A story in images: the abortion media storm of 2019
Spring 2019's onslaught of state-level abortion bans across the country was like a surprising cold shower on an even colder day. I found this landslide of legislation disturbing on many levels. How and why is the abortion topic so divisive well after 50 years of what many consider “settled law”? What are the narratives that continue to provoke? What is the religious framing, the secular framing, the feminist framing? Can the visual language around these stories partially answer these questions?
To find out, I used MediaCloud, our awesome media analysis platform here at MIT’s Center for Civic Media. I downloaded headlines and images from this spring’s popular U.S. News stories. I then created interactive maps that sorted the data for me. From these maps, I could clearly see that not only has the term “unborn child” entered conservative mainstream vernacular, but the accompanying image often features a baby or infant. Women who grow and birth these infants have far less visual representation. It looks like images of already-born cute cuddly children are essentially weaponized to malign the choice of abortion.
It seems so obvious to me — a liberal feminist — that a woman’s right to bodily autonomy “trumps” a not-yet-living-and-breathing-developing-life’s “right” to exist and grow in that woman’s body. And certainly I don’t want the government deciding whether and when I should or should not become a mother. But it is not so obvious to others, especially to my southern kin down in Georgia. Especially if cute cuddly babies are discussed more than the pregnant women who suffer from tremendous burdens: illness, abuse, poverty and more.
Are there images that are as effective — and reductive — for the pro-choice movement in favor of abortion accessibility and against coercive pregnancies?
Initial Finding #1: Blastocytes, embryos, babies: conflating stages of development
Apparently there is a term for the visual equating and equivocation of vastly different stages of fetal development: metonymy. A survey of pro-life media leads one to believe that images of a blastocyst, an embryo, a fetus or a baby can be used interchangeably, as metonyms of each other. According to Elena Souris, “because this kind of [emotional, value-based] argument is much better suited for attention-getting rhetoric, the pro-life movement has been much more successful in this modern debate.”
Initial Finding #2: Different stories, different reality
Having reviewed the images over these weeks, it became abundantly clear that left-leaning and right-leaning news sources hardly even cover the same story! For example, analysis over the last week in June showed that while progressive media sources shared a horror story of a woman being arrested for a miscarriage after being shot, conservative sources shared stories about Clarence Thomas, Clint Eastwood, and the Democratic presidential candidates. The difference is fascinating, and also concerning.
Given these initial findings, I can expect that any media analysis would likely determine that conservative/right-leaning sources have more likely paired stories with various images of babies. In comparison, I should also expect progressive media to share images of odious-looking men, coat-hangers and depictions of slavery. What surprises might I find — if any?
Second, what is the effect of these reductive narratives? How can we make any progress in a society when we don’t even talk about the same issues in the same way?
For the past three years, I’ve helped develop a super cool platform for exactly this kind of analysis. Media Cloud is the MIT Center for Civic Media’s suite of tools to analyze the attention an issue gets in the news from week to week, and how effective those stories are at garnering traction. We can even track narrative shifts over a period of time. As of this summer, Media Cloud now pulls a story’s top image (the header image that goes alongside a story) along with the text that is available in front of the firewall (thanks to Leon Yin and Rahul Bhargava for their work on this as well). You can pull these images yourself using our Github tools, or wait until we finish integrating this capability in 2020 to do this type of research using the platform.
Using TopicMapper and MediaCloud to target coverage
My colleague Aashka Dave and I set up a MediaCloud topic “Abortion in the U.S” in TopicMapper, covering the four months of the abortion-related media storm, March 1–July 31, 2019.
My first inspection of the topic showed that most stories were published during the week of May 13–20, topping out at about 17,400 stories.
Kentucky’s governor, Matt Bevin, started the stampede of abortion bans in April 2019, with Ohio and Mississippi following suit. Georgia Governor Brian Kemp made headlines when he signed the ban on May 7 with Alabama Governor Kay Ivey signing another on May 15. These so-called “heartbeat” bills assert that if the not-yet-fully-formed heart cells of the fetus contract in unison creating a detectable heartbeat, then the fetus should be considered alive (as opposed to “viable”) and that this life should be protected under the law. Clearly with 17000+ news stories resulting from this coordinated legislative campaign, there were plenty of narratives to extract.
Progressive vs. Conservative Coverage
Using our MediaCloud API, I downloaded partisanship metadata along with headlines and images in order to visualize differences in coverage between the left and right. As stated earlier, depending on the partisanship of the sources, very different stories were covered and the news was hardly ever the same.
Traditional Press vs. Social Media
Differentiating between publisher inlinking and Facebook sharing is important as well. This is not only because MediaCloud inlink date counts are more accurate, but also because analyzing which media sources are sharing what narratives can track how certain narratives might change over time.
I decided to look at the images a month before the legislative wave (week of April 22–29), two weeks during May (May 13–20 and May 27–June 3), and one month (week of June 24–July 1) and two months (July 29–August 5) after the Georgia decision. Below you can see the jump from 4,770 stories with 1,903 links to 17,418 stories with 11,598 links from April 29 to May 20.
