CHRIS J. VARGO, LEI GUO, DR. MAXWELL MCCOMBS
AND DR. DONALD LEWIS SHAW
Network Issue Agendas On Twitter During
e 2012 U.S. Presidential Election
AGENDA SETTING, LEVEL 1
Time1/Media Time2/Public
Economy
Equality
Drugs
War
Economy
Equality
Drugs
War
NETWORK AGENDA SETTING MODEL
¢ e salience of network relationships among
objects/attributes can be transferred from one
agenda to another.
—  McCombs’ new book designates this the “3rd level of
agenda-setting theory.”
¢ News media not only tell us what to think about
and how to think, they might also be capable of
telling us what and how to associate.
Economy
Federal
Programs
Foreign
Policies
Indiv.
Liberties
Immigration
Economy
Federal
Programs
Foreign
Policies
Indiv.
Liberties
Immigration
Media issue agenda network
Public issue agenda network
TEST GROUPS: MEDIA (AGENDAMELDING)
¢  CBS, NBC, CNN, PBS..
¢  USA Today, NYT, WSJ, US Weekly..
¢  Denver Post, Pittsburgh Post Gazette..
General Public
Horizontal
Media
(Democratic)
Specific
PublicVertical
Media
¢  MSNBC, Hosts & Talk Shows
¢  Hardball, Maddow, John
Stewart, Anderson Cooper..
Horizontal
Media
(Republican)
Specific
Public
¢  FoxNews, Hosts & Talk Shows
¢  Rush Limbaugh, Glenn Beck,
O’Reilly, Hannity..
TEST GROUPS: CANDIDATE SUPPORTERS
¢  Users that Tweeted <12 times about one candidate, and not
another were subjected to a sentiment analysis
¢  Average sentiment was calculated for each user
¢  Average was subjected to a one-way t-test to see if sentiment
was significantly greater than 0 at the .10 level
—  DF = 1 - # of Tweets broadcasted about candidate
¢  2,875 Obama supporters found
¢  2,457 Romney supporters found
H1: QAP CORR. & VERTICAL MEDIA
¢ From the networked perspective
suggested by the NAS model, we
examined agendamelding by
investigating the correlations
between candidate supporters and
vertical media during the 2012
election.
¢ We hypothesized that the network
issue agenda of (a) Obama
supporters and (b) Romney
supporters was positively
correlated with the vertical
media’s network issue agenda.
H2: QAP CORR. & HORIZONTAL MEDIA
¢ As agendamelding theory
states, we also expected
candidate supporters to
correlate with horizontal
media.
¢ Specifically, we assume that
the network issue agenda of
(a) Obama supporters and (b)
Romney supporters is
positively correlated with the
network issue agenda of
horizontal media.
Obama &
Romney
Supporters
Horizontal
Democratic
Horizontal
Republican
Vertical
Media
Horizontal
Media
Republican
Specific
Public
H3: VERTICAL OVER HORIZONTAL
¢ According to the
findings in the 2008
study, we hypothesize
that vertical media will
offer more explanatory
power for the network
issue agendas of (a)
Obama supporters and
(b) Romney supporters
than horizontal media.
Democratic
Specific
Public
Republican
Horizontal
Media
Democratic
Horizontal
Media
H4: “TENDING TO THEIR OWN” MEDIA
¢  We expected to see
Republican horizontal
media offer more
explanatory power for the
network issue agendas of
Romney supporters.
¢  Likewise, we expected the
network issue agendas of
Obama supporters as
better explained by
Democratic horizontal
media than by Republican
horizontal media.
METHOD-NETWORK ANALYSIS
¢ Step 1: Create a matrix for each group
A.  Economy
B.  Foreign Policy
C.  Individual Liberties
D.  Federal Programs
E.  Immigration
F.  Education
G.  Environment
H.  Big Government
METHOD-NETWORK ANALYSIS
¢ Hypotheses 1 & 2:
—  Step 2: Conduct QAP correlations (H1/H2)
—  Step 3: Compare QAP correlations to see which
media predicted supporters the best (H1/H2)
—  Step 4: Create degree centrality scores for every
possible issue combination (H3/H4)
¢  Hypotheses 3 & 4:
—  Step 5: Preform linear regression to see if centrality
measures of one issue combination could predict the
others.
—  Step 6: Compare regression models and see which
media predicted supporters the best.
RESULTS H1: VERTICAL MEDIA
¢  Network issue agenda of Obama supporters was
positively correlated with the vertical media’s network
issue agenda
—  15 out of the 17 weeks analyzed.
