Social Welfare Final Paper
Local Television News Coverage of the Affordable Care Act: Emphasizing Politics Over Consumer Information Sarah E. Gollust, PhD, Laura M. Baum, MURP, Jeff Niederdeppe, PhD, Colleen L. Barry, PhD, MPP, and Erika Franklin Fowler, PhD Objectives.To examine the public health and policy-relevant messages conveyed through local television news during thefirst stage of Affordable Care Act (ACA) implementation, when about 10 million Americans gained insurance.
Methods.We conducted a content analysis of 1569 ACA-related local evening tele- vision news stories, obtained from sampling local news aired between October 1, 2013, and April 19, 2014. Coders systematically collected data using a coding instrument tracking major messages and information sources cited in the news.
Results.Overall, only half of all ACA-related newscoveragefocusedonhealthinsurance products, whereas the remainder discussed political disagreements over the law. Major policy tools of the ACA—the Medicaid expansion and subsidies available—were cited in less than 10% of news stories. Number of enrollees (27%) and Web site glitches (33%) were more common features of coverage. Sources with a political affiliationwerebyfarthemost common source of information (>40%), whereas research was cited in less than 4% of stories.
Conclusions.The most common source of news for Americans provided little public health–relevant substance about the ACA during its early implementation, favoring political strategy in coverage. (Am J Public Health.2017;107:687–693. doi:10.2105/ AJPH.2017.303659) See also Parekh, p. 639. I nthefirst year of full implementation of the Affordable Care Act (ACA; Pub LNo.111–148), more than 10 million Americans gained insurance through state and federal marketplaces or through the expansion of the Medicaid program in 24 states to low-income, childless adults. 1These insurance gains were both a public health and health communication success, because those who gained insurance had to recognize their need for insurance, become aware of their plan options, seek help if needed with the application process, and purchase or enroll in health insurance. Survey data indicate that an important source of in- formation about health insurance options was the local news. 2An April 2014 survey of new enrollees conducted by Enroll America revealed that news was the top source of information cited by enrollees, and of those citing news as a top source, 42% cited local television news—more than any othernews media type—as their most influential information source. 3 Local news both provided information to new enrollees and exposed many more millions of Americans to information with which they might interpret the success or failure of ACA implementation. Jacobs and Mettler reported significant shifts in public attitudes toward the ACA between 2010 and 2014, an attitude change that they attribute to Americans’evaluations of the benefits followingfrompolicyimplementation. 4 Given that a relatively small proportion of Americans experienced the law directly, most Americans’perceptions likely came from media depictions, at least atfirst. In fact, in February 2014, 44% of Americans in- dicated that their impression of the ACA was based on the media, and only 23% from one’s own experience. 2New evidence suggests that individuals exposed to higher volumes of ACA-related media in October 2013 were more likely to believe that they were suffi- ciently knowledgeable about the law. 5 The aforementioned evidence suggests a relationship between public attitudes about health policy and media coverage volume, but the content of these media portrayals (e.g., the specific messages and tone con- veyed in news coverage) may be even more important. Although a few previous studies have examined news media coverage of the lengthy congressional deliberations leading up to the passage of the ACA, 6–8 surveys confirm low levels of knowledge about the law prior to the start of the open enrollment period. 9Therefore, the initial policy implementation stage is when Americans could have actually learned about the details of the new insurance options available to them. Yet no previous research has explored in depth the public health– relevant content of news coverage of the ACA in this early implementation phase. 10 The current study provides a compre- hensive analysis of the content of local ABOUT THE AUTHORSSarah E. Gollust is with the Division of Health Policy and Management, University of Minnesota School of Public Health, Minneapolis. Laura M. Baum is with the Wesleyan Media Project, Wesleyan University, Middletown, CT. Jeff Niederdeppe is with the Department of Communication, Cornell University, Ithaca, NY. Colleen L. Barry is with the Department of Health Policy and Management, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD. Erika Franklin Fowler is with the Department of Government and the Wesleyan Media Project, Wesleyan University, Middletown, CT.
