Comparing the moral language of television to that of its viewers
This post is the second in a series on the moral language expressed on television. The previous post is available here.
The words we choose can convey our attitudes about the topic of conversation. For example, consider conversations about laboratory-grown meat. One speaker, with one ideological agenda, might use an expression like “Frankenmeat” to mean a monstrosity created by someone playing God. Another speaker, with a different agenda, might use an expression like “ethical meat” to mean steak without slaughter. A third speaker might opt for the expression “cultured meat” in an effort to avoid any stigma the listener might attach to laboratory-made products.
At HI, we seek to better understand entertainment messages, including the moral sentiments that color those messages. Television depicts any number of topics that might incite a viewer’s moral judgment. Word choices matter in this process. When a character on TV talks about “Frankenmeat,” he or she expresses and invites a preferred moral judgement: revulsion towards lab-grown meat. Over several posts, we explore these notions more rigorously to characterize the moral palette of the 2011-12 network television season on ABC, CBS, CW, FOX, and NBC.
In the previous post, we described the overall distribution of moral themes on television. Our analysis relied on Moral Foundations Theory , which proposes that moral systems, across cultures, share six themes: 1. Security (kindness vs. cruelty), 2. Justice (fairness vs. prejudice), 3. Autonomy (freedom vs. coercion), 4. Community (loyalty vs. betrayal), 5. Authority (respect vs. affront), and 6. Purity (innocence vs. corruption). Different cultures and subcultures, at different times, emphasize particular moral foundations over others—for example, liberals describe the moral universe differently than conservatives. To measure how the language of television maps to the actual moral worldview of viewers, we directly compared the overall moral palette of network television to the moral priorities of survey respondents on YourMorals.org .
Visitors to YourMorals.org respond to the Moral Foundations Questionnaire (MFQ), which assesses which foundations people consider morally relevant and the extent to which people agree with various moral judgments . For example, in terms of the Authority foundation, respondents rate the moral relevance of “whether or not someone showed a lack of respect for authority” on a scale from 0 = “not at all relevant” to 5 = “extremely relevant.” Later, again regarding Authority, respondents rate their agreement (0 = “strongly disagree” to 5 = “strongly agree”) with “respect for authority is something all children need to learn.” Respondents  also reported their ideological sympathies in terms of liberal, moderate, conservative, and libertarian. One can glean the moral priorities of each ideological subgroup from the mean (relevance and judgement) rating on each of five moral foundations .
The moral palette of network television, as described in the previous post, derives from percentages of words and phrases that observe or violate each of six moral foundations. To allow for direct comparisons, we summed observances and violations for each of the five foundations shared across data sets (1. Security, 2. Justice, 3. Community, 4. Authority, 5. Purity). We then converted the ratings from YourMorals.org into percentages  to simulate four model televisions shows: model.liberal, model.moderate, model.conservative, and model.libertarian. Figure 1 (below) displays a side by side comparison of the moral palettes used by the actual series on network television in 2011-12 and the four simulated shows. Differences between the green bar and the other model bars indicate the extent to which the actual language of television diverges from the simulated language of these four ideological subgroups.
Figure 1. Comparing moral palettes of network television (2011-12) and simulated shows that reflect the moral priorities of ideological subgroups: percentages of words & phrases that relate to each of five moral foundations.
This graph does not reflect the dramatic ideological polarization that one might expect to see given the current political climate . The moral language of television (the green bar) and the various simulated model shows do not seem far apart from one another.
Geographic metaphors can help us better understand these more subtle distinctions. The differences we observed aren’t dramatic—we are not comparing New York City to Omaha. Rather, the differences we observed are more like measuring the distance between neighborhoods in the same city—say, the Flatiron District (current home of HI) and Morningside Heights (the author’s home). As the crow flies, 4.45 miles separate these Manhattan neighborhoods, but using city-block, or taxicab geometry, the person on the street must travel up and across 93 city blocks (the blue line in Figure 2).