—  QAP correlation coefficients: .54 to .91
—  Median of .72.
¢  Network issue agenda of Romney supporters was
significantly correlated with the vertical
media’s network issue agenda
—  13 out of the 17 weeks analyzed.
—  QAP correlation coefficients: .44 to .88
—  Median value of .77.
¢  Considerable support for H1
RESULTS H2: HORIZONTAL MEDIA
¢  Network issue agenda of Romney
supporters was positively correlated with
Republican horizontal media
—  13 out of 17 weeks
—  QAP correlation coefficients
during that period ranging from
.50 to .85. e median is .74.
¢  Network issue agenda of Obama
supporters and that of the Democratic
horizontal media were significantly
correlated
—  14 out of 17 weeks.
—  QAP correlation coefficients range from .56 to .
86, and the median is .72.
¢  Considerable support for H2.
Obama &
Romney
Supporters
Horizontal
Democratic
Horizontal
Republican
H3 RESULTS
¢  For Obama, the vertical media offered the
highest explanatory power
—  Highest r2 value for six of the eight issues analyzed.
—  e averaged r2 value across all eight issues suggests
that the vertical media were the most explanatory, an
increase of .20 r2 over the Democratic horizontal
media.
¢  For Romney, horizontal media offered the
highest explanatory power.
—  For the other seven issues, the compatible
(Republican) horizontal media were more
explanatory.
—  Averaged r2 value across all eight issues also suggests
the Republican horizontal media was the most
explanatory, an increase of .20 r2 over the vertical
media.
¢  e Republican horizontal media better
explained the network issue agendas of
Romney supporters on Twitter during the 2012
election.
Republican
Specific
Public
Democratic
Specific
Public
Democratic
Horizontal
MediaH4 RESULTS
¢  For Obama supporters, Democratic
horizontal media offered higher
explanatory power
—  Higher r2 value, for all of the eight
issues than the Republican horizontal
media.
—  Averaged r2 value across all eight issues
suggests that the Democratic horizontal media offered an
increase of .20 r2 over the Republican horizontal media. H4a
was strongly supported.
¢  For Romney supporters the Republican horizontal
media offered higher explanatory power,
—  Higher r2 value, for all but one of the eight issues than the
Democratic horizontal media.
—  Averaged r2 values across all eight issues indicate the
Republican horizontal media offered an increase of .13 r2 over
the Democratic horizontal media.
Republican
Horizontal
Media

Vargo, Guo, McCombs & Shaw - Network Issue Agendas on Twitter during the 2012 U.S. Presidential Election

  • 1.
    CHRIS J. VARGO,LEI GUO, DR. MAXWELL MCCOMBS AND DR. DONALD LEWIS SHAW Network Issue Agendas On Twitter During e 2012 U.S. Presidential Election
  • 2.
    AGENDA SETTING, LEVEL1 Time1/Media Time2/Public Economy Equality Drugs War Economy Equality Drugs War
  • 3.
    NETWORK AGENDA SETTINGMODEL ¢ e salience of network relationships among objects/attributes can be transferred from one agenda to another. —  McCombs’ new book designates this the “3rd level of agenda-setting theory.” ¢ News media not only tell us what to think about and how to think, they might also be capable of telling us what and how to associate.
  • 4.
  • 5.
    TEST GROUPS: MEDIA(AGENDAMELDING) ¢  CBS, NBC, CNN, PBS.. ¢  USA Today, NYT, WSJ, US Weekly.. ¢  Denver Post, Pittsburgh Post Gazette.. General Public Horizontal Media (Democratic) Specific PublicVertical Media ¢  MSNBC, Hosts & Talk Shows ¢  Hardball, Maddow, John Stewart, Anderson Cooper.. Horizontal Media (Republican) Specific Public ¢  FoxNews, Hosts & Talk Shows ¢  Rush Limbaugh, Glenn Beck, O’Reilly, Hannity..
  • 6.
    TEST GROUPS: CANDIDATESUPPORTERS ¢  Users that Tweeted <12 times about one candidate, and not another were subjected to a sentiment analysis ¢  Average sentiment was calculated for each user ¢  Average was subjected to a one-way t-test to see if sentiment was significantly greater than 0 at the .10 level —  DF = 1 - # of Tweets broadcasted about candidate ¢  2,875 Obama supporters found ¢  2,457 Romney supporters found
  • 7.
    H1: QAP CORR.& VERTICAL MEDIA ¢ From the networked perspective suggested by the NAS model, we examined agendamelding by investigating the correlations between candidate supporters and vertical media during the 2012 election. ¢ We hypothesized that the network issue agenda of (a) Obama supporters and (b) Romney supporters was positively correlated with the vertical media’s network issue agenda.
