Spent a little while trying to track down a source report for the Amnesty conversion therapy claims
"We mapped 12 groups promoting the harmful and abusive practice of so-called 'conversion therapy,' aiming to ‘change’ or suppress people’s sexual orientation or gender identity. The spending of 4 of these surged by nearly 165%, marking the fastest growth of all categories tracked."
Was not able to find anything - anyone else found the report?
I did find some Amnesty published documents, for example
Intersectionality from a racial justice perspective: Submission to the UN Special Rapporteur on Contemporary Forms of Racism, Racial Discrimination, Xenophobia and Related Intolerance - Amnesty International
Was a bit shocked to find that about 95% of the references in this paper were for to other Amnesty documents - all a bit circular.
There seemed to be a need to use numbers to backup claims,
"In the realm of technology-facilitated gender-based violence (TfGBV), Amnesty International found that racialized women (Black, Asian, Latin American and mixed-race women) in UK and USA were 34% more likely than white women to be mentioned in abusive content"
but on closer inspection the numbers do not seem to be based in scientific method - e.g. the 34% above is based on anecdotal evidence
Amnesty International, “Crowdsourced Twitter study reveals shocking scale of online abuse against women”, 18 December 2018
Crowdsourced Twitter study reveals shocking scale of online abuse against women - Amnesty International
The study appears to use crowdsourcing to inspect tweets, pick out the ones with women, divide them up by race and then use AI to rate them as abusive or non-abusive.
There might be a further step where the AI scales the results of the small tweet sampling to reflect the total number of tweets posted in the UK and USA.
The exact definitions of abuse, woman and methodology are clearly very important in interpreting the study's results.
At the end of the report, there is a useful link - Click For full methodology.
This link appears broken