffed by
professional journalists with experience in traditional media. TechnicallyPhilly.com
focuses on the city's tech community. Public School Notebook covers Philly schools
and local education issues. PlanPhilly.com concentrates on planning and zoning.
SeptaWatch.org provides coverage of local transportation. The Broad Street Review
provides coverage of the local arts scene.

The Philadelphia media market provides the fourthlargest panel in the sample,
making it easier to find lowmarketreach sites here than it is almost anywhere else.
PlanPhilly.com shows up just in the February data, with 7 visitors out of 7967
panelists. None of the other online news sources show up at all.

The Project for Excellence in Journalism (2010) also conducted a detailed look at
Baltimore's online news environment in "How News Happens: A Study of the News
Ecosystem of One American City." PEJ found 10 unaffiliated digital news sources
based in Baltimore. Half were hosted on larger sites such as Blogspot or Twitter,
FCC PUR11000027

DRAFT SUBMITTED 4.6.2011

HINDMAN
LESS OF THE SAME: LOCAL NEWS ON THE INTERNET
19
meaning that they were not visible in our data. Sites hosted on independent
domains included BaltimoreBrew.com (founded by former Baltimore Sun staffers),
BMoreNews.com, ExhibitANewsBaltimore.com, InsideCharmCity, and
InvestigativeVoice.com (run by former Baltimore Examiner employees). None of
these sites appeared in the comScore data.

How are we to make sense of these null findings? First, it is worth remembering just
how much traffic one visitor in the comScore panel represents. As a rule of thumb,
one comScore panelist approximates--very roughly--600 real life audience
members. The New York City television market, for example, has an online audience
of slightly more than 11 million people, which comScore tries to track with a New
York panel of 19,998. Assume for the moment that the sample construction is
perfectly random: in that case, a site that averages 3,000 unique, withinmarket
visitors a month will still appear in our data less than half the time. Since our data
have a traffic threshold for each market, local sites with some crossmarket reach
can receive even more traffic without being likely to appear.

Measuring the size of tiny groups with panel methods is a known problem in the
social sciences. In these cases, even small amounts of bias or measurement error can
exceed the size of the group to be estimated (e.g. Gelman 2010). While the comScore
data set may be enormous by the standards of national surveys, it still cannot make
precise audience estimates for the smallest Websites in the smallest local markets.
Still, the fact that such sites are too small to measure is a powerful substantive
finding in its own right. Our data can provide strong bounds on their maximum
audience.


Five Markets Under the Microscope


Is there anything more that we can say about these smallest news and information
outlets? Blogging is a widespread phenomenon, with 11 percent of those over 30
reportedly engaged in the activity to at least a minimal degree (Lenhart et. al 2010).
Any large metropolitan area can expect to have a number of bloggers that deal, at
least in part, with city issues. While the lack of traffic to local news in general is
surprising, a better understanding of local blogs can inform our conclusions about
the Internetonly news and information sources.

With these aims in mind, this study takes a closer look at five media markets, using
search engines and blog lists to find local news content that might be undetected in
the comScore data. This more detailed study does not aim for a comprehensive
census of local blog or hyperlocal news content in these markets: that would require
a larger and specially designed study. Rather, the goal is to serve as a robustness
check on the rest of the analysis, and to search for relatively popular, highquality
sites that might nonetheless have escaped notice. If we understand the "best of
FCC PUR11000027

DRAFT SUBMITTED 4.6.2011

HINDMAN
LESS OF THE SAME: LOCAL NEWS ON THE INTERNET
20
breed" blogs and hyperlocal Websites in these markets, we can understand better
what the broader survey might have missed.

Much previous research has show that blogs--include small subcategories and sub
subcategories of blogs--follow a highly skewed, roughly power law distribution (e.g.
Adamic and Glance 2005, Hindman 2009). Those findings mean several things if
they hold in this context. First, the most popular blogs should be easy to find either
through either directed search or through following links from other blogs. Second,
because links and traffic to blogs are highly skewed, this handful of top sites should
receive a large fraction of all local site links and traffic. As a rough guess, one might
expect that the top 10 local news sites to receive between onefourth and onehalf of
all category traffic. Third, this research suggests delving further into the blog
rankings add little, because the additional sites each garner little traffic.

One goal of this research is to examine the impact of televisionnewspaper cross
ownership on local media diversity. In this vein, the obvious two case studies to
examine are DallasFt. Worth and Houston. In DallasFt. Worth the Dallas Morning
News and the ABCaffiliated television WFAA are both owned by the Belo
Corporation; Houston does not have an instance of crossownership. These two
cities are of similar size, with more than 3 million people in both metro areas. They
are located in the same state, helping to isolate statelevel variables that might affect
media consumption or the number of outlets available.

The other three cities examined are: Portland, Oregon; Cincinnati, Ohio; and
Charlotte, North Carolina. These three smaller cities are from different parts of the
country with divergent cultural reputations. All three are within one standard
deviation of the national norm in terms of broadband penetration, income, elderly
population, and nonHispanic white population.

In all five cases, search engines were used to look for local news and information
sources. Three types of searches were used: first, the city name and the term news
(i.e. "dallas news"); second, the city name and the term blog ("dallas blog"); and
third, the mayor's name along with the city name ("dwaine caraway dallas"). While
the rationale for searching for "dallas news" or "dallas blog" is obvious, political
scientists often measure political knowledge by asking about familiarity with public
officials. If local news is to promote democratic accountability, acquainting readers
with the highestprofile local public official is an essential part of that process.
Where appropriate, related search terms were excluded; "maine" was added to
Portland searches to prevent confusion with Portland, Maine.

The top 100 searches for each query were examined, for a set of 1500 results
overall. Sites returned were categorized as news and information sources using the
same standards used with the comScore data. Compared to the previous coding
above, classification required greater subjectivity and a larger number of unclear
coding decisions. Particularly problematic were blogs that referenced local news
and local issues, but only rarely. If at least 20 percent of the preceding month's
FCC PUR11000027

DRAFT SUBMITTED 4.6.2011

HINDMAN
LESS OF THE SAME: LOCAL NEWS ON THE INTERNET
21
content had at least some local news component, broadly construed, a site was
considered to be a local news source. Near the bottom of these results, many results
were discovered to be abandoned local blogs, or sites that had not been updated
recently. Only blogs or local sites that have been updated within the previous two
weeks were included in the roster. Lastly, sites focusing exclusively on dining, arts
and entertainment, or other similarly narrow topics were excluded, on the grounds
that they did not meet established definitions of a "local news and information
source." For example, a blog about the local music scene would be included only if it
referenced local politics, local public officials, or local issues. When local blogs or
local news sites were identified, references to other local sites were followed to
expand the number of candidate sites.

