“Men Appear Twice as Often as Women in News Photos on Facebook”

Onyi Lam, Stefan Wojcik, Adam Hughes, and Brian Broderick write:

A new study of the images accompanying news stories posted publicly on Facebook by prominent American news media outlets finds that men appear twice as often as women do in news images, with a majority of photos showing exclusively men. . . .

Researchers chose to study news images on Facebook because the site standardizes the presentation of news images and text across outlets. News posts that appear in social media feeds like Facebook feature large photographs and contain only a small amount of text and a link to a longer article. In contrast to other formats such as print media, the photograph in a Facebook post occupies more screen space than the accompanying text and is the main object that Facebook users see when they scroll through the news feed. Academic research based on data collected between July 2014 and January 2015 finds that Facebook users only clicked on about 7% of national news, politics and world affairs posts that they viewed in their news feeds. Previous research from the Center has examined how representations of men and women in Google Image Search results can sometimes be at odds with real world data. And academic researchers have leveraged similar tools to study the depiction of women in the news. . . .

The 17 media organizations included in the study were selected according to several criteria. These included: whether they conduct original reporting on general topics, whether they primarily covered national news rather than local news, whether the site was for a news organization based in the U.S., and whether their websites received at least 20 million unique visitors in the third quarter of 2018, according to data from Comscore. The study does not include media outlets that focus their coverage on one topic, such as business, politics, entertainment or sports. The study also excludes local media outlets. The full list of sites included in the study appears in the Methodology. . . .

They also break things down by topic:

I’m surprised that 17% of sports stories were exclusively women. I’d’ve expected this would be lower. I didn’t realize there was this much coverage of women’s sports.

Lots more at the link.

I wonder what Alison Bechdel would say here? I guess just that she’s not surprised.

16 thoughts on ““Men Appear Twice as Often as Women in News Photos on Facebook”

  1. “academic researchers … study the depiction of women in the news”

    Why and to what end?

    Why this among the hundreds of news aspects that could be studied?
    Seems a highly biased mindset as to what is important in news coverage.

    • Let’s just say, I don’t know, tomorrow – Facebook and other media content contained ‘representation’ that was fully reflective of the composition of US society. We still would not have 51 women in the US Senate (as, I feel, we should). Women would also still be paid less than men (even at the same academic rank – so much for the left-leaning bias in academe), take may more years to move from Assistant Professor to Professor, and experience a whole host of other troubling realities.

      Studies like this are underpinned by the idea that insufficient information is the cause of the problem. It isn’t.

      It’s a shame so much energy goes into content issues when we could really do something more useful – from the ground – up to gain parity for women in all levels of our culture.

  2. What makes the Bechdel test great is its simplicity. Not only is it easy to apply as an objective measure of film bias, but it’s incredibly difficult to conceive of any plausible cause other than bias in Hollywood. By far the major determinant of which movies get written, funded, produced, marketed, etc. is the POV of (mostly male) decisionmakers in the entertainment industry.

    What makes this study great is the relative complexity of the mechanisms underlying its measure of bias. There’s a lot more to break down here: bias in news photos seems likely to be a consequence of, yes, editorial bias, but also the genders of individuals who are impacting the world, the genders of those impacted, gender-correlated interest in types of news stories, even whether stories use stock photos. So many inputs! So many arrows! This study seems to set the stage for future researchers to construct complex models with predictions at each stage, which they can then test using huge existing datasets that continue to grow in real time. That’s exciting for what it will end up telling us and, hopefully, as a model for investigating similarly complex systemic bias of all kinds.

  3. I wish I had been asked to guess, before seeing the numbers. Andrew, maybe you can sometimes put the question in the lede, and put the answer below the fold.

    Although I can’t return myself to the state of ignorance I was in five minutes ago, I still think I would have guessed something not terribly far from the right numbers. Thinking of the past few years, on the political side we’ve had a male president and a male presidential candidate, and mostly male governors and U.S senators. Yes, we’ve got Nancy Pelosi, but also Mitch McConnell. I would think a fair proportion of ‘economic’ photos would include politicians, and that’s going to be skewed heavily male. And economics itself is a famously male-dominated field.

    I don’t know whether every news story was defined so as to fit one of their categories, but if so then it’s probably relevant that most foreign leaders are also men. I haven’t been on Facebook in the past year, but in the news coverage I have seen I’ve seen plenty of Angela Merkel photos, but not as many as I’ve seen Boris and Justin and Macron and Putin.

    Sports..I think that, like Andrew, I would have guessed even lower than the observed number. Other than women’s tennis I rarely see coverage of women’s sports unless I look for it, which I do only rarely and pretty much only for bike racing and for a couple of elite athletes I know personally. But I do occasionally see stories about a woman who rowed across the Atlantic or won an ultramarathon or whatever. Still, it feels very rare.

    Of course, not all news photos are of the decision-makers, some are photos of people affected by government policies or economic changes, or of sports fans.

    Putting it all together, the numbers are arresting at first read, but not surprising with even moderate thought. I think they largely illustrate the extent to which we live in a male-dominated society, rather than reflecting bias in the news reporting itself. Most politicians are men, most pro athletes are men, most prominent economists are men (as are most non-prominent ones), most prominent lawyers are men, most prominent doctors are men…it’s not a happy state of affairs, to the extent that I think a large component of the dominance of men in these areas is unjustified and unfair, but it’s about what I would have expected.

  4. I only read part of the article and did not try to go to the original source at all. But the article did raise the issue that I find far more interesting that the title suggests: how does the bias in the news photos compare with the underlying bias in what is being reported? Since men are more common in economics and high level business positions, is the reporting bias representing that bias accurately, exaggerating it, or attempting to mitigate it? Without taking a position on what media reporting should do (which I think is worthy of discussion), just finding out which is the case seems important to me. The fact that photos of men appear more often than women just doesn’t seem so interesting to me. Male national leaders are also in the news more than female national leaders – what should we make of that?

    • I think it would be reasonable (and maybe interesting) also to ask whether reporting is too biased toward major institutions and the people at the top of them. However factual and straightforward the journalism, the picture you’d come away with is hardly a representative swath of society. The perennial focus on CEOs and presidents as point sources of news seems complicit with the gender inequities in our institutions in creating imbalances like the one in the article. It would be hard to do journalism without mentioning them, of course, but if I had seen only half as many photos of Trump and Pelosi and McConnell and Musk over the past several years I’m not sure I’d feel less informed.

  5. Also relevant to our discussion from the other day is this sentence from the “Glossary” at the bottom of the methods:

    “Deep learning is a class of machine learning models that is inspired by how biological nervous systems process information.”

    I admit, I wasn’t really aware of how these models were being described in popular media, but I guess they really are trying to make it seem like these models are “brain-like”.

    The sentence itself is not wrong, since the models are certainly “inspired by” brains. By the same token, airplanes were certainly “inspired by” birds, but ultimately they work quite differently and do different tasks.

  6. In a sentence on cross validated accuracy

    > It achieved 95% accuracy on the set of images that were not used to train the model.

    1. I wish people would stop saying “accuracy” when it comes to probabilistic predictions

    2. I wonder if they’ve propagated the classifier’s cross validated error rate into their confidence intervals for the final inference. I remember the kerfluffle here about Ionaddis’s seroprevalence study not propagated uncertainty in the test false positive rate, and I’m wondering how often studies take “good enough” classifiers as essentially perfect.

Leave a Reply to Michael Nelson Cancel reply

Your email address will not be published. Required fields are marked *