死刑应该取消么。—————— 婴儿惨案有感。

Dracula

There was similar discussion 2 years ago.

5 N$ T4 h8 p3 p' m( Uhttp://www.aswetalk.org/bbs/thread-1852-1-2.html

This is my comment then.

http://www.aswetalk.org/bbs/foru ... amp;page=1#pid42135

Dracula

code_abc 发表于 2013-3-6 12:52
我总觉得取消死刑是一个很奇葩的想法。
* F4 D; M5 `" q5 y从正义的角度看，正义不但要得到声张，还必须被公众看到（借用《间 ...

There is a large literature in economics studying the relationship between homicide rate and death penalty. The result is mostly inconclusive. There is a excellent examination of these evidence in Donohue and Wolfers (2005) “Uses and Abuses of Empirical Evidence in the Death Penalty Debate.”, Stanford Law Review. This is a link.

http://www.stat.columbia.edu/~ge ... L.%20Rev.%20791.pdf
, j. F  T  y. h2 |8 O
The most impressive evidence for me is the graph in page 799 comparing homicide rates in US and Canada from 1950 to 2005. These 2 countries are very similar both culturally and economically, thereby good comparison groups. Canada de facto abolished death penalty in 1963. US Supreme Court abolished death penalty in 1972 and reinstated it 1976. If death penalty has a significant deterrence effect, we should observe differing patterns of change in homicide rates for these 2 countries between 1963-1972, 1973-1976 and after 1977. However, as can be seen from the graph, Canadian homicide rate was in virtual lockstep with US, being one third lower and one third less variable. This casts serious doubt on any large deterrence effect of death penalty.

8 i0 _3 L0 [( ~5 h* P

Dracula

code_abc 发表于 2013-3-6 16:22
" o& [2 Q. |; e- p) m* a% N9 z粗略看了一下，我认为这个分析有一定道理。不过还有值得商榷的地方，文中以死刑执行率的谋杀率做相关性研 ...

As I said, Canada de facto abolished death penalty in 1963. This is from wikipeida:
& L3 Q& L/ V' ~/ f
& v" f# G, D) a5 G, V"Following the success of