The first and most obvious finding was that, in each of these five markets, searches
for news returned the same top sites seen in the comScore data. Local news
searches pointed users first to local TV sites and local newspaper sites in every case;
so did searches for the local mayor. This validation was welcome. These searches
also found alternative print publications focusing on smaller communities that did
not show up in the comScore data, such as Websites for black press newspapers, or
sites for local LGBT periodicals.

The methodology turned up dramatically different numbers of blogs and hyperlocal
sites in the different markets. Dallas produced only three examples of Webnative
local content. But the same methods found nine local sites in Houston, six in
Charlotte, 12 in Cincinnati, and 18 in Portland.

This qualitative assessment included both good news and bad news. Some
observers have worried that local blogs would produce only inferior content. If our
concern is the ability of news sites to inform citizens, a high quality news and
information source does several things: it provides accurate, regularly updated
accounts of local news stories and events; it discusses the actions and views of local
officials, stakeholders, and other residents; and it is wellwritten. Though the quality
of local sites found varied enormously, every market had at least one Webnative
news site that was judged to be high quality according to these standards. In Dallas,

DallasSouthNews.com

is a nonprofit news organization that combines traditional
reporting, citizen journalism, and local blogging. In Houston

BlogHouston.net

and

BigJolly.com

produce commentary on local and regional news.

CLT Blog

and

The

Meck Deck

produce opinionated updates about life and politics in Charlotte.
Cincinnati and Portland do even better in our sample, with several high quality local
blogs each.

Simply because some of these sites produce highquality content, however, does not
mean that they are substitutes for a traditional media outlet. Previous studies have
long found that blogs produce little original news reporting (e.g. Blood 2003). With
a couple of exceptions, that finding holds true for the blogs this study identifies. A
majority of posts involve commentary on stories and features found in traditional
media outlets.
FCC PUR11000027

DRAFT SUBMITTED 4.6.2011

HINDMAN
LESS OF THE SAME: LOCAL NEWS ON THE INTERNET
22

Ideally, we would like to measure both the amount of news content that each local
news site produces, and the audience that these sites receive. In fact, it is possible to
examine rough measures of both. These blogs and hyperlocal news outlets typically
publish their entire content to an RSS feed. Google Reader, a Webbased newsreader
tied to Google's Gmail, automatically calculates the average number of postings per
week for any feed requested. It also lists how many of its users are subscribed to a
given feed. Google Reader users are only a fraction of those who use such services,
and they may not necessarily be representative of Web users as a whole. Still,
subscriber numbers can provide a rough measure of the relative audience that even
small blogs receive.

The blogs identified range widely in their number of subscribers. Several have no
subscribers in Google reader. The most successful site,

Bike Portland

, had more
than 5300 subscribers as of March 2011. While most of its content is cycling news
and events, the blog does contain some discussion of public policy, and the blog
itself is a potential vehicle for citizen organizing. Still, Bike Portland is an extreme
outlier; the median number of subscribers to the 18 Portland blogs is 35.

Still, perhaps the biggest problem with content in these local blogs is that there isn't
much of it. Added together, the nine Cincinnati blogs in the Google reader data
produce 5.9 posts a day; the six Charlotte blogs produce 5. Even in Portland, adding
up the 11 sites that have at least 20 subscribers each produces less than 20 postings
a day.

Cincinnati and (especially) Portland are the limiting cases in our study. If blogs add
little in the way of local content in the cities where they are most successful, one
would be hardpressed to argue that they make a substantial contribution to media
diversity elsewhere. Thus, one full week of blog posts for the 11 Portland sites with
20+ subscribers were examined, from Sunday February 27th through Saturday
March 5th, 2011. All together, these blogs produced approximately 27,400 words
worth of content. That is slightly less than 4,000 words a day, small enough to be
printed on a single page of a fullsize daily newspaper.


Regression Analysis


Local news may be a tiny subset of the content that citizens consume, but the
comScore data does show that markets differ in both the number of online local
news outlets available and the traffic that local news receives. Are there systematic
factors that predict the number of local outlets of all kinds found in a given market?
In which sorts of markets do we find Webnative news outlets? What marketlevel
variables are associated with greater or lesser local news consumption? Which
factors promote concentration in local Web news markets?

FCC PUR11000027

DRAFT SUBMITTED 4.6.2011

HINDMAN
LESS OF THE SAME: LOCAL NEWS ON THE INTERNET
23
In considering these questions, it is worth emphasizing what regression analysis
cannot tell us. The initial research design hoped to use regression analysis as a tool
for understanding which markets would produce Webnative news outlets capable
of filling gaps in offline coverage. But the answer, as we have seen, is "none of
them." The number of Webnative outlets is too few, and the traffic they receive too
sparse.

Still, some types of markets do provide more online news outlets--and more Web
native news outlets. Some types of markets show higher levels of local news
consumption. And some cities do show more (or less) concentration in the online
news marketplace. To shed light on these patterns, I perform regression analysis on
our 100 local markets, combining the comScore data with additional data provided
by the FCC.

There is little previous scholarly research to turn for guidance about the market
level factors that predict the number of online local news outlets, the level of online
news consumption, or the degree of online news market concentration. However,
we do know the personal demographics associated with online news consumption,
and established veins of scholarship have examined local news in traditional media
both at the market and individual level (Hindman 2009, Hamilton 2004, Napoli
2003). To the extent possible, this study examines both structural market
characteristics and aggregate market demographics.

Detailed summary statistics for our explanatory variables can be found in

Table 2

(appended at the end of the document). One obvious question concerns the link
between the size of a market, the number of local news outlets we find there, and
the quantity of local news consumed. The variable TV Market Population is based on
estimates from the Census Bureau's American Community Survey. Previous work
has found that larger markets produce more broadcast news (Shiman 2007,
Crawford 2007).

Some research has shown that broadband users use the Web substantially
differently from dialup users, and that they consume more and richer types of Web
based content (Smith 2010). To investigate potential impacts of highspeed access,
Broadband % includes the percentage of the market that subscribes to 768 bps or
faster broadband.

