So here's one of the "evaluations" linked to off the front page of the KC website [1].
It lists no author, date or other bibliographic information. It has no page numbers, making referencing difficult. It shows no evidence of being reviewed. It is, to judge from internal evidence, the unreviewed, unpublished research data from someone's PhD.
Parts of it defy belief. It claims to be comparing KC clients (interestingly, aged 16-24) to a "control group" matched for age and socio-economic status. It makes the dramatic claim that compared to this control, KC's clients "are 13 times more likely than control participants to have experienced severe to extreme levels of sexual abuse". But turning to Fig.2, we see the graph showing 0% sexual abuse in the control, 13% in the KC cohort. Given n=108, 0% means 0 cases.
First, zero times thirteen is not thirteen (and other figures in the graph fail to match the summary, too). Let's assume the figure is correct and the summary is wrong, however.
But secondly, how seriously are we to take a claim that a group of 16-24 year olds, matched for socio-economic status with KC clients, does not contain a single case of sexual abuse? And indeed, that only 2% (again, matched with KC's cohort) have experienced physical neglect? Those numbers don't pass a basic sniff test: they seem extraordinary compared to those quoted in Table 1 of [2].
Then it goes on to make detailed psychiatric claims about clients. 35% depression, 35% PTSD, 41% "dissociation" - the latter dramatically undefined, which given the high rate of dissociative experiences in the normal population seems a little odd. Again, this is unpublished preliminary work from someone's PhD, with no explanation as to who's doing the diagnosis.
And ladies and gentlemen, the coup de grace: While approximately 0.02% of the general population are thought to have a learning disability, this number is far higher amongst Kids Company clients, where 28% have been identified as living with a learning disability. Yep, we can deploy a new scepticism about SPLDs, as Kids Company claim that learning disabilities present in only 2 per ten thousand of the population, so the typical school will have at any one time a grand total of zero children with SPLD. This 0.02% claim isn't referenced (which is odd, given the welter of citations used elsewhere).
Mencap [3] reckons roughly 2%, one hundred times the rate the evaluation claims. Surely this evaluation can't have got confused between percentages and rates per hundred? 2% is 2 per hundred, not two per ten thousand.
[1] kidsco.org.uk/download/clients.pdf
[2] www.nspcc.org.uk/globalassets/documents/research-reports/child-abuse-neglect-uk-today-research-report.pdf
[3] www.mencap.org.uk/about-learning-disability/information-professionals/more-about-learning-disability