Welcome to European Tribune. It's gone a bit quiet around here these days, but it's still going.
Display:
You're right - I overlooked that. But the proof of bias is there... whether you use a weighted average or vanilla average.
by vladimir on Mon Mar 16th, 2009 at 03:57:19 AM EST
[ Parent ]
You call a 1:1 correspondence in indictments a proof of bias? A 10:9 correspondence in convictions? There is a possible bias in favour of Albanians, but that does not mean that there is a bias against Serbs.

- Jake

Friends come and go. Enemies accumulate.

by JakeS (JangoSierra 'at' gmail 'dot' com) on Mon Mar 16th, 2009 at 05:20:28 AM EST
[ Parent ]
Then we're not comparing the same thing. You said that the group's average should be weighted... and indeed that can be an approach - which would change the estimated factor of bias - but it wouldn't change the t test results - which confirm that there is a statistically significant difference between the two groups being compared.

By weighing the group's average, you don't change any individual ethnic group's results. If you look at these results (in the last graph provided in the diary), you'll see that in the Bosnian war, it appears that all ethnic groups are treated equally. However, these figures work on the wrong assumption that there was no killing of Croats by Muslims and vice-versa...

Compared to Muslims & Croats in Bosnia we have:
> a bias of 100:135 against Serbs in Croatia
> a bias of 100:179 against Serbs in Kosovo
> a bias of 22:100 in favour of Albanians
> a bias of 0:100 in favour of Croatian Croats

by vladimir on Mon Mar 16th, 2009 at 05:44:01 AM EST
[ Parent ]
No, I'm saying that all your averages and standard deviations should be weighted by the denominators that went into your ratios.

And your comparisons of conviction rates are pure, mean-spirited garbage, because there are still cases outstanding which will in all probability change most of those figures significantly. Compare indictments, or break down the indictments into convictions, acquittals and outstanding cases, if you like to. But using only convictions is nonsense as long as there are cases outstanding.

- Jake

Friends come and go. Enemies accumulate.

by JakeS (JangoSierra 'at' gmail 'dot' com) on Mon Mar 16th, 2009 at 05:55:23 AM EST
[ Parent ]
Bull shit Jake. That's something you could eventually argue had the tribunal been set up last year. But after 16 years of the tribunal's work, taking the convictions is perfectly legitimate. Look at Seselj, he's been in jail for 6 years, with no exit date in view... and is without a conviction and even without a trial (it's been indefinitely adjourned)! Where would you put him?

It's perfectly plausible to say that the outstanding cases will follow the same pattern as those already finished. Unless you have some inside information that there is an army of Croats and Albanians in the dock waiting to be convicted.

by vladimir on Mon Mar 16th, 2009 at 06:14:52 AM EST
[ Parent ]
vladimir:
Unless you have some inside information that there is an army of Croats and Albanians in the dock waiting to be convicted.
Bull shit vladimir, by your own data there are 6 Croats whose trial is "ongoing" (and 2 have been transferred to national courts) vs. 9 (+8) Serbs. So there are proportionally more Croatian cases outstanding than Serb.

Regarding Kosovo, it is a well-known fact that when the number of people in a category drops below about 5, statistical tests become insufficiently powerful. In the case of Kosovo the expected numbers are small enough you can't really draw any conclusions.

Most economists teach a theoretical framework that has been shown to be fundamentally useless. -- James K. Galbraith

by Migeru (migeru at eurotrib dot com) on Mon Mar 16th, 2009 at 06:23:09 AM EST
[ Parent ]
Bull shit migeru ;)

That's a ratio of 8 Croats : 17 Serbs.
Croatian population= 4,5 M
Serbian population= 8 M (not counting Kosovo Albanians)

by vladimir on Mon Mar 16th, 2009 at 06:44:52 AM EST
[ Parent ]
it looks even
by vladimir on Mon Mar 16th, 2009 at 06:45:34 AM EST
[ Parent ]
You don't compare to the population, you compare to the number of indictments in which the ratio is closer to 3:1 than to 2:1.

You are beginning to appear disingenuous. You keep moving the goalposts.