Correspondence should be sent to Sarah E. Gollust, Division of Health Policy and Management, University of Minnesota School of Public Health, 420 Delaware St SE, MMC 729, Minneapolis, MN 55455 (e-mail: [email protected]). Reprints can be ordered at http://www.ajph.org by clicking the“Reprints”link.
This article was accepted January 2, 2017.
doi: 10.2105/AJPH.2017.303659 May 2017, Vol 107, No. 5AJPHGollust et al.Peer ReviewedPublic Health Policy687 AJPHPOLICY television news coverage about the ACA during thefirst open enrollment period running from October 1, 2013, to April 19, 2014. Although viewership has declined somewhat (particularly among younger people), local television is still the news source most watched by Americans over- all. 11This study provides important con- textual information for policymakers, health policy scholars, and enrollment advocates seeking to better understand the information environment in the period during which 10 million Americans gained insurance. METHODS To identify local TV news stories about theACA,weconductedsearchesfor“health care,”“healthcare,”“Obama care,”“Oba- macare,”or“Affordable Care Act”in closed captions of the top 2 highest-rated local evening news broadcasts in 208 US media markets for the period that defined thefirst ACA Open Enrollment Period: October 1, 2013, to April 19, 2014 (including the 2-week special enrollment period following the official end on March 31). Two media markets, Lake Charles (Louisiana) and Bowling Green (Kentucky), did not have closed captioning and therefore were not captured at all. These searches yielded a total of 150 534“hits,”or matches to the keyword search strings. We next constructed a sample of news content to identify a man- ageable set of stories for comprehensive content coding, using methods employed and validated in previous work. 12,13 For each monthinoursample,wecreateda“con- structed week”sample of news coverage, randomly selecting 1 Monday, Tuesday, Wednesday, Thursday, Friday, Saturday, and Sunday from each month. This sampling procedure accounts for the cyclic nature of news content. We then limited the sample to just the news hits that aired on dates within the constructed weeks. This process yielded 5301 total hits.
Thefirst step in the data collection process was to check whether the hits were actually stories about the ACA. Trained coders either watched the entire news clip or, if video was not available, relied on the closed caption transcripts to determine whether it was an ACA-related story. For a hit to beclassified as an ACA-related story, (1) it had to mention the ACA, Obamacare, the health care law, health care reform, or Medicaidexpansioninthestates,and(2) it could not be a duplicate of an already identified ACA story. We excluded a sub- stantial number of hits that coders identified as not being ACA stories, leaving a sample of 2976 identified stories about the ACA.
More than a third of the identified stories (n = 1063) came from the month of October. November had the next highest (n=588) followed by March (n=467; Figure A, available as a supplement to the online version of this article at http:// www.ajph.org).
Missing data from certain news programs or stations are relatively common when collecting data from closed captioning, and our study is no exception. From the sample of 2976 ACA stories, 1407 had missing video or video–audio issues, such that they could not be coded. Fortunately, given in- formation on which videos had problems, we confirmed that the missing data were well dispersed across markets and months and not systematically related to particular time periods, with the exception of October 21 and October 25, 2013, dates for which we have no video for any market. We present proportions in ourfindings and have no reason to believe that these proportions generated from coded data would system- atically differ in the missing data. Our final sample for content coding included 1569 ACA-related stories. Content Coding Instrument We developed a coding instrument to capture information about the ACA in local news through an iterative process. After screening each hit to identify whether it was an ACA story, the research team then proceeded to code stories for a variety of content features. Coders assessed whether the story concerned mainly (1) the politics of the ACA (i.e., disagreement over the role of government, political disagreements over implementation, plans to repeal and replace the law), (2) the health insurance products available through the ACA and the enroll- ment process, or (3) a combination of politics and health insurance product–related content.Within the subsample of news stories that had at least some health insurance product–related content, we then coded for aseriesofhealthpolicy–relevant variables.