Attention was still high at the end of May, waning through July into August
For each week, I then downloaded the top image (pulled using the Python Newspapers api) from the 30 most inlinked stories and the top 30 images from the most shared stories on Facebook. I then created several “Treemaps” to better showcase the stories’ attention relative to one another. The results were pretty interesting!
Visualizations of Shared Stories and Images
I used a D3 Treemap to relationally size the inlinked stories and stories shared on Facebook so that the story with the highest count is given the largest rectangular region in the treemap. As noted above, upon hovering on the interactive version of these maps, you can see the full image as well as its metadata. You can also follow the link to the original story.
Images are grouped by similarity of the story headline (not visual similarity). The border color around the story indicates whether the source is left or right leaning (dark blue, light blue, gray, light red, dark red respectively).**
** This is according to our analysis of the Twitter landscape during the 2016 Election. You can read more here.
Findings: Week of May 13–20, 2019
Inlinks vs Facebook Share Count:
Because we have noted different behaviors around social media as compared to news media (the more traditional media such as the press), I needed to differentiate between the top count of inlinks (the number of times an online news stories has been cited by the press, or linked to by, other online stories) from the count of Facebook shares for a given story. It is an interesting — but not surprising — observation that the stories most inlinked from digital news feeds are not the stories most shared on Facebook.
Grouping stories together by subject, we easily see that the Alabama ruling topped the media headlines for May 13–20, 2019 with 581 inlinks, followed by stories about Ohio’s ban with 276 inlinks.
If we hover over each of these stories, we can see how many times each story was linked, as well as the title of the story, the publisher (source), and to what partisan category we think it belongs.
The most popular singular story at 239 inlinks is the story about “abortion reversal,” in which different publishers focused on state (Ohio) and national coverage respectively and chose different images than the Facebook image. Neither left nor right leaning media sources seemed to outweigh the other in number of inlinks.
Across the top 30 inlinked stories, none of the images are particularly stimulating nor surprising, the titles themselves are more compelling. The pictures run the gamut displaying government buildings like the Alabama state house, political figures like Kay Ivey, state legislators, celebrities like Lady Gaga, and many protesters.
Viewing MediaCloud’s capture of stories shared on Facebook and associated images for the overall timeframe of the abortion topic, we see a woman and infant, a pregnant woman, an ultrasound, and loads of pictures of men in suits. Unsurprisingly, these images fall into typical left and right wing use of metonymic symbols and metaphors. For example, there are more pictures of baby-related visuals (mother and baby, empty bassinet, newborns, fetuses) from right-leaning news sources and more slavery-related visuals from left-leaning ones (references to Margaret Atwood’s The Handmaid’s Tale, tape over a woman’s mouth, a juxtaposition of all the white men voting for the Alabama abortion ban).
However, the above summary shows us the overall Facebook counts of a story since it was published. What does the coverage look like for just the week of May 13–20, in particular? Does it look different?
Ah, interesting! It seems that in the week of May 13–20, more center and left-leaning Facebook shared stories were inlinked that week than those on the right. The stories associated with the image of the silenced woman, the ultrasound, The Handmaid’s Tale, and the all-male Alabama state legislature were the most shared. Stories about Alabama were shared over 600,000 times and stories about Ohio were shared a third as much.
In comparison, the open date range shows us that more right-leaning sources had greater share counts. For example, New York’s ruling that abortion could occur up to birth — which was published in January 2019—has been shared 1,070,623 times since that date and was linked to 18 times in the week of May 13–20. The specific week seems to show more left-leaning sources being shared via Facebook.
As noted above, because it’s difficult to programmatically scale and position variable images so they visually make sense, here’s a Photoshopped image of the heat map with the corresponding images.
Pretty interesting — and accurate — visual language describing that week, don’t you think?
Findings: Week of May 27–June 3, 2019
For fun, let’s look at the terrain two weeks later, between May 27–June 3. While the attention on the issue had died down a bit, it was still a very active space with 11,842 stories published overall.
Here we definitely see more inlinking from central or left-leaning sources. Coverage is generally about the abortion bans — Kay Ivey, Louisiana, Missouri state governors, Mike Pence.
282,046 Facebook shares top out for one particular story “‘Dark Day’: Illinois House passes abortion bill far worse than New York’s.” While MediaCloud’s 2016 partisan categorization doesn’t recognize this media source, we know that liveactionnews.org is a far-right source. Looks like they were busy! The story was published on May 28, so we can reasonably say that the share count overall is fairly high if we view it along with the New York story published on January 1, five months earlier, with 1,070,623 shares.
Findings: Week of June 24–July 1, 2019
If we take a quick view a month later, at the end of June, we see the attention on the abortion conversation has died down quite a bit — almost by half.
