So let's start with a critically flawed, if well meaning, source, to demonstrate why you have to be careful and the type of misinformation that is extremely easy to happen. The human rights campaign is one of the first sources people are going to come up against here. They also horribly mangle the data in ways that are, frankly, shocking. I'm pretty sure that they didn't have anyone qualified to do statistical analysis look over their claims. Maybe not even anyone with basic training.
So the hrc bases a lot of their claims on identified homicide numbers. Which, of course, introduces massive selective bias. Using a pool of 372 identified trans victims of homicide, the hrc claims that 83% of trans victims of homicide were women.
The problem being that identified victims are not the same as all victims. Especially when you are comparing a known highly visible demographic to demographics known for low visibility.
The discrepancy in those claims is subtle to those who are not trained in data analysis, but it is such a basic and major error as to call into question the ability of the HRC to do effective data analysis at all. It's classic survivorship bias, I'm sure you've seen the story of the WW2 planes with bullet holes, it's that sort of thing.
And it is true that all sources agree that trans women (especially black trans women) are far more often identified. We have more names of murdered trans women than other demographics. You might be forgiven thinking this means trans women suffer the most violence.
But competent data analysis paints a different picture.
Lets look at "Global Burden of Violence Against Transgender and Gender-Diverse Adults: A Systematic Review and Meta-Analysis".
The first thing you will notice is that this source spends a huge amount of time going into exact methodology, what data was used, the source, what data was not used and why, and how and why it was analyzed in the way it was. This is what competent data analysis looks like. Lets pull some critical quotes from the source:
"Disproportionate violence burdens are experienced by transgender and gender-diverse people who are multiply marginalized.13 For example, transphobia, racism, and classism intersect to compound violence exposure for transgender women of color.14,15 According to the Trans Murder Monitoring project, 94% of the 321 identified transgender and gender-diverse people who were murdered in 2023 were transgender women or transgender feminine people, and the majority were Black, women of color, and engaged in sex work.16 Such disparities suggest the importance of considering intersecting identities in violence research and intervention strategies.13"
You will notice they acknowledge the data on identified victims is almost entirely trans women. What they do not do is suggest that this means trans women are the primary victims of anti trans violence, instead drawing the conclusion that intersecting identities are a key factor. This is important because of the results of the actual data of properly conducted data gathering. Here is their conclusion about violence across trans gender identities:
"Our meta-analyses did not show significantly different levels of violence globally among transgender women, transgender men, and nonbinary adults."
"While we would have included homicide, we identified no studies that reported its prevalence."
The first part is pretty plain. There is not a statistically significant difference in violence levels between trans identities, according to the data that was collected in a scientific manner.
However, homicide data has been completely neglected to be collected in a scientific manner (likely due to the difficulty of doing so). Instead reports are anecdotal - that is, we identify trans homicide victims after the fact, which means it is entirely based on demographic visibility. The source does suggest that it seems likely that black trans women in particular are disproportional victims of homicide, but no real conclusion can be drawn without further studies.
Given that general violence against trans people seems fairly flat across identity, and that intersectional factors seem to play a critical part in violence against trans people, I would suggest that anecdotally gathered evidence is the source of the discrepancy. Between a demographic known for hyper visibility (trans women), and demographics known for being practically invisible (other trans identities), it draws a very obvious picture of why the anecdotal evidence looks the way it does.
As a side note and purely my own opinion, I think there is a tendency for white trans women to claim the violence experienced by trans women of color, especially black trans women, and report it as general violence against trans women. But the data suggests that race (and class, which are often intertwined) play such a massive role in violence against trans women that I feel it is honestly kind of racist how the trans women community so often glosses over it, if it is even discussed at all. The simple fact is that a white trans woman like myself is much, much less likely to experience violence than a black trans woman, and that is a key part of the discussion.
Getting back to data, there is one more important piece I want to highlight. Specifically, how data has been gathered, and on who.
"Of the 94 studies, nearly all (90 studies [96%]) included transgender women, whereas 49% (46 studies) included transgender men, and 37% (35 studies) included nonbinary adults"
This is a massive discrepancy in visibility. It seems almost laughable that a study would include trans women specifically but not bother with trans men, a staggering display of systematic bias against trans men, and yet it appears that half of available studies that passed the minimum bar of quality to be included in the meta analysis did just that. And nonbinary people have it even worse. It is no wonder that violence against trans demographics other than trans women is less well understood and under reported in anecdotal analysis. Even the people actively looking for violence against trans people are heavily biasing their studies towards discovering violence against trans women.
I could go on for a long, long time breaking down this one meta analysis, but I think you get the point, and in particular how important proper data analysis is. The simple fact of the matter is many "reputable" sources are clearly not conducting proper data analysis when breaking down anti trans violence by demographic and making wildly incorrect claims because of it.
I have one more source I will be brief with. "Gender Identity Disparities in Criminal Victimization: National Crime Victimization Survey, 2017–2018". The critical quote for our purpose:
"Transgender women and men had higher rates of violent victimization (86.1 and 107.5 per 1000 persons, respectively) than did cisgender women (23.7 per 1000 persons; OR = 3.88; 90% CI = 0, 8.55) and cisgender men (19.8 per 1000 persons; OR = 5.98, 90% CI = 2.09, 9.87), but there were no differences between transgender men and women (Δ = 21.4; SE = 68.7; P = .76)."
You will note that the raw data shows higher violence against trans men, but with proper data analysis it is revealed to not be statistically significant. This is something you have to be careful of, there are tools experts use for data analysis that reveal that obvious conclusions from raw numbers are not always correct.
There are more sources out there, but I am not going to bother tracking them down. It was not hard to find these sources, these two major sources I found through simple internet searches on the first page (duck duck go, in my case). I don't know why you had problems, for me it was as simple as searching for something like "trans violence by gender identity" and looking for actual studies that talk about methodology.
This is a skill people really need to develop, but its not something I know how to teach. I personally think the queer community on tumblr is very over reliant on being spoon fed sources, and on top of that is really terrible at evaluating sources. I think it is genuinely quite likely that most trans people go straight to the HRC page, see their big bold graphics proclaiming that 83% of trans homicide victims are trans women, and don't even consider that it could be bad information.