Given that local newspapers and commercial TV stations provide most local news
online, it is particularly important to examine the structure of offline local media.
Regarding newspapers, the models test several variables. Newspaper Copies Per
Capita
reports the total newspaper circulation per person in a given market. # of
Daily Newspapers (5%)
measures the number of daily newspapers that reach at least
5 percent of the market population. # of Newspaper Parent Companies captures the
number of parent daily newspaper firms in the television market. These variables
hopefully provide some leverage over which aspect of the newspaper environment
FCC PUR11000027

DRAFT SUBMITTED 4.6.2011

HINDMAN
LESS OF THE SAME: LOCAL NEWS ON THE INTERNET
24
matters most: the number of newspaper firms, the size of the newspaper audience,
or just the number of outlets that reach a minimum threshold of readership.

Television broadcasters are approached similarly. Since we find few local news
outlets associated with noncommercial television broadcasters, we focus on the
number and type of commercial broadcasters. # Commercial TV Stations records the
number of fullpower commercial television broadcasters. # Commercial TV Parent
Co.
calculates the number of companies that own at least one station in the local
market. The expectation is that a greater number of print and television offline
outlets, from a greater number of companies, should be associated with more online
news sources and greater news consumption.

Additionally, we are interested in whether ownership patterns impact the number
of online outlets found and the amount of local news consumed. LocallyOwned TV
Stations
measures the number of commercial stations with owners in the market.
MinorityOwned Stations captures the number of local television stations under
minority ownership. NewspaperTV CrossOwnership reports the number of parent
companies in a market that own both a daily newspaper and a full power
commercial television station.

While newspaperTV crossownership is of particular concern, the local radio
market is also potentially important. RadioTV CrossOwnership measures the
number of parent entities that own both a broadcast TV station and a radio station.
# of NewsFormat Radio Stations captures the number of broadcast radio stations
with a newsbased format.

We also investigate possible ties between the racial and ethnic makeup of a market
and its online news production. Percent Black and Percent Hispanic capture the
portion of the market population that is AfricanAmerican and Hispanic
respectively. One might hypothesize that markets with larger Hispanic populations
will consume less Englishlanguage local news. Also included are interaction effects
between racial and ethnic makeup and market size: Hispanic % * Market Population
and Black % * Market Population. These terms test whether racial and ethnic market
characteristics have divergent effects in larger vs. smaller markets.

The analysis also controls for possible effects of income and age. Per Capita Income
uses earnings data provided by BIA. Age 65+ is the fraction of the population in the
market that is 65 or older.

Lastly, because there may be monthspecific factors that impact the amount of news
consumed, two dummy variables for the months of February and March are included.
April is the omitted category. These coefficients should be interpreted as the
difference between consumption in April and the listed month.


FCC PUR11000027

DRAFT SUBMITTED 4.6.2011

HINDMAN
LESS OF THE SAME: LOCAL NEWS ON THE INTERNET
25
Regression Results

The data is analyzed using two different sorts of regression models. First, negative
binomial models are used to estimate both the number of Webnative news outlets
found in a given market, and the number of local online news outlets of all kinds.
Second, OLS regression models are used analyze local news consumption and the
level of market concentration.

In all cases, the analyses use robust standard errors clustered by market. An
observation in February is not independent of the repeated observation of the same
market in March, and this reduces our effective sample size. Under these
circumstances regression coefficients are unbiased, but uncorrected standard errors
are misleading and usually anticonservative. Because our effective sample size is
small, this analysis should expect to find only the largest and most consistent
relationships.

Table 3

presents the results of the negative binomial models. These models, which
assume that the dependent variable is a nonnegative integer, are often a better
choice for count data than OLS regression. Negative binomial models are closely
related to poisson regression models. Whereas poisson models have only a single
parameter , which governs both the mean and the variance of the distribution,
negative binomial models add the parameter to capture overdispersion. While
overdispersed count data are common in the social sciences, in this case both
models produce estimates of very close to zero (see Table 3). When = 0, poisson
and negative binomial models are equivalent.

Let us start with the number of Webnative news outlets found in a given market.
Greater numbers of Internetonly news sites are found in markets with lower per
person print circulation. The relationship is highly statistically significant, and it
persists even when we exclude markets where print newspapers recently stopped
their presses while maintaining their Web sits. Whether or not they can effectively
close gaps in coverage, this suggests that Internetonly news sources have indeed
sprung up in areas where print newspaper readership is lower.

The models also show that, all else being equal, markets that are both large and
heavily Hispanic have fewer Internetonly news sites. Though less precisely
measured, the coefficients suggest a similar interaction between the size of a market
and its proportion of AfricanAmerican residents.

There is modest (though not statistically significant) evidence that larger markets,
and markets with higher per capita income, may also produce more Webnative
news outlets. No other variable comes close to statistical significance.

The story looks somewhat different when examining the total number of online
news outlets in a market--a category where daily newspapers sites and television
FCC PUR11000027

DRAFT SUBMITTED 4.6.2011

HINDMAN
LESS OF THE SAME: LOCAL NEWS ON THE INTERNET
26
station sites dominate. First, there is strong evidence that larger markets have more
news outlets. With a zscore > 5, this finding is extremely unlikely to have been
produced by chance.

There is also a strong association between the total number of online outlets, and
the number of newspaper parent companies operating in the market. This comports
with our analysis above, which shows that the biggest variance in outlets comes
from the number of newspaper sites. Larger numbers of commercial television
stations are also associated with more local news sites, though this coefficient is just
shy of statistical significance.

As with the analysis of Internetonly news outlets, there is a complicated
relationship between the racial and ethnic makeup of a market and the number of
online news outlets we find there. The coefficients for Percent Hispanic and Percent
Black
are both positive (only the latter is significant). However, there are also strong
and highly significant interaction effects between a market's racial and ethnic
composition and its size. The model finds fewer online outlets in cities that are both
large and heavily minority.

There is suggestive evidence that markets with a greater percentage of 65andolder
residents have fewer local online outlets, though the result is not quite significant (p
< .13 in a twotailed test). None of the other explanatory variables approach
statistical significance.

This data also allow us to examine the consumption of local news, measured both by
page views and minutes. Here we return to standard OLS linear regression models
(

Table 4

).

