1/ UCU free speech survey🧵. It is not good practice to invite people to take part in a survey by telling them what their views should be. One likely effect is to dissuade those who disagree with you from participating, thus biasing the results.
2/ Questions are highly disclosive, taken together they would give a high risk of identifiability. Although questions are optional, this isn't made clear. Given that the ruling faction in the union is hostile to some members, leading to low trust, these considerations are
3/ important. Some reassurance should have been provided that the data and analysis would be handled independently, both to ensure ethical treatment of the data and impartial data analysis.
4/ "The next questions are about your identity". This will obviously antagonise some of the respondents whose academic freedom has been most threatened in recent years.
5/ If you want to know people's response to a statement like this, attributing the statement to a source (e.g. UK govt) will bias how people respond.
6/ Not sure how meaningful this question is.
7/ How did they come up with this list of options? Doesn't include the key threats in my view, and mentioning IHRA but not biological sex and women's rights seems perverse.
8/ List of options continued...
9/ We are generously allowed to make some other suggestions.
10/ The next part is quite repetitive
11/ No "don't know" option, seriously? How many people do you think have seriously engaged with this legislation?
12/ I wonder how they'll analyse this?
13/ Finally, the most important question.
14/ This is a survey designed for propaganda purposes. I hope that people will complete it in order to undermine this. It is unbecoming for an academic union to do bogus "research".
• • •
Missing some Tweet in this thread? You can try to
force a refresh
"The Office for National Statistics “hugely overestimated” the number of transgender people in the UK, Whitehall sources have claimed, as the body admitted it could have carried out “additional probing” before releasing the controversial data"
"Michael Biggs... claimed that confusion over the question may explain why the London boroughs of Newham and Brent, which have a significant percentage of residents who speak English as a second language, recorded the highest proportion of transgender people in the UK."
"A Whitehall source suggested the ONS executive, led by Prof Sir Ian Diamond, may have lost its “credibility” to accurately record sex and gender, based on its handling of the trans issue together with its separate loss of a legal challenge over the wording of the 2021 census."
1/ It is appropriate that @StatsRegulation are looking into the 2021 Census.
According to the Code of Practice for official statistics, "Quality means that statistics fit their intended uses, are based on appropriate data and methods, and are not materially misleading."
2/ Michael Biggs @SociologyOxford has shown that the responses generated by the gender identity question are materially misleading.
3/ This is not surprising given that the question asked, 'Is the gender you identify with the same as your sex registered at birth?’ is not in plain language.
This fails questionnaire design 101. Questions must be clear and comprehensible to all.
When a he/him Professor with no expertise in social statistics makes an assertion about the likely direction of error in a statistic, perhaps I should just ignore it.
But misinformation and attempts to whip up hate are indeed dangerous, so I will respond.
Prof Walters implies that the Trans Murder Monitoring Project figures must be an underestimate.
On what basis?
The project is funded to record all homicides of people who may fall under the trans umbrella. Activists are highly motivated to report any possible incidence.
In fact, the Trans Murder Monitoring Project has over-stated the numbers in the past, corrected by @STILLTish (who sadly is not funded by the EU to do this work).
Analysis of the geographical concentrations of trans identity throws up some odd results. osf.io/preprints/soca…
2/ "The most likely explanation is that the question on gender was misinterpreted by a non-trivial number of immigrants whose main language was not English."
This highlights an important principle of questionnaire design.
Questions need to make sense to the whole population
3/ It's particularly important if you are trying to gauge the size of a small minority group that the majority who are not in that group understand what is being asked.
Otherwise a small proportion of false positives from confused people may represent a substantial % ...
2/ It is hard to overstate the importance of Census data: it provides a snapshot of the entire population; it furnishes the benchmark against which we judge whether other data sources are representative.
And it is vital for equalities monitoring.
3/ If an organisation wants to know how representative it is, Census data is vital.
E.g. We can compare the proportion of people who say they are LGB at the BBC (8 per cent) to the proportion in the 2021 Census (3 per cent).
Michael Biggs, a professor of sociology at Oxford, said the university “is supposed to be governed democratically by its academics, and yet here is an outside lobby group ordering the administration to change its policies without the normal scrutiny”. thetimes.co.uk/article/212a31…
"The university had tried to keep its correspondence with the lobby group secret"...
"However, it was ordered to release the documents after the information commissioner ruled that participation in the Workplace Equality Index allowed Stonewall to “exercise a significant degree of influence” over the policies of public organisations."