Most economists teach a theoretical framework that has been shown to be fundamentally useless. -- James K. Galbraith

by Migeru (migeru at eurotrib dot com) on Mon Mar 16th, 2009 at 06:47:59 AM EST
[ Parent ]
My fault.
by vladimir on Mon Mar 16th, 2009 at 06:54:08 AM EST
[ Parent ]
But even with 3:1... we're in the same ball park. It's not going to change your end result by very much.
by vladimir on Mon Mar 16th, 2009 at 06:55:16 AM EST
[ Parent ]
Look, I'll be the first to argue that keeping people locked up for decades without trial is a monstrous travesty of justice. I'd argue that Adolf fucking Hitler should be released if you couldn't convict him after five years. But there are a couple of points here that you fail to account for:

  1. Not all the cases started at the same time, because some of the indictees were better at evading capture than others.

  2. You can reasonably expect (at least to first order) that the ratio between convictions and acquittals hold up - so you could use the figure [convictions*(1 + ongoing/acquittals)] if you wanted to. But using just convictions remains nonsense.

- Jake

Friends come and go. Enemies accumulate.
by JakeS (JangoSierra 'at' gmail 'dot' com) on Mon Mar 16th, 2009 at 06:26:19 AM EST
[ Parent ]
You know, you really shouldn't be making even an implicit comparison between Seselj and Hitler.
by vladimir on Mon Mar 16th, 2009 at 01:25:08 PM EST
[ Parent ]
Actually, even better than convicted, we could use the severity of the conviction (number of years sentenced to prison). But that's also something that would take time to compile. Really, all the other indicators can be politically charged - including for example, convicted and then released 1 year later.

The cases outstanding only favour the Croats and this by a very small margin. So I'd still go with the convicted indicator.

by vladimir on Mon Mar 16th, 2009 at 01:41:27 PM EST
[ Parent ]
Remember that the more variables you throw into your model the least significant are the results.

Most economists teach a theoretical framework that has been shown to be fundamentally useless. -- James K. Galbraith
by Migeru (migeru at eurotrib dot com) on Mon Mar 16th, 2009 at 01:47:18 PM EST
[ Parent ]
vladimir:
The cases outstanding only favour the Croats and this by a very small margin. So I'd still go with the convicted indicator.
Are you cherry-picking your indicators?

Most economists teach a theoretical framework that has been shown to be fundamentally useless. -- James K. Galbraith
by Migeru (migeru at eurotrib dot com) on Mon Mar 16th, 2009 at 01:48:52 PM EST
[ Parent ]
Why are you insinuating that? I'm just reacting to Jake's comment about me being mean spirited (which I really didn't appreciate) and explaining why I think the convicted indicator is best.
by vladimir on Mon Mar 16th, 2009 at 01:51:23 PM EST
[ Parent ]
My being mean spirited because I chose the 'convicted' indicator and not the 'acquitted indicator' which in his opinion was much better.
by vladimir on Mon Mar 16th, 2009 at 01:52:40 PM EST
[ Parent ]
You misunderstand the use of "mean-spirited". Focus on this instead
Compare indictments, or break down the indictments into convictions, acquittals and outstanding cases, if you like to. But using only convictions is nonsense as long as there are cases outstanding.


Most economists teach a theoretical framework that has been shown to be fundamentally useless. -- James K. Galbraith
by Migeru (migeru at eurotrib dot com) on Mon Mar 16th, 2009 at 01:55:43 PM EST
[ Parent ]
First, it's the comparison that I call mean spirited, not you - apologies if that wasn't clear.

Second, I'm not arguing that you should use acquittals - I'm arguing that you should use [convictions*(1 + ongoing/acquittals)], which, unlike convictions, would make sense... kind of.

- Jake

Friends come and go. Enemies accumulate.

by JakeS (JangoSierra 'at' gmail 'dot' com) on Mon Mar 16th, 2009 at 02:00:40 PM EST
[ Parent ]

Display:

Top Diaries

Impeachment gets real

by ARGeezer - Jan 17
24 comments

A Final Warning

by Oui - Jan 10
112 comments

Environment Anarchists

by Oui - Jan 13
4 comments

More Spanish repression

by IdiotSavant - Jan 6
8 comments

Occasional Series