I took your data table and added a couple of columns to analyse the data by population percentage so that it is more representative of recorded prosecutions/ conviction rates per head of population.
There seem to be very significant differences between the likelihood of rape prosecutions/ convictions in different ethnic groups, which cannot be accounted for by “police/ jury prejudice” or suchlike given that some of the categories with the lowest recorded offending rates are not white/ European (e.g. Chinese, Arab).
Obviously there may also be differences in reporting levels within some ethnic groups as well, which makes it harder to analyse what is happening. But it’s probable that this factor would actually make the gap between actual offending rates and recorded prosecutions higher, not lower, given that the women in the ethnic groups with higher recorded prosecution/ conviction rates generally also tend to also be those where the women are more likely to be subject to coercive or physical or financial abuse and therefore less likely to feel able to make reports to the police at all.
What is disturbing is that nearly 30% of prosecutions and convictions were in categories with this data not recorded or specified at all, or unclear categorisations. It’s unclear why the data is being recorded in some cases and not others. We cannot make appropriate evidence-based public policy if authorities like the police and justice system do not keep accurate records. It’s unacceptable.