How many women consider themselves to be a feminist?
Feminist self-identification was the subject of a recent BBC article by Dr Christina Scharff (King’s College London).
On social media, Kate Andrews (Institute for Economic Affairs) said that 90% of British women don’t consider themselves to be feminists. In response, the barrister Jessica Simor QC said that figure was actually 66%.
Which is correct: do one in ten or one in three British women consider themselves to be feminist?
10% or 34%?
When a Conservative leadership candidate was questioned about feminism, Kate Andrews (Institute for Economic Affairs) wrote that:
The *90% of women* who don’t identify as feminist won’t have viewed it that way.
The barrister Jessica Simor QC replied that the figure was actually 66%, citing Dr Scharff’s BBC article.
Ms Andrews then stated:
No, I meant to say 90% (I think it’s just over 90% to be exact). What [Jessica Simor QC] has done is crop out the lead story in a BBC article about 2019 data and only share a poll quoted further down in the article that is from 2018. Very poor form.
The contention is whether 10% or 34% of British women regard themselves as feminists.
The Survation internet panel poll of women was conducted in December 2015 on behalf of the Fawcett Society (not “2019 data”). This is being compared against the female sub-sample from a YouGov internet panel poll conducted in February 2018. Nor is there a need to be “exact” here: surveys provide estimates.
However, the main difference is not timing, but the response options.
In the Survation poll, five substantive options are given (“which of the following statements best describes your view?”):
I describe myself as a feminist
I believe in equality for women and men but I don’t describe myself as a feminist
I feel excluded by feminism
I think feminism is irrelevant
I don’t know what feminism stands for
9% in the sample clicked “I describe myself as a feminist” — roughly 1 in 10. The most popular option was “I believe in equality for women and men but I don’t describe myself as a feminist” (65%). 10% clicked “none of the above”.
A Binary Choice
The YouGov question offers two attitudinal options (“do you consider yourself to be a feminist?”), with the option to be unsure:
Yes, I do
No, I do not
In the female sub-sample, 34% clicked ‘Yes, I do’. Sub-samples are not ‘internally weighted’, and we should somewhat cautious about the central estimate. (The BBC article compares to another YouGov poll in October 2013, with the more direct variant: “Are you a feminist?”. This wording difference is not noted in the article.)
When a binary choice offered, those self-describing as ‘feminist’ increases dramatically. This is plausibly due to the limited options and acquiescence bias. In Survation’s question, the response options allowed women to express a belief in gender equality, whilst rejecting the term ‘feminist’.
In High Definition
In February 2018, YouGov split their samples and asked a binary question of feminist self-identification in three different ways: direct self-reporting, providing a definition plus asking if they were a feminist, and agreement with a definition.
The definition given was:
One definition of a feminist is someone who thinks men and women should have equal rights and status in society, and be treated equally in every way. Are you a feminist?
Across seven European countries, there was a rise in self-reporting when a definition is given, and agreement with that definition is even higher.
As Tanya Abraham of YouGov notes, there appears to be a serious image problem with the word ‘feminism’ — whilst people generally subscribe to gender equality.
Both figures used by Kate Andrews and Jessica Simor QC are from polls conducted by British Polling Council members.
The response options differed — with Survation’s five options giving a lower estimate than YouGov’s binary choice. When citing survey data, people need to be clear on when the survey was conducted, what the question was, and what options people were offered.
Different response options can lead to different estimates.