A particularly consistent predictor of local news consumption, in both page views
and minutes, is the portion of the population that is Hispanic. Communities with a
proportionally larger Latino population consume less local news than otherwise
comparable cities. Moreover, interaction effects between market size and ethnic
composition amplify this finding. Not only do heavily Hispanic markets have lower
local news consumption than average, and local news traffic in bigger heavily
Hispanic markets is lower still. (The portion of market residents who are African
American does not produce similar findings.)

The model also suggests that media ownership patterns predict the level of local
news consumption. The presence of a minorityowned television station is
associated with greater local news usage in both page views and minutes (though
only minutes is significant). Given the findings above, it should be noted that many
of these minorityowned stations are in large, heavily Hispanic markets (such as
MiamiFt. Lauderdale or Los Angeles). Similarly, the presence of locallyowned TV
stations also predicts higher levels of news consumption, though here too only
minutes spent is statistically significant.
FCC PUR11000027

DRAFT SUBMITTED 4.6.2011

HINDMAN
LESS OF THE SAME: LOCAL NEWS ON THE INTERNET
27

The level of TVnewspaper crossownership also seems to matter. Markets with
crossowned newspapertelevision firms show an extra four monthly page views
per person going to local news sites ( p < .10, twotailed). Findings for local news
minutes are similar, though not significant.

Curiously, markets with greater per capita income are estimated to consume less
news than poorer markets. This finding emerges with both minutes and page views,
and it is statistically significant for both measures. The percentage of local residents
age 65 or older is also associated with lower news consumption, though the results
are significant only with minutes spent.

Lastly, we look at predictors of market concentration as measured by the HHI.
Concentration in minutes spent online and in page views is examined, and both
metrics tell a very similar story. The patterns in market concentration are less
predictable than the preceding models. Most of our predictors do not approach
statistical significance, with two exceptions. First, all else being equal, more
populous markets have lower levels of estimated concentration. This result is
significant with page views, and approaches significance in minutes spent online.

Second, markets with newspapertelevision crossownership show dramatically
higher levels of concentration in both minutes spent and page views. There are 19
such markets in our data, and the estimated effect size is enormous: with TV
newspaper crossownership the model predicts an 1115point jump in the HHI in
page views, and a 1201point jump in the HHI by minutes spent. Both metrics are
statistically significant.


Conclusion


Has the Internet significantly expanded the number of local news voices? The
answer that emerges from the comScore data is firm but qualified "no." We can say
the least about the very smallest online news sources--those that receive less than a
few thousand unique visitors monthly, and are thus unlikely to appear in our data.
But above this threshold, we find almost no evidence that the Internet has expanded
the number of local news outlets.

Most television markets have fewer than a dozen local news Websites. Those sites
that do receive an audience are overwhelmingly newspaper and local television
station Websites, rather than new and independent sources of local news. Only 16 of
our top 100 markets have an unaffiliated Internet news source that reaches our one
percent audience threshold. Even the exceptions prove the rule: the four most
successful Internetonly news sites were all related to the closure of a traditional
print newspaper. The fact that sites like SeattlePI.com continue with a skeleton crew
is welcome, but it does not represent an expansion of media diversity. Online local
FCC PUR11000027

DRAFT SUBMITTED 4.6.2011

HINDMAN
LESS OF THE SAME: LOCAL NEWS ON THE INTERNET
28
news markets resemble downsized versions of traditional media news markets,
with the same news stories produced by the same newspapers and television
stations.

Even more surprising than the small number of outlets, or the lack of new Web
native news organizations, is just how small the online news market is. Discussions
about the newspaper crisis often start with the claim that online news has a revenue
problem, not a readership problem. John Morton's (2010) recent assessment is
typical, arguing that the problem with newspaper sites is that "Lots of people came,
but lots of advertising didn't."

The comScore data show that this diagnosis is wrong. The central problem facing
local online news sites is that their audiences are small--and proportionally much
smaller than even many publishers and journalists seem to realize. Metrics such as
monthly audience reach are often falsely inflated, and deceptive even when
measured accurately. If a particular news startup gets a few tens of thousands of
page views a month, the site is hailed as a success--even though many citizens view
thousands of pages a month each, and even though page views last less than 30
seconds each on average. Online local news has a revenue problem largely because it
has a readership problem.

A detailed economic analysis of the state of local news is outside the scope of this
study. Nonetheless, the fact that local news sites capture little of citizens' attention
has obvious economic implications. If we want to understand the financial viability
of advertisingsupported local news on the Web, we should focus on two questions.
First, how valuable is the entire online advertising space in a given media market?
Second, how much of that online space do local news sites control? In 2009, the last
year for which full data are available, online advertising revenue in the U.S. totaled
$22.7 billion dollars (IAB 2010). That amounts to $74 per person. In the long run,
how much of that $74 is going to accrue to a group of sites that gets onehalf of one
percent of page views in a typical market?

For more than a decade, some have suggested that the Internet and other
technologies (such as cable television) have made it less necessary to regulate
broadcast media. According to this reasoning, the Internet has increased the
number of local news and information outlets available to citizens, strengthened
news competition, and the broadened diversity of news voices.

Arguments that the Internet has expanded the number of local news voices, or
allowed new Webbased news outlets to fill gaps in news coverage, find little
support in this data. In deciding Prometheus v. FCC (2004), the court's majority
worried that local news sources might just be repackaged versions of television and
newspaper content. The comScore data show that this is indeed the case.

Some have found evidence of consumer substitution between online and traditional
news sources (Waldfogel 2003). For national news, and particularly for commodity
FCC PUR11000027

DRAFT SUBMITTED 4.6.2011

HINDMAN
LESS OF THE SAME: LOCAL NEWS ON THE INTERNET
29
news content, this finding likely holds. But the comScore data make it difficult to
sustain the same argument with regard to local news content. We find few examples
of Webnative news sites that are straightforward substitutes for the product of a
television station or a newspaper. Even in markets with relatively successful blog
communities, the top blogs produce only a trickle of content. The lack of traffic these
sites receive is a strong clue that citizens themselves do not think that they are
comparable to television and newspaper Websites.

The low levels of traffic that local news sites receive should color regulators'
assessments in other ways as well. The small audience for local news online makes
it implausible that a midsized or smaller media market can support numerous
onlineonly news organizations with adequate staff and resources. The story of
hyper local journalism thus far has been paved with economic failure, as the long list
of such failed experiments shows.