Specifically, we coded for the overall tone of coverage about the new health insurance marketplace, the major messages about the lawanditseffects,theframingofthenameof the law, the frequency with which stories featured anecdotes of people affected, and the sources cited in coverage (coding in- strument is given in the online Appendix).
We coded most variables dichotomously— that is, we coded each story element (e.g., whether the story described Web site mal- functions) as either“present”or“absent.” For 1 content code (Medicaid content), we also assessed whether that theme was a“focus of the story”(coded 2), comprised “only a mention”(coded 1), or was“not mentioned”at all (coded 0). We assessed the tone of news coverage about health insurance enrollment using a 4-category variable, coded as neutral, discouraging, encouraging, or mixed (both encouraging and discouraging). We also created a tone summary score by coding discouraging coverage as–1, mixed or neutral coverage as 0, and encouraging coverage as 1, so average valuesabove0arepositiveorencourag- ing and values below 0 are negative or discouraging. Coding Process and Interrater Reliability A team of 32 student coders coded all stories after an extensive training period to ensure reliable application of the coding instrument. For all categorical variables, we estimated item-level interrater reliability (IRR) usingkstatistics, a measure that es- timates coder agreement for categorical variables and adjusts for the likelihood of chance agreement. 14For 1 ordinal variable (tone of coverage), we estimated Krippen- dorff’salpha. 15For some relatively low- frequency items, IRR statistics fell below 0.70, and these variables are not reported here. The IRR for our main news coverage type variable (political coverage vs insurance product coverage vs a mix) was 0.76, whereas the IRR for tone was 0.74. The IRR of all other variables reported herein ranged from 0.70 to 1.0. AJPHPOLICY 688Public Health PolicyPeer ReviewedGollust et al.AJPHMay 2017, Vol 107, No. 5 Analysis Thefirst stage of the analysis was to calculate descriptive statistics of the key messages in news coverage about the ACA.
Because we anticipated that local news coverage would differ both geographically and over time, we constructed geographic information system (GIS) maps of news coverage tone and also compared all news coverage variables for the 2013 period (October 1–December 31, 2013) and the 2014 period (January 1–April 19, 2014) or between states that operated their own marketplaces versus those that relied on the federal marketplace, both using 2-sample tests of proportions. RESULTS About one quarter (26.5%; n = 416) of the 1569 stories in our analytic sample were all about the politics of the ACA (i.e., political disagreements over imple- mentation, discussion of repealing the law) and did not mention any information about health insurance products. By contrast, just under half (44.9%; n = 705) of the storieswere all about the health insurance options available through the ACA. The remaining group (28.6%; n = 448) presented a mix of coverage of both politics and health in- surance options available. Overall, then, more than half of all news stories (55.1%, n = 865) presented at least some coverage of politics. This composition changed over time. Figure 1 displays the composition of content over the 6 months, showing that news coverage of political disagreements (those stories that had some or all of their content devoted to politics) was dominant during thefirst 2 months of open enroll- ment; although political coverage declined somewhat thereafter, politics persisted as a major component of news coverage throughout the period.
Relatively few of the 1569 news stories mentioned Medicaid (7.4%) or had a Med- icaid focus (5.0%); more of these stories focusing on Medicaid were in the subsample that was determined to be all about the politics (10.1% vs 3.1%;P<.01). More than half of the news stories (51.9%) referred to the law by its full name, the Affordable Care Act, or its abbreviation, the ACA. About a third referred to Obamacare (38.8%); other references included the health care law(11.2%) and the President’s health care law (3.8%), or referred to healthcare.gov or state-based exchanges. Stories within the health insurance product–focused subsample were more likely than politics-focused stories to reference the Affordable Care Act or ACA (57.8% vs 35.3%,P<.01), whereas stories in the politics-focused subsample were more likely than product-focused stories to refer to Obamacare (43.8% vs 36.9%,P= .015).
