Update: What the Florida SYG numbers seem to be telling me [View all]

[Bazinga]

Last edited Fri Jul 26, 2013, 09:37 AM - Edit history (2)

I spent some time on the Tampa Bay Times site and ran the numbers on every combination I could conceive. I did t-tests for two proportions using this site comparing acquittal rates for every group:

White defendant
Minority defendant
White victim
Minority victim
White on white
White on minority
Minority on white
Minority on minority
Inter-racial vs intra-racial

I compared every single option that gave a number of acquittals and a number of cases to every single other option (though I have only finished the fatal cases so far). Only two comparisons, yep two out of over 100, achieved statistical significance. They were the following:

Acquittal rate if victim is white: 56.5% (69 cases, 39 justified)
if victim is black: 78.1% (32 cases, 25 justified)
p-value: 0.0246

Acquittal rate for white on white: 58.8% (51 cases, 30 justified)
for white on minority: 83.3% (12 cases, 10 justified)
p-value: 0.0694

You'll note that these comparisons are very similar in that only the victim's race appears to be important. Interestingly the comparison of whites killing blacks vs blacks killing blacks was as follows:

Acquittal rate for white on black: 85.7% (7 cases, 6 justified)
Acquittal rate for black on black: 65.4% (26 case, 17 justified)
p-value: 0.2456

Similarly if we compare white defendants to black defendants we get:

Acquittal rate for white defendants: 61.5% (65 cases, 40 justified)
Acquittal rate for black defendants: 68.6% (35 cases, 24 justified)
p-value: 0.4816

My conclusion (and I reserve the right to be convinced otherwise) is that Florida's SYG law is applied equally to defendants of all races, however, there is a disparity in cases where the victim is a minority. I can't explain that disparity, but I have a hard time believing that it is racism because the the judicial system deals with defendants, not victims, and there appears to be no racial favoritism in the case of defendants.

I would very much like to hear your opinions. Also, if there are other comparisons that you would like to see, let me know and I will do my best to run the numbers. The Tampa Bay Times has a tally of how many cases are initiated by the victim (as opposed to the defendant), but unfortunately it does not break down how many of those case were deemed justified. I think it would be very telling if the acquittal rate is higher when the victim initiates vs when the defendant initiates, as this is particularly a propos to the Zimmerman trial.

ps- I also reserve the right to make mistakes with the numbers. I'm a bit of a statistics amateur, but I've done my best to avoid any gross errors.

Update with ALL cases:

The patterns I highlighted above remained true. In fact, the majority of comparisons became less significant. Most apparent was the comparison between white victims and black victims. When all cases are considered, this comparison becomes:

Acquittal rate if victim is white: 62.8% (129 cases, 81 justified)
Acquittal rate if victim is black: 75% (64 cases, 48 justified)
p-value: 0.0782

Notice that, while still significant, the confidence level has dropped from 95% confidence to 90%. Also, the acquittal rates are only 12.2 points different instead of 21.6. Perhaps that is why the Tampa Bay times article chose to use the fatal case numbers for their article, it is obviously the most inflammatory.

There were a few comparisons that became more significant, but none that reached significance that weren't already significant. Two, however, caught my eye because they improved and came pretty close to 90% significance. They are the following:

Acquittal rate if victim is white: 62.8% (129 cases, 81 justified)
Acquittal rate if victim is hispanic: 83.3% (12 cases, 10 justified)
p-value: 0.1048

Acquittal rate if case is Intra-Racial: 65.5% (148 cases, 97 justified)
Acquittal rate if case is Inter-Racial: 76.9% (52 cases, 40 justified)
p-value: 0.1082

The first statistic reaffirms the disparity between cases with white vs minority victims, but I'm not quite sure what to make of the second. It is likely confounded by the fact that the acquittal rate for white victims is much lower and the majority of white victim cases involve a white defendant. I don't think there's a racial argument to be made with this specific statistic, and it doesn't meet my definition of significant, but it is interesting.

Because rdharma will want to know, the p-value for the comparison of white on black cases vs black on white cases did increase from 0.4392 to 0.4206.

note- In all of these numbers I ignored pending cases (for obvious reasons). And always remember there are liars, there are dirty rotten liars, and then there are statisticians.