Lastly, there is evidence that media concentration offline carries over into online
media markets. Most local news markets on the Web are dominated by just a few
firms. If online local news were to be considered as a separate market, half the 100
largest markets would qualify as highly concentrated under Department of Justice
and Federal Trade Commission HHI guidelines, and nearly all would be considered
at least moderately concentrated.

Perhaps the most striking example of offline media structure intersecting with local
news on the Web is seen with newspapertelevision crossownership. In cities
where a firm owns both a newspaper and a television station, we find a jump in the
HerfindahlHirschman Index greater than 1000 points. While the underlying causal
relationship deserves more study, these numbers make a strong argument for
regulatory caution. Restrictions on media crossownership do not just matter in
print and on the airwaves: they likely impact news diversity on the Web as well.


FCC PUR11000027

DRAFT SUBMITTED 4.6.2011

HINDMAN
LESS OF THE SAME: LOCAL NEWS ON THE INTERNET
30

References




Adamic, Lada, and Natalie Glance . 2005 . "The Political Blogosphere and the 2004
U.S. Election: Divided They Blog." In LinkKDD '05: Proceedings of the 3rd
international Workshop on Link Discovery. Online at:
http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.59.9009

Blood, Rebecca. 2003 . "Weblogs and Journalism: Do They Connect"? Nieman
Reports. Fall Online at: http://uu.rowacreative.com/wp
content/Weblogandjournalism.pdf

Cook, William A. & Raymond C. Pettit. 2009 . "comScore Media Metrix U.S.
Methodology." Advertising Research Foundation . Online at: http://thearforg
auxassets.s3.amazonaws.com/downloads/research/comScoreRReview.pdf

Crawford, Gregory S. 2007. "Television Station Ownership Structure and the
Quantity and Quality of TV Programming." FCC Report. Online at:
http://hraunfoss.fcc.gov/edocs_public/attachmatch/DA073470A4.pdf

Gelman, Andrew. 2010 . "When Small Numbers Lead to Big Errors." Scientific
American, October . Online at:
http://www.scientificamerican.com/article.cfm?id=whensmallnumberslead
tobigerrors

Graves, Lucas, John Kelly, and Marissa Gluck. 2010. "Confusion Online: Faulty
Metrics and the Future of Digital Journalism." Tow Center for Digital Journalism,
Columbia University Graduate School of Journalism. Online at:
http://www.journalism.columbia.edu/system/documents/345/original/online_
metrics_report.pdf

Hamilton, James. 2004 . All the News That's Fit to Sell . Princeton, NJ: Princeton
University Press.

Hindman, Matthew. 2009. The Myth of Digital Democracy. Princeton, NJ: Princeton
University Press.

Interactive Advertising Bureau. 2010. IAB Internet Advertising Revenue Report. Full
year. April 2010. Online at:
http://www.iab.net/insights_research/947883/adrevenuereport

Kirchhoff, Suzanne. 2010 . "The U.S. Newspaper Industry in Transition."
Congressional Research Service. Online at:
http://digitalcommons.ilr.cornell.edu/key_workplace/634/

FCC PUR11000027

DRAFT SUBMITTED 4.6.2011

HINDMAN
LESS OF THE SAME: LOCAL NEWS ON THE INTERNET
31
Kopytoff, Verne G. 2011. "AOL Bets on Hyperlocal News, Finding Progress Where
Many Have Failed." The New York Times. January 16. Online at:
http://www.nytimes.com/2011/01/17/business/media/17local.html


Lenhart, Amanda, Kristen Purcell, Aaron Smith, and Kathryn Zickuhr. 2010. Social
Media and Young Adults. Pew Internet and American Life Project . February 3.
Online at: http://www.pewinternet.org/Reports/2010/SocialMediaand
YoungAdults.aspx

Morton, Andrew . 2011 . "Costly Mistakes." American Journalism Review.
December/January . Online at: http://www.ajr.org/Article.asp?id=4994

Napoli, Philip. 2003. Audience Economics: Media Institutions and the Audience
Marketplace. New York: Columbia University Press.

Newspaper Association of America. 2010. Trends and Numbers, Advertising
Expenditures . Online at:
http://www.naa.org/Trendsandnumbers/AdvertisingExpenditures.aspx.

Noam, Eli M. 2009. Media Ownership and Concentration in America . Oxford: Oxford
University Press.

Purcell, Kristen, Lee Rainie, Amy Mitchell, Tom Rosenstiel, and Kenny Olmstead.
2010. "Understanding the Participatory News Consumer." Pew Internet and
American Life Project . March. Online at:
http://www.pewinternet.org/Reports/2010/OnlineNews.aspx

Prior, Markus. 2007. Post Broadcast Democracy. Cambridge: Cambridge University
Press.

Project for Excellence in Journalism . 2010a . "How News Happens: A Study of the
News Ecosystem of One American City." Pew Research Center . Online at:
http://www.journalism.org/analysis_report/how_news_happens

Project for Excellence in Journalism. 2010a . State of the News Media 2009. Online
at: http://www.stateofthemedia.org/2009/

Project for Excellence in Journalism. 2011. State of the News Media 2011. Online at:
http://www.stateofthemedia.org/2011/

Prometheus Radio Project v. FCC. 2004. 373 F. 3d 372.

Shafer, Jan. 2010. "Exploring a Networked Journalism Collaborative in Philadelphia:
An Analysis of the City's Media Ecosystem with Final Recommendations." JLab
FCC PUR11000027

DRAFT SUBMITTED 4.6.2011

HINDMAN
LESS OF THE SAME: LOCAL NEWS ON THE INTERNET
32
report. April. Online at: http://www.j
lab.org/publications/philadelphia_media_project/

Shiman, Daniel . 2007 . "The Impact of Ownership Structure on Television Stations'
News and Public Affairs Programming." FCC Report. Online at:
http://hraunfoss.fcc.gov/edocs_public/attachmatch/DA073470A5.pdf

Smith, Aaron. 2010 . "Home Broadband 2010." Pew Internet and American Life
Project. Online at: http://pewinternet.org/Reports/2010/HomeBroadband
2010.aspx

Smith, Erica. 2009. "Layoffs and Buyouts at U.S. Newspapers in 2009." Online at:
http://graphicdesignr.net/papercuts

Waldfogel, Joel. 2002. "Consumer Substitution Among Media." Media Ownership
Working Group, Federal Communication Commission . Online at:
http://www.fcc.gov/ownership/materials/already
released/consumer090002.pdf


FCC PUR11000027

DRAFT SUBMITTED 4.6.2011

HINDMAN
LESS OF THE SAME: LOCAL NEWS ON THE INTERNET
33

Table 1: Summary Statistics for Variables of Interest



Mean

Std. Dev.