Overall, 57.2% of local TV news coverage was encouraging, 20.6% was discouraging, 13.3% was mixed, and 8.8% was neutral.
Figure 2 maps the tone of coverage across the 208 included local media markets, using a measure ranging from–1 (all discouraging) to +1 (all encouraging), and averaged for each market and across the full open en- rollment period. This map reveals high levels of geographic variation in the tone of ACA news coverage across the United States. A few states’sampled local TV news coverage was entirely positive (e.g., North Dakota, Nevada, Arizona, Hawaii), but most states had intrastate variation in news coverage tone. No state had entirely neg- ative coverage, on average. We tested whether tone of coverage varied by whether 0 10 20 3040 50 60 70 80 90 100 October 2013 November 2013 December 2013 January 2014 February 2014 March 2014 April 2014* Coverage is somewhat or all about poli cs Coverage is all about health insurance op ons Percentage Month Note.April 2014 was a partial month (April 1–19). Lines display the percentage of total news coverage in each month that had at least some or all of its content about politics (e.g., political disagreements over implementation, discussion of repealing the law) vs coverage that was all about the health insurance options available through the ACA.
FIGURE 1—Composition of News Coverage of the Affordable Care Act (ACA) in Local Television News Broadcasts (n = 1569 stories): United States, October 2013–April 2014 AJPHPOLICY May 2017, Vol 107, No. 5AJPHGollust et al.Peer ReviewedPublic Health Policy689 the states were implementing their own marketplaces versus relying on the federal marketplace; we found no differences be- tween the tone of coverage in state-based marketplace states (mean tone = 0.341; SD = 0.80) compared with federal market- place states (mean tone = 0.374, SD = 0.80; t=–0.60,P= .55). There were statistically significant differences in the tone of cov- erage over time. News coverage in 2013 (mean tone = 0.238; SD = 0.84) was sig- nificantly more discouraging than news coverage in 2014 (mean tone = 0.513; SD = 0.74;t=–5.88;P<.01).
Table 1 displays the major messages in- cluded in the overall sample of news that was at least somewhat about health insurance products (n = 1153), divided into the 2 time periods (late 2013 vs early 2014). Overall, about a third of product-related news coverage—and a significantly larger share in 2014, about 4 in 10 news stories—included some factual policy details about enrollment such as mention offines, subsidies, or how to get help to enroll. A small proportion of news coverage (13.9%) provided viewers with information on where to go or a number to call to get help. Less than 7% of all stories indicated that subsidies areavailable to make health insurance more affordable. A comparison of states that of- fered their own marketplaces and those that relied on the federal marketplace revealed no difference in reporting on subsidies; overall, however, the coverage in news media markets within states defined as fed- eral marketplace states reported on policy details about enrollment (in 39.2% of stories) more than in state-based marketplace states (27.9%;P<.01).
About one fourth of health insurance product–related news stories overall—and about 38% in 2014—presented information on the number of new enrollees to date.
Another relatively common message, par- ticularly in 2014 (14.2% of stories), de- scribed federal or state enrollment goals.
Very few news stories (1.8%) mentioned any research or evaluation evidence about the ACA.Websiteglitcheswereacommon message in news coverage; one third of news stories (33.3%) mentioned some type of Web site glitch (including healthcare.gov or state-based marketplace sites). As ex- pected, such stories were significantly more common in 2013 (41.0%) than in 2014 (24.4%;P<.01). Finally, narratives or exemplars describing particular individualsaffected by the law were featured in few news stories overall (less than 10%).
Storiesweremorelikelytodescribepeople helped by the law in 2014 than in 2013 (8.4% vs 2.4%;P<.01).
Table 2 displays the sources of in- formation cited in local TV news stories about the ACA, combining neutral in- formation sources and those cited with particular positions about the law. The most common sources overall were President Obama (cited in 38.9% of news stories) and a White House or other federal executive agencyfigure (28.7%); also common were Republican (22.3%) or Democratic (15.9%) politicians or officials.