Tidings of Two Tales
For this week especially, it looks like the inlinked sources were primarily left-leaning, and centered on the Alabamian woman who was shot and was then (unbelievably) charged for her fetus’ death. The Facebook shares however are split between the more left-leaning sources discussing the Alabamian woman and more right-leaning sources talking about nothing of the sort, choosing to cover Clint Eastwood, Clarence Thomas, the Democratic presidential candidates, and of course Donald Trump. It’s really remarkable and perhaps a little surprising that the headlines about the unfairly imprisoned Alabamian woman didn’t seem to gain any traction from the right-leaning media.
The May 13–20 Facebook story around the Ohio 11-year-old who was raped and is now pregnant and could be forced to deliver a child due to the abortion bans wasn’t covered by the conservative media, nor was the story about an Alabamian 13-year-old in a similar condition. Rather, right-wing news sources had very little overall coverage on abortion that week through either the press inlinks or through Facebook shares, the most popular being the headline that Vermont made abortion a “fundamental right.”
Comparison of image maps over time
So what do we make of these graphs and the images that we see proliferating online? Well, let’s review the photoshopped maps back-to-back from May through the beginning of July:
Without knowing the context or the headlines, these image clusters display helpful information. We see men in ties, in vitro fetuses or tiny babies, legislators, protest posters. We see tape over a woman’s mouth and large pictures of a black woman. Perhaps we recognize Clint Eastwood, Clarence Thomas, Donald Trump, and Kay Ivey. If we notice the colored borders, we see there are generally more blue or grey borders than red. To someone familiar with the abortion debate, there are few surprises. Perhaps the most interesting fact here is that it’s fairly clear that the left-leaning and right-leaning sources rarely share the same story. In a polarized political climate, this is not surprising — it is as if discussing the same news event would lend too much credibility to the “other” side.
Salacious headlines proliferate rapidly when emotional issues are discussed. Drawing from my own experience, I too find myself sharing stories according to the outrage elicited by the story headline; companion images may additionally spur that sharing. It is wise to remember the headlines and content I regularly see are only part of the overall context of this controversial issue. Not only does my newsfeed fail to offer full comprehension, it creates difficulty when conversing with someone whose understanding of the topic doesn’t even address the same information.
Needless to say, this kind of split-personality coverage does not bode well for people to meet in the middle. Because there is no middle. Or if there is, it is so murky with misinformation, it’s hard for anyone to discern facts from opinion.
Having been raised by a single mother in the South, I am firmly committed to empowering women—and men—to raise their families as best as they can. This means people get to decide when they should become parents — and with whom. When my mother was first married, she couldn’t have a credit card of her own. Needless to say, after the divorce, this made providing for her children a challenge, especially since my lawyer dad used every trick in the book to avoid paying child support.
To understand how the “rights of the unborn child” argument has succeeded in the South is difficult, especially considering that “the war of Northern Aggression” is still a phrase used to describe governmental overreach which lead to the outbreak of the Civil War. For me it is even more difficult if viewed through the context of women’s rights and the myriad risks associated with pregnancy. It is further astonishing that the very people who can give birth are demonized if they choose to abort an embryo for one reason or another.
Southern religious leaders have not always been against abortion. In Randall Balmer’s “The Real Origins of the Religious Right,” he wrote, “In 1968, for instance, a symposium sponsored by the Christian Medical Society and Christianity Today, the flagship magazine of evangelicalism, refused to characterize abortion as sinful, citing ‘individual health, family welfare, and social responsibility’ as justifications for ending a pregnancy.”
Fifty years later, a spate of abortion bans threaten what was once considered a reasonable ruling. It’s easier to talk and share messaging about babies and families than talk sex education, pregnant teenagers, and the fallout from coercive sex. Over time, repetition and redundancy of message, combined with an active and singularly focused religious constituency, dismiss the complex challenges around pregnancy and abortion. This is a deeply rooted social phenomenon and it seems to me that much of this stems from how vocal Southerners feel that women should be in voicing their own needs apart from a family context.
Progressive Strategy Going Forward?
On the other hand, left-leaning sources address a myriad of sub-topics, discussing more of the issues around family planning, pregnancy, and birth—though perhaps less consistently and less insistently. For example, in one month, the stories ranged from how Roe v. Wade is at risk; how a brown woman is being treated horribly; women’s bodily autonomy is at stake; women are underrepresented in state legislatures; Kay Ivey and Susan Collins are traitors to women; etc., etc.
In many cases the wide array of subtopics likely dilutes the messaging and makes it more difficult for any one progressive narrative to gain traction. It seems the left is still struggling to how to best visualize women and their families, who are at the heart of this matter, in a manner that honors and respects women as individuals trying to do their best.
All in all, this was a fascinating foray into the visual language of the abortion debate. While the visual imagery did not evidence much change before and the bans, there was much nuance and it was a widely shared topic. I am keen to continue to observe this topic into 2020.
As we begin to ingest different platforms in MediaCloud like Reddit, Twitter, and 4chan, images containing themes and memes can offer valuable insight into the context of any topic. The 2020 political campaign will be very interesting visually, for sure. We will observing thematic visual changes as well as textual ones.