Min

Max

# of WebNative Local News Outlets
.19
.44
0
3
Total # of Local Online News Outlets
10.5
4.2
4
28
Total Number of Local News Page Views / Capita
13.8
10.0
1.8
90.2
Total Number of Local News Minutes / Capita
10.6
7.6
1.3
63.4
Local News Page Views As % of All Page Views
.51
.27
0.6
3.4
Local News Minutes As % of All Online Minutes
.54
.39
0.6
3.2
Total Nonlocal News Page Views / Capita
60.0
30.8
28.0
370
Total Nonlocal News Minutes / Capita
59.0
16.4
23.4
126
HHI in Page Views
2749
1297
921
9003
HHI in Minutes Spent
2943
1444
939
8955



FCC PUR11000027

DRAFT SUBMITTED 4.6.2011

HINDMAN
LESS OF THE SAME: LOCAL NEWS ON THE INTERNET
34

Table 2: Summary Statistics for Explanatory Variables



Mean

Std. Dev.

Min

Max

TV Market Population (in thousands)
2543
2972
142
20841
Broadband
.56
.11
.32
.98
Newspaper Copies Per Capita
1.00
.25
.46
1.7
# of Daily Newspapers (5%+)
1.86
.92
1
5
# of Newspaper Parent Companies
6.4
3.3
1
19
# Commercial TV Stations
8.2
3.4
3
23
# LocallyOwned TV Stations
1.0
1.4
0
8
# MinorityOwned Stations
.38
.81
0
6
NewspaperTV CrossOwnership
.19
.39
0
1
RadioTV CrossOwnership
1.6
1.5
0
7
# of NewsFormat Radio Stations
8.4
5.6
1
30
Percent Hispanic
.12
.15
.01
.91
Hispanic % * Market Population
405
953
10.7
7712
Percent Black
.13
.105
.01
.49
Black % * Market Population
329
486
2.8
3563
Per Capita Income (thousands)
25.0
3.9
13.0
38.9
Percent 65 or older
.14
.02
.09
.24
FCC PUR11000027

DRAFT SUBMITTED 4.6.2011

HINDMAN
LESS OF THE SAME: LOCAL NEWS ON THE INTERNET
35

Table 3: Negative Binomial (Count) Models



# of WebNative News
Total # of Local Online
Outlets
News Outlets
TV Market Population
.0007
.00026
(.0005)
(.00005)
Broadband %
1.80
.045
(2.84)
(.277)
Newspaper Copies Per Capita
2.92
.135
(1.36)
(.129)
# of Daily Newspapers (5%+)
.142
.030
(.321)
(.035)
# of Newspaper Parent Companies
.0049
.031
(.0887)
(.009)
# Commercial TV Stations
.088
.005
(.086)
(.012)
# LocallyOwned TV Stations
.094
.044
(.258)
(.029)
# MinorityOwned Stations
.063
.012
(.452)
(.040)
NewspaperTV CrossOwnership
.209
.045
(.529)
(.091)
RadioTV CrossOwnership
.186
.003
(.218)
(.029)
# of NewsFormat Radio Stations
.064
.0047
(.066)
(.0055)
Percent Hispanic
1.56
.462
(2.55)
(.291)
Hispanic % * Market Population
.0016
.00043
(.0009)
(.00008)
Percent Black
.656
1.35
(3.76)
(.52)
Black % * Market Population
.0014
.0008
(.0015)
(.0002)
Per Capita Income
.123
.002
(.108)
(.010)
Percent 65 or older
2.92
1.46
(11.2)
(.95)
February
.223
.012
(.101)
(.008)
March
.223
.014
(.118)
(.006)
Constant
3.07
1.77
(2.05)
(.25)
/ln(Alpha)
14.5
18.6
(.63)
(.38)
Alpha
.0000
.0000
(.0000)
(.0000)
N
294
294


FCC PUR11000027

DRAFT SUBMITTED 4.6.2011

HINDMAN
LESS OF THE SAME: LOCAL NEWS ON THE INTERNET
36

Table 4: OLS Regression Models



Page Views (per Minutes
HHI in Page
HHI in
capita)
(per capita)
Views
Minutes
TV Market Population
.0002
.0003
.525
.403
(.0016)
(.0014)
(.279)
(.318)
Broadband %
16.6
17.3
896
2174
(13.0)
(11.2)
(1453)
(1573)
Newspaper Copies Per Capita
4.63
1.92
581
975
(3.31)
(2.83)
(550)
(652)
# of Daily Newspapers (5%+)
1.21
.590
47.7
98.5
(.88)
(.744)
(114)
(156)
# of Newspaper Parent Companies
.134
.070
9.20
43.0
(.421)
(.317)
(55.7)
(60.4)
# Commercial TV Stations
.091
.217
.53
18.9
(.386)
(.317)
(57.8)
(64.8)
# LocallyOwned TV Stations
3.11
2.13
52.1
37.9
(1.95)
(1.27)
(170)
(173)
# MinorityOwned Stations
1.60
2.20
125
254
(1.23)
(1.18)
(234)
(264)
NewspaperTV CrossOwnership
3.96
1.68
1115
1201
(2.41)
(1.62)
(514)
(559)
RadioTV CrossOwnership
.521
.095
125
94.2
(.921)
(.656)
(156)
(164)
# of NewsFormat Radio Stations
.516
.247
57.3
25.4
(.399)
(.269)
(35.9)
(39.1)
Percent Hispanic
10.5
7.15
586
27.8
(5.5)
(4.14)
(953)
(1049)
Hispanic % * Market Population
.0058
.0061
.345
.045
(.0024)
(.0021)
(.327)
(.373)
Percent Black
2.26
1.09
507
1589
(10.3)
(7.38)
(2705)
(2543)
Black % * Market Population
.0046
.0028
1.02
.82
(.0040)
(.0041)
(1.03)
(1.16)
Per Capita Income
.670
.480
10.8
28.4
(.334)
(.254)
(44.9)
(46.6)
Percent 65 or older
109
75.9
9670
8643
(71)
(45.2)
(7207)
(7239)
February
1.52
1.80
81.0
77.9
(.54)
(.53)
(85.7)
(101)
March
.088
.608
67.0
78.1
(.35)
(.35)
(68.6)
(93.1)
Constant
26.2
16.0
4435
4456
(10.8)
(7.2)
(1233)
(1292)
R2
.374
.323
.269
.225
Root MSE
8.26
6.47
1155
1323