Each of the other sources was cited in less than 10% of news stories, including governors, health insurance navigators, or insurers. Researchers or those with an academic, think tank, or foundation affil- iation were cited in only 3.9% of news stories. President Obama, Republicans, and Democrats were all more likely to be included as sources of information in 2013 than in 2014 (P<.001). Insurance company representatives were also more likely to be cited in 2013 (9.1%) than in 2014 (2.8%;P<.001). News Coverage Tone –1.00 to –0.51 –0.50 to –0.11 –0.10 to 0.10 0.11 to 0.50 0.51 to 1.00 Note.For 208 of the 210 television media markets in the United States, trained coders identified the number of local news stories covering the ACA and assessed the story tone (discouraging, neutral, or encouraging) for the insurance product–oriented news segments (n = 1153) using closed captioning searches. Story tone displayed is the average tone within a media market, from–1 (very discouraging) to 0 (neutral or balanced) to +1 (very encouraging).
FIGURE 2—Geographic Variation in Tone of Local News Coverage of the Affordable Care Act (ACA): United States, October 1, 2013–April 19, 2014 AJPHPOLICY 690Public Health PolicyPeer ReviewedGollust et al.AJPHMay 2017, Vol 107, No. 5 DISCUSSION This study is thefirst to our knowledge to comprehensively examine local TV news media messaging about the ACA during its critical implementation stage, the fullfirst open enrollment period. We found that only half of all news coverage about the ACA focused on the health insurance products available through the ACA. Even within the news coverage that did discuss health insurance products, key policy as- pects of the ACA content were surprisingly uncommon. For instance, policy analysts attribute a large portion of insurance gains from the ACA to the voluntary state Medicaid expansions to low-income childless adults, 1yet Medicaid was only mentioned in 7% of news stories. Only 7% of news stories discussed the availability of subsidies, a critical provision for en- couragingenrollmentbymakingcoverage more affordable. Overall, fewer than 4 in10 news stories presented policy details that would be useful for consumers interested in enrollment, such as where to sign up, subsidies, or the availability of navigators for assistance.
Rather, news stories provided compar- atively more attention to reporting on figures on enrollment expected or achieved or to Web site glitches. These emphases are consistent with the typical political strategy or“horse race”–focused journalistic style of covering campaigns or policy contests—with an emphasis on who is winning or losing. 16–18 This manner of coverage is also consistent with how the political deliberations prior to the passage of the ACA were covered in news media in 2009 and 2010, 7andalsowithhowthenews media tended to cover the health care re- form legislative debates in the 1990s. 19,20 In this case, political strategy framing was represented by a strong emphasis on“wins” for ACA proponents (e.g., mentions in newscoverage of the numbers of Americans who had established accounts, signed up, and gained insurance) compared with“losses” for ACA proponents (e.g., mentions of Web site or enrollment glitches). Attention to Web site glitches decreased but persisted through the latter half of thefirst open enrollment period, appearing in a full quarter of news stories in 2014.
