About that twitter misogyny research (and bad science reporting)

CONTENT NOTE: research of rape “threats” and threatening language.

I first saw the story today claiming that “50% of misogynistic tweets are made by women” on the BBC News website. I noted other sources being cited on Twitter later, with various criticisms; but this is for the thoughts that immediately occurred to me without reference to anyone else.

(You’d be surprised how little social media networking I do while at the day job!)

My first thought was that maybe this is bad science reporting, rather than bad science. Lord knows, it happens often enough that maybe any intelligence and nuance has been scrubbed away by editors anxious for teh clicks.

A quick click through to the PDF file reveals there is indeed somewhat more nuance than the story suggests. The authors did not simply, “count the number of uses of two particular words as indicators of misogyny”, but rather:

We subjected each data set to a number of analyses, using both qualitative and quantitative methods:
1) Volume over time
2) Different types of use
3) Who is using these words?
4) Case study: what drives traffic?

To conduct the analysis we conducted both automated analyses using a technique called natural language processing; and qualitative analysis where a researcher carefully reviewed random samples of the data.

My second thought was that there are at least two ways in which women might be using the words “slut” or “whore” in ways that are not themselves misogynistic. The first I thought of was “claiming the name”, and self-referring (perhaps in a positive way) as either slut or whore – perhaps as sex workers, or “kinky” types, or just celebrating their own sexuality. For example, the “Slutwalk” marches that started a few years ago. The second I thought of was reporting on men’s misogyny towards them, for example, “I didn’t answer when the guy told I looked sexy, so he called me a…” with a hashtag about street harassment.

The way the research was reported made it seem as though these sorts of responses were lumped in with genuine “you tried to steal my boyfriend, I hate you, you slut!” (or whatever – some of my ideas of women’s insulting twitter exchanges may be based on half-remembered secondary school overheard conversations!).

The actual paper looks like just maybe both of these were screened out in some way!

Here are the findings that the paper reports:

  • Between 9 January and 4 February 2014 there were around 131,000 cases of ‘slut’ and ‘whore’ used in English from UK-based Twitter accounts. We estimate that approximately 18 per cent of them appears misogynistic.
  • There was a high proportion of ‘casual’ misogyny. Approximately 29 per cent of the ‘rape’ tweets appeared to use the term in a casual or metaphorical way; while approximately 35 per cent of the ‘slut’ and ‘whore’ tweets appeared to use the term in a casual or metaphorical way.

(You’ll notice that there was also analysis of the use of “rape”; the headline figure is that around 12% of 100 thousand uses seemed to be threatening)

The researchers then, “split the data into ‘comment’ (tweets which were about the use of word itself) and ‘conversation’ (tweets which included the word as part of a conversation).”

So my question regarding reporting versus using seems to have been answered. The researchers go on to record, “We found 7,993 tweets that were commenting on usage of these words, 108,409 that were actual conversational usage.” A quick bit of approximate doing the sums in my head, that’s only around 1 in 14 or 15 that was discussing usage versus conversation.

In their analysis of usage, they broke down use into “Serious/non-offensive”, “colloquial/casual”, “Generally misogynistic”, “Abusive”, and “Other (inc. subversive and porn)” and used a sample of 500 manually assessed tweets to estimate proportions (with the headline finding of 18% misogynistic; also 20% classed as “abusive”). [If you click through to the paper, be aware – the example of “abusive” usage the authors chose also includes threats of violence]

I do have one issue with the paper. In the “Key findings”, the authors write,

Women are as almost as likely as men to use the terms ‘slut’ and ‘whore’ on Twitter. Not only are women using these words, they are directing them at each other, both casually and offensively; women are increasingly more inclined to engage in discourses using the same language that has been, and continues to be, used as derogatory against them.

This does not seem to be supported by the analysis as presented. It may be true (certainly, given the graph of usage by gender shows roughly equal usage, which means a sizeable proportion of women’s tweets must have been “conversation” as opposed to “comment”) but it hasn’t been demonstrated. To demonstrate it, I would need to see a breakdown by gender of the types of usage table, which isn’t provided.

So, once again, bad science journalism trumps a fairly reasonably conducted piece of research. The research itself does have issues, not all of which occurred to me and not all of which I have mentioned here. The authors themselves admit: “To give a rough and ready illustration, we ran a series of short studies in order to better understand the volume, degree and type of misogynistic language used on Twitter.” (emphasis mine).


About ValeryNorth

I overthink everything.
This entry was posted in Gender, Language and tagged , , , , . Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s