FCC PUR11000027

DRAFT SUBMITTED 4.6.2011

HINDMAN
LESS OF THE SAME: LOCAL NEWS ON THE INTERNET
37

Table 5: MarketLevel Data On Local News Websites


This table presents marketlevel statistics for each of the 100 markets in our
sample. Included are the following:

Loc. Page Views:

the number of page views received by all online local news
sites, divided by comScore's estimated online population. This is the average
of the February, March, and April data.

Page Views (%):

Local news page views as a percent of all page views in the
market. (FebruaryMarchApril average.)

Loc. Min.:

number of minutes spent on local news Web sites, divided by the
estimated online population. This is the average of the February, March, and
April data. (FebruaryMarchApril average.)

Loc. Min. (%):

minutes spent on online local news sites as a percentage of all
online usage in the market. (FebruaryMarchApril average.)

Print:

number of local print news Websites found in the April 2010 data. Both
daily and weekly print publications are included. This statistic (and TV, Radio,
and Web below) are subject to censoring, as sites that receive less than 6 raw
visitors do not appear in the April data, even if they have appeared previously.

TV:

number of local television station news sites in the April 2010 data.

Radio:

number of local radio station news sites in the April 2010 data.

Web:

number of Webnative news sites in the April 2010 data.

HHI (page):

HerfindahlHirshman index for all local online news sources,
calculated by the percentage of page views that each site receives. The
reported statistic is the average of the February, March, and April 2010
numbers.


Market

Loc.

Page

Loc.

Loc.

Print TV

Radio Web HHI

Page

Views

Min.

Min.

(page)

Views

(%)
(%)
ALBANY-SCHENECTADY-TROY
19.2
.68
11.0
.54
9
5
0
0
2363
ALBUQUERQUE-SANTA FE
5.5
.22
3.4
.18
5
3
0
0
2531
ATLANTA
9.6
.36
8.4
.44
6
4
2
0
4282
AUSTIN
7.3
.25
6.1
.27
3
5
0
0
2275
BALTIMORE
13.0
.47
11.5
.56
5
4
1
0
2430
BATON ROUGE
26.1
1.0
16.0
.90
2
3
0
0
8320
BIRMINGHAM
12.4
.48
8.1
.44
9
3
0
0
4840
BOSTON
15.4
.50
12.6
.57
21
5
2
0
1606
BUFFALO
13.4
.50
9.3
.50
8
3
1
0
2792
BURLINGTON-PLATTSBURGH
13,4
.51
8.1
.42
6
3
0
0
1588
CEDAR RAPIDS-WATERLOO&DUBQ
26.4
.87
18.2
.89
4
3
0
0
3660
CHAMPAIGN&SPRNGFLD-DECATUR
20.5
.75
13.1
.64
5
3
0
0
2892
CHARLESTON, SC
6.1
.25
4.5
.25
4
3
0
0
2940
CHARLESTON-HUNTINGTON
14,6
.50
10.8
.51
5
1
0
0
2730
CHARLOTTE
14.1
.52
12.3
.67
9
5
0
0
1805
CHATTANOOGA
12.6
.46
7.4
.36
3
3
0
1
2513
CHICAGO
9.9
.36
7.9
.39
11
6
1
0
1816
FCC PUR11000027

DRAFT SUBMITTED 4.6.2011

HINDMAN
LESS OF THE SAME: LOCAL NEWS ON THE INTERNET
38
CINCINNATI
15.7
.57
17.6
.90
5
3
1
1
2786
CLEVELAND
19.6
.65
15.4
.73
13
5
2
1
1791
COLORADO SPRINGS-PUEBLO
3.0
.12
2.1
.12
2
5
0
0
2130
COLUMBIA, SC
13.6
.54
12.3
.66
3
3
0
0
2081
COLUMBUS, OH
24.5
.85
11.6
.54
10
5
1
0
3620
DALLAS-FT. WORTH
5.5
.22
5.5
.30
4
4
0
1
3256
DAVENPORT-R.ISLAND-MOLINE
16.2
.59
7.9
.42
7
2
0
0
2446
DAYTON
10.2
.34
11.2
.53
4
3
1
0
4500
DENVER
11.1
.38
8.5
.38
9
4
0
0
2723
DES MOINES-AMES
15.6
.48
9.4
.48
4
2
0
0
3487
DETROIT
12.9
.48
12.7
.67
9
4
0
0
1546
EL PASO
7.2
.29
5.7
.35
2
3
0
0
3070
FLINT-SAGINAW-BAY CITY
12.3
.48
10.2
.53
5
3
0
0
4973
FRESNO-VISALIA
6.9
.25
4.2
.22
5
5
0
0
2718
FT. MYERS-NAPLES
8.3
.32
6.9
.36
4
2
0
0
2436
FT. SMITH
7.1
.34
5.2
.35
2
3
0
0
3119
GRAND RAPIDS-KALMZOO-B.CRK
15.8
.57
13.7
.71
2
4
0
0
3298
GREEN BAY-APPLETON
16.7
.84
11.1
.59
5
4
0
0
1852
GREENSBORO-H.POINT-W.SALEM
24.3
.88
15.7
.84
9
2
0
1
2682
GREENVILLE-SPART-ASHEVILLE
18.8
.71
13.6
.72
7
4
0
1
2328
HARLINGEN-WESLACO-BRNSVLLE
3.6
.16
3.5
.23
3
4
0
0
2233
HARRISBURG-LNCSTR-LEB-YORK
27.9
.95
22.3
1.0
9
5
0
0
2239
HARTFORD & NEW HAVEN
13.0
.46
10.0
.47
14
3
0
0
1219
HONOLULU
7.2
.25
6.9
.36
6
3
0
0
3250
HOUSTON
7.8
.30
10.3
.56
5
4
0
0
2435
HUNTSVILLE-DECATUR,FLOR
13.0
.59
7.5
.46
2
3
0
0
2297
INDIANAPOLIS
22.2
.81
15.9
.78
12
5
0
0
1764
JACKSON, MS
11.1
.51
7.6
.48
6
3
0
0
1965
JACKSONVILLE, BRUNSWICK
10.8
.40
11.3
.60
3
3
0
0
2856
KANSAS CITY
10.9
.42
7.1
.39
5
4
0
0
1879
KNOXVILLE
15.6
.60
16.1
.85
6
3
0
0
3070
LAS VEGAS
3.6
.13
2.1
.10
4
4
0
0
2063
LEXINGTON
12.4
.51
9.6
.55
5
4
0
0
2643
LITTLE ROCK-PINE BLUFF
14.1
.62
8.9
.53
7
3
0
0
2289
LOS ANGELES
3.9
.14
3.3
.17
9
5
0
0
1878
LOUISVILLE
8.3
.33
7.2
.39
4
4
1
0
2325
MADISON
14.7
.48
8.4
.42
5
3
0
0
2278
MEMPHIS
5.6
.23
5.1
.30
4
6
0
0
2004
MIAMI-FT. LAUDERDALE
5.0
.19
4.3
.22
4
4
0
0
2232
MILWAUKEE
12.7
.78
18.7
.93
6
4
1
1
3098
MINNEAPOLIS-ST. PAUL
18.2
.61
18.2
.88
12
5
1
1
1598
MOBILE-PENSACOLA
9.3
.35
7.4
.40
3
4
0
0
1682
NASHVILLE
8.7
.34
7.5
.38
8
5
0
0
1819
NEW ORLEANS
24.8
.99
26.4
1.5
7
5
1
0
2221
NEW YORK
11.3
.37
8.3
.37
12
8
0
0
1355
NORFOLK-PORTSMTH-NEWPT NWS
11.7
.44
10.3
.53
7
3
0
0
2656
OKLAHOMA CITY
12.0
.45
10.5
.52
6
4
0
0
2663
OMAHA
17.2
.60
9.6
.46
4
3
0
0
2204
ORLANDO-DAYTONA BCH-MELBRN
7.9
.28
9.0
.44
3
5
0
0
1702
PADUCAH-C.GIRD-HARBG-MT VN
9.5
.30
4.4
.25
3
3
0
0
2954
FCC PUR11000027