This emphasis on strategic over sub- stantive reporting in local TV news is further supported by the sources that reporters solicited for their perspectives. Political and partisan sources dominated; these elected officials’viewpoints were far more repre- sented in local news coverage about the ACA than any other group. Surprisingly, few news stories included any public health, medical, research, or health advocacy per- spectives. This dominance of political sources suggests that viewers’impressions of the law, already polarized dating back to the initial legislative debates in 2009, would remain so during the policy implementation stage. Research indicates that when political source cues are presented in news media, particularly for issues that are already po- liticized, changes in public perceptions are unlikely because viewers use these political cues to interpret andfilter the information presented, in accordance with their own predispositions. 21 Indeed, public opinion data demonstrate strong and persistent po- litical polarization in public views of the ACA. 4Thesefindings help explain why the law’sfateisinjeopardyasofthiswriting:
even during the early phase of ACA implementation, the local news discussed the law using a political strategy frame (similar to political campaign coverage), thus limiting citizens’exposure to the substance of ACA policy content while heightening the public’s likelihood of per- ceiving the law via politically charged heuristics. Limitations This research has several limitations worth noting. First, although this large-scale content analysis comprehensively analyzed the messages in a constructed week sam- ple of video and transcripts for more than 1100localTVnewsstoriesabouttheACA across media markets, because of missing TABLE 1—Affordable Care Act (ACA) Implementation Messages in Local News Stories About the ACA, Over Time: United States, October 1, 2013–April 19, 2014 MessageOverall Sample, % (n = 1153)2013 Stories, %(n=617)2014 Stories, %(n=536) Pof Difference by Yeark a Factual policy details for individual enrollees Any mention offines, subsidies, navigators, questions and answers, enrollment events or information36.3 31.9 41.4<.001 0.7989 Fines or penalties for not enrolling 16.4 12.3 21.1<.001 0.8908 Information on where to go or number to call 13.9 12.3 15.7 .10 0.8508 Subsides are available 6.9 6.0 7.8 .22 0.9020 Coverage of a past enrollment or informational event4.6 3.6 5.8 .07 0.7456 Population-level contextual information about law Number of enrollees to date 27.0 17.7 37.7<.001 0.8662 Federal or state enrollment goal 9.1 4.7 14.2<.001 0.8032 Number of Web site visitors 5.2 3.2 7.5 .001 0.7136 Mentions research or evaluation about ACA 1.8 1.3 2.4 .15 1.0 Negative information about ACA implementation or enrollment Any Web site glitch 33.3 41.0 24.4<.001 0.8118 Healthcare.gov Web site glitch 19.1 27.1 9.9<.001 0.7880 State Web site glitches 4.8 3.1 6.7 .004 0.7247 Narratives about people affected Any narrative 9.3 8.4 10.3 .28 0.7838 Narrative of an individual or family helped by law5.2 2.4 8.4<.001 0.9501 aEstimate of the interrater reliability for each variable across the team of coders. AJPHPOLICY May 2017, Vol 107, No. 5AJPHGollust et al.Peer ReviewedPublic Health Policy691 data, this sample is not representative of all local television news coverage about the ACA. However, our investigations into the missing data by network and station rep- resentation indicate that any bias in in- cluded versus missing data is small (e.g., our sample included 20% NBC affiliates whereas the original sampling frame included 22% NBC affiliates).
Second, this study only includes local broadcast TV news coverage of the ACA, not other sources of news coverage including national news, online news, print news, or radio or other media. However, local TV news remains the most common source of news for Americans overall 11and the most common source of information for new 2014 enrollees in particular.
3Furthermore, given the localized nature of health insurance—with plans and prices varying dramatically across the United States—local news sources are likely among the most relevant sources of information about health insurance enrollment for consumers.
Third, as with all media content coding, there is some subjectivity to the process of interpreting video or text. Given the sheer volume of news to code, we relied on more than 30 coders. However, interrater re- liability was greater than 0.70 for all codedmeasures included, increasing our confi- dence in the coding process. That said, we required precise and specific references for particular terms; thus, we may have missed more oblique messages that could be better captured using a qualitative approach to content analysis.
Finally, this is a descriptive study of news content. This work is important for understanding what Americans were plau- sibly exposed to in the local news environ- ment; however, we cannot assert claims of actual news effects with the data and design presented here. In other work, we are linking media data to survey data at re- spondents’county of residence to better understand how ecological measures of media are associated with various outcomes, including public attitudes and health in- surance enrollment. 5,22 Implications and Conclusions As we look back at 2014, the ACA achieved great insurance gains in spite of a local TV news environment that relatively infrequently provided detailed and policy- relevantinformationtoconsumers.Many other media sources—including paid media by insurers, states, and healthcare.gov—mayhave at least partiallyfilled this information vacuum. It will be important to build on this analysis by quantifying the extent to which various media sources and channels were associated with attitudinal or behav- ioral changes among both potential new health insurance enrollees and the general public during this critical phase of ACA implementation, as well as understanding how the media will contribute to public perceptions in a post-2016 political era wherein repeal or changes to the ACA loom large. CONTRIBUTORSS. E. Gollust drafted the article, provided project lead- ership, and secured funding. L. M. Baum coordinated the data collection and conducted the data analysis and data visualization. J. Niederdeppe and C. L. Barry provided critical revisions and interpretation. E. F. Fowler led the data collection and analysis process. All authors collec- tively contributed to the study design and data in- terpretation and approved thefinal version of the article.
ACKNOWLEDGMENTSThis research was funded by the Robert Wood Johnson Foundation State Health Access Research and Evaluation (SHARE) program, grant no. 72179. This research was also supported through the McKnight Land-Grant Professorship at the University of Minnesota; additional support was provided by Wesleyan University.
We presented an earlier version of this article at the Association for Public Policy Analysis and Management Fall Research Conference on November 5, 2016, in Washington, DC.
We thank the numerous student researchers at Wes- leyan who made this project possible, especially Courtney Laermer and Alison Mann. Finally, we thank CommIT Technology Solutions for coding system support and acquisition of data.
HUMAN PARTICIPANT PROTECTIONInstitutional review board approval was not needed because there were no human participants involved in the study.
REFERENCES1. Sommers BD, Musco T, Finegold K, Gunja MZ, Burke A, McDowell AM. Health reform and changes in health insurance coverage in 2014.N Engl J Med. 2014;371(9):
867–874.
2. Kaiser Family Foundation. Kaiser Health Policy News Index: February 2014. Available at: http://kff.org/health- reform/poll-finding/kaiser-health-policy-news-index- february-2014. Accessed January 30, 2017.
3. Enroll America.“Why did some people enroll and not others?”Survey for Enroll America, conducted by Perry- Undem Research/Communication. 2014. Available at:
https://s3.amazonaws.com/assets.enrollamerica.org/ wp-content/uploads/2014/05/EA-PerryUndem-Survey- Report.pdf. Accessed January 30, 2017.
4. Jacobs LR, Mettler S. Liking health reform but turned off by toxic politics.Health Aff (Millwood). 2016;35(5):
915–922.
5. Fowler EF, Baum L, Barry C, Niederdeppe J, Gollust SE. Media messages and perceptions of the Affordable TABLE 2—Sources Cited in Local News Stories About the Affordable Care Act (ACA), Over Time: United States, October 1, 2013–April 19, 2014 SourceOverall Sample, % (n = 1153)2013 Stories, % (n = 617)2014 Stories, % (n = 536) Pof Difference by Yeark a President Obama 38.9 46.4 30.2<.001 0.8146 Executive: White House Administration, HHS, other federal agency28.7 27.2 30.4 .23 0.7240 Republican politicians or officials at any level 22.3 30.6 12.7<.001 0.8832 Democratic politicians or officials at any level (not Obama or administration)15.9 24.6 5.8<.001 0.7858 Governor 4.1 3.9 4.3 .73 0.8191 Navigators or certified application counselors 6.9 5.5 8.4 .053 0.7140 Insurance representative 6.2 9.1 2.8<.001 0.7716 Researcher, academic, think tank, foundation 3.9 3.2 4.7 .21 0.7933 Note.Table does not include sources cited in less than 1% of news coverage, including religious or- ganizations and the advocacy group Enroll America. We also attempted to code for additional sources, including representatives of business, patient advocacy organizations, state exchange or other state authority, physicians, and average citizens, but reliability fell below 0.70 for these codes, likely because they had very low frequency (below 5% of news stories). HHS = Department of Health and Human Services.
aEstimate of the interrater reliability for each variable across the team of coders.
AJPHPOLICY 692Public Health PolicyPeer ReviewedGollust et al.AJPHMay 2017, Vol 107, No. 5 Care Act during the early phase of implementation. J Health Polit Policy Law . 2017;42(1):167 –195. 6. Conway BA. Addressing the “medical malady ”: second-level agenda setting and public approval of“ Obamacare. ”Int J Public Opin Res . 2013;25(4):535 –546. 7. Fahmy S, McKinley CJ, Filer CR, Wright PJ. Pulling the plug on Grandma: Obama ’s health care pitch, media coverage & public opinion. Adv Journal Commun . 2013; 1(3):19 –25. 8. Pelika SL, Clark JH, Rigby E. “Don ’t let government touch my Medicare ”: how media framing shapes public support for health care reform. 2010. Working paper.Available at: http://opensiuc.lib.siu.edu/pnconfs_2010/17. Accessed January 30, 2017.
9. Kaiser Family Foundation. Kaiser Health Tracking Poll: September 2013. Available at: http://kff.org/health-reform/poll- finding/kaiser-health-tracking- poll-september-2013. Accessed January 30, 2017.
10. Hopper J. Obamacare, the news media, and the politics of 21st-century presidential communication. Int J Commun . 2015;9:1275 –1299. 11. Pew Center for People & the Press. The modern news consumer: news attitudes and practices in the digital era. ” July 2016. Available at: http://assets.pewresearch.org/ wp-content/uploads/sites/13/2016/07/PJ_2016.07.07_ Modern-News-Consumer_FINAL.pdf. Accessed Janu-ary 30, 2017.
12. Slater MD, Long M, Bettinghaus EP, Reineke JB. News coverage of cancer in the United States: a nationalsample of newspapers, television, and magazines. J Health Commun . 2008;13(6):523 –537. 13. Luke DA, Caburnay CA, Cohen EL. How much is enough? New recommendations for using constructedweek sampling in newspaper content analysis of healthstories. Commun Methods Meas . 2011;5(1):76 –91. 14. Lombard M. A call for standardization in content analysis reliability. Hum Commun Res . 2004;30(3): 434 –437. 15. Hayes AF, Krippendorff K. Answering the call for a standard reliability measure for coding data. Commun Methods Meas . 2007;1(1):77 –89. 16. Gans H. Deciding What ’s News . New York, NY: Vintage; 1979.
17. Patterson TE. Out of Order . New York, NY: Vintage Books; 1994.
18. Lawrence RG. Game-framing the issues: tracking the strategy frame in public policy news. Polit Commun . 2000; 17(2):93 –114. 19. Cappella JN, Jamieson KH. Spiral of Cynicism: The Press and the Public Good . Oxford, UK: Oxford University Press; 1997.
20. Jacobs L, Shapiro R. Politicians Don ’t Pander: Political Manipulation and the Loss of Democratic Responsiveness . Chicago, IL: University of Chicago Press; 2000.
21. Druckman JN, Peterson E, Slothuus R. How elite partisan polarization affects public opinion formation. Am Polit Sci Rev . 2013;107(1):57 –79. 22. Karaca-Mandic P, Wilcock AD, Baum LM, et al. Advertising and the uninsured: relationship between health insurance marketing and county-level insurance enrollment in 2014. Poster presented at the Academy-Health Annual Research Meeting; Boston, MA; June2016. Available at: https://academyhealth.confex.com/academyhealth/2016arm/meetingapp.cgi/Paper/11440.Accessed January 30, 2017. AJPH POLICY May 2017, Vol 107, No. 5 AJPH Gollust et al. Peer Reviewed Public Health Policy 693 Copyright ofAmerican JournalofPublic Health isthe property ofAmerican PublicHealth Association anditscontent maynotbecopied oremailed tomultiple sitesorposted toa listserv without thecopyright holder'sexpresswrittenpermission. However,usersmayprint, download, oremail articles forindividual use.