  • 8.
    H2: QAP CORR.& HORIZONTAL MEDIA ¢ As agendamelding theory states, we also expected candidate supporters to correlate with horizontal media. ¢ Specifically, we assume that the network issue agenda of (a) Obama supporters and (b) Romney supporters is positively correlated with the network issue agenda of horizontal media. Obama & Romney Supporters Horizontal Democratic Horizontal Republican
  • 9.
    Vertical Media Horizontal Media Republican Specific Public H3: VERTICAL OVERHORIZONTAL ¢ According to the findings in the 2008 study, we hypothesize that vertical media will offer more explanatory power for the network issue agendas of (a) Obama supporters and (b) Romney supporters than horizontal media. Democratic Specific Public
  • 10.
    Republican Horizontal Media Democratic Horizontal Media H4: “TENDING TOTHEIR OWN” MEDIA ¢  We expected to see Republican horizontal media offer more explanatory power for the network issue agendas of Romney supporters. ¢  Likewise, we expected the network issue agendas of Obama supporters as better explained by Democratic horizontal media than by Republican horizontal media.
  • 11.
    METHOD-NETWORK ANALYSIS ¢ Step 1:Create a matrix for each group A.  Economy B.  Foreign Policy C.  Individual Liberties D.  Federal Programs E.  Immigration F.  Education G.  Environment H.  Big Government
  • 12.
    METHOD-NETWORK ANALYSIS ¢ Hypotheses 1& 2: —  Step 2: Conduct QAP correlations (H1/H2) —  Step 3: Compare QAP correlations to see which media predicted supporters the best (H1/H2) —  Step 4: Create degree centrality scores for every possible issue combination (H3/H4) ¢  Hypotheses 3 & 4: —  Step 5: Preform linear regression to see if centrality measures of one issue combination could predict the others. —  Step 6: Compare regression models and see which media predicted supporters the best.
  • 13.
    RESULTS H1: VERTICALMEDIA ¢  Network issue agenda of Obama supporters was positively correlated with the vertical media’s network issue agenda —  15 out of the 17 weeks analyzed. —  QAP correlation coefficients: .54 to .91 —  Median of .72. ¢  Network issue agenda of Romney supporters was significantly correlated with the vertical media’s network issue agenda —  13 out of the 17 weeks analyzed. —  QAP correlation coefficients: .44 to .88 —  Median value of .77. ¢  Considerable support for H1
  • 14.
    RESULTS H2: HORIZONTALMEDIA ¢  Network issue agenda of Romney supporters was positively correlated with Republican horizontal media —  13 out of 17 weeks —  QAP correlation coefficients during that period ranging from .50 to .85. e median is .74. ¢  Network issue agenda of Obama supporters and that of the Democratic horizontal media were significantly correlated —  14 out of 17 weeks. —  QAP correlation coefficients range from .56 to . 86, and the median is .72. ¢  Considerable support for H2. Obama & Romney Supporters Horizontal Democratic Horizontal Republican
  • 15.
    H3 RESULTS ¢  ForObama, the vertical media offered the highest explanatory power —  Highest r2 value for six of the eight issues analyzed. —  e averaged r2 value across all eight issues suggests that the vertical media were the most explanatory, an increase of .20 r2 over the Democratic horizontal media. ¢  For Romney, horizontal media offered the highest explanatory power. —  For the other seven issues, the compatible (Republican) horizontal media were more explanatory. —  Averaged r2 value across all eight issues also suggests the Republican horizontal media was the most explanatory, an increase of .20 r2 over the vertical media. ¢  e Republican horizontal media better explained the network issue agendas of Romney supporters on Twitter during the 2012 election. Republican Specific Public Democratic Specific Public
  • 16.
    Democratic Horizontal MediaH4 RESULTS ¢  ForObama supporters, Democratic horizontal media offered higher explanatory power —  Higher r2 value, for all of the eight issues than the Republican horizontal media. —  Averaged r2 value across all eight issues suggests that the Democratic horizontal media offered an increase of .20 r2 over the Republican horizontal media. H4a was strongly supported. ¢  For Romney supporters the Republican horizontal media offered higher explanatory power, —  Higher r2 value, for all but one of the eight issues than the Democratic horizontal media. —  Averaged r2 values across all eight issues indicate the Republican horizontal media offered an increase of .13 r2 over the Democratic horizontal media. Republican Horizontal Media