DRAFT SUBMITTED 4.6.2011

HINDMAN
LESS OF THE SAME: LOCAL NEWS ON THE INTERNET
39
PHILADELPHIA
11.5
.41
11.0
.54
12
5
1
0
1099
PHOENIX
10.1
.34
11.0
.55
5
4
1
0
3712
PITTSBURGH
25.3
.91
21.9
1.1
12
4
0
0
2011
PORTLAND, OR
9.8
.33
8.5
.40
9
3
1
0
1761
PORTLAND-AUBURN
7.8
.28
5.7
.27
4
3
0
0
2410
PROVIDENCE-NEW BEDFORD
10.8
.37
7.6
.36
9
5
1
0
1789
RALEIGH-DURHAM
23.3
.87
33.3
1.7
5
4
1
0
6939
RICHMOND-PETERSBURG
8.6
.40
6.1
.33
2
4
1
0
3334
ROANOKE-LYNCHBURG
31.8
1.3
20.4
1.1
3
3
0
0
3229
ROCHESTER, NY
13.3
.46
9.3
.41
3
3
0
1
2550
SACRAMNTO-STKTN-MODESTO
9.8
.28
5.4
.27
10
6
0
0
1901
SALT LAKE CITY
89.0
3.2
58.4
3.1
6
5
0
0
8642
SAN ANTONIO
9.7
.40
7.6
.43
3
4
0
0
2986
SAN DIEGO
9.3
.30
9.6
.48
3
6
1
1
3311
SAN FRANCISCO-OAK-SAN JOSE
6.1
.21
5.9
.28
7
4
1
1
2699
SAVANNAH
6.8
.29
3.2
.18
4
3
0
0
2740
SEATTLE-TACOMA
10.9
.36
8.7
.41
11
4
1
1
1065
SHREVEPORT
10.0
.46
5.3
.32
2
3
0
0
4962
SOUTH BEND-ELKHART
18.7
.65
12.2
.57
6
3
0
0
2838
SPOKANE
4.6
.17
4.1
.21
4
2
0
0
3263
SPRINGFIELD, MO
15.4
.62
7.2
.38
4
2
1
0
3451
ST. LOUIS
15.9
.60
14.7
.95
7
5
0
0
2213
SYRACUSE
25.9
.99
22.5
1.1
4
2
1
0
3679
TAMPA-ST. PETE,SARASOTA
11.2
.37
8.2
.37
9
3
0
0
1831
TOLEDO
9.2
.35
8.3
.43
5
4
0
0
2403
TRI-CITIES, TN-VA
17.9
.73
11.9
.65
5
1
0
0
2444
TUCSON (NOGALES)
9.7
.35
7.8
.39
1
3
0
1
4629
TULSA
12.8
.50
10.7
.60
4
5
1
0
2625
WACO-TEMPLE-BRYAN
14.0
.49
6.2
.33
4
4
0
0
3235
WASHINGTON, DC
10.7
.37
7.5
.36
8
4
1
1
3103
WEST PALM BEACH-FT. PIERCE
8.8
.33
5.5
.27
3
4
0
0
3059
WICHITA-HUTCHINSON PLUS
13.3
.51
10.4
.55
6
4
0
0
1769
WILKES BARRE-SCRANTON
22.2
.76
11.8
.56
13
3
0
0
1938

FCC PUR11000027

DRAFT SUBMITTED 4.6.2011

Note: We are currently transitioning our documents into web compatible formats for easier reading. We have done our best to supply this content to you in a presentable form, but there may be some formatting issues while we improve the technology. The original version of the document is available as a PDF, Word Document, or as plain text.

close
FCC

You are leaving the FCC website

You are about to leave the FCC website and visit a third-party, non-governmental website that the FCC does not maintain or control. The FCC does not endorse any product or service, and is not responsible for, nor can it guarantee the validity or timeliness of the content on the page you are about to visit. Additionally, the privacy policies of this third-party page may differ from those of the FCC.