Figure 2. Illustrating the city-block distance metric: a Google Map showing the Euclidean distance (red) and city-block distance (blue) between HI’s offices and the author’s home.
Much as Manhattan’s streets and avenues form a two-dimensional grid of city blocks, the five moral foundations form a five-dimensional grid. The percentages of words and phrases related to each moral foundation can serve as coordinates on that grid. Figure 3. displays the city-block distances between actual television and the various simulated shows—the higher the bar, the greater the difference. In this graph, it is easier to see that the conservative model differs most from the moral language that we observed on the actual television shows.
Figure 3. Measuring differences between moral palettes: city-block distances between network television (2011-12) and simulated shows that reflect the moral priorities of ideological subgroups.
On average during the 2011-12 season, network television appeared to speak equally to the moral priorities of self-identified liberals, moderates, and libertarians in the U.S. Our data show that conservatives may indeed have reason to decry the moral biases of network television . Television did not avoid the language of Community, Authority, and Purity, but conservatives might have preferred an extra splash of those moral colors. We leave you with the question of when, if ever, did entertainment speak with a voice that resonates with conservatives?
Disclaimer and Caveats
Readers should accept our interpretations with caution. In addition to the Disclaimers & Caveats offered in the previous post, the reader should exercise some healthy skepticism about our approach to simulating model shows. The method described in note 5 take some liberties with standard statistical methods; an equivalence between ratings and foundation-related lexical choices may be the wrong simplifying assumption. Nevertheless, as note 6 concludes, equivalence is the most cautious assumption about how moral priorities might translate into foundation-related lexical choices. Interested readers may request technical details about methods and data. We welcome any ideas or suggestions about how to proceed with further research.
We compare the overall moral palette of 2011-12 season of network television to the entertainment of yesteryear.
References & Notes
2. Specifically, 32,263 of the 34,476 respondents who were included in Graham, J., Nosek, B. A., Haidt, J., Iyer, R., Koleva, S., and Ditto, P. H. (2011). Mapping the moral domain. Journal of Personality and Social Psychology, 101(2):366-385. The report is freely available from http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3116962/
5. Specifically, we converted the mean ratings into standard scores, then converted the standard scores into percentages. For example, the mean Security rating for all respondents was 3.42, with a standard deviation of 0.86; for “liberals” the mean rating was 3.62. So, the standard Security score for “liberals” is (3.62 – 3.42) / 0.84 = 0.24. The mean percentage of Security-related words for television was 5.80, with a standard deviation of 1.01; so, the percentage for the simulated model.liberal show is 5.80 + (0.24 * 1.01) = 6.04.
6. This lack of polarization likely has several sources. First, the YourMorals.org data arguably reflects the moral priorities of a wide range of liberals, moderates, conservatives, and libertarians, not just the extremes of each subgroup. Moreover, within that range, we used the mean ratings from each subgroup, not their extreme ratings. Finally, we assumed some degree of equivalence between ratings and foundation-related lexical choices; it might be that every unit of difference in moral rating equals two or more units of difference in moral language.
7. In fact, statistical tests (the Student’s t-test) corroborated what seemed obvious to our eyes. The mean city-block distance between network television shows and the model.conservative show (M = 4.31) is significantly greater than the average distance between network television and the other three models (M = 2.99), t = 5.80, df = 168, p < 0.001. Other tests failed to reach significance: model.liberal (M = 2.88) seems no farther from network television than the average to the remaining models (M = 3.04), t = -0.62, df = 168, p = 0.54; and model.moderate (M = 3.09) seems no farther from network television than model.libertarian (M = 3.09), t = 0.37, df = 168, p = 0.71. Hint: “t” is the magnitude of the test statistic; “df” is the number of degrees of freedom; and “p” meaning the probability of wrongly rejecting the null (no difference) hypothesis, see.
This is the second part of a three post series.
Part One: The moral color of network television
Part Three: Moral resonance across entertainment history
Image: Family watching television, c. 1958 by Evert F. Baumgardner | Via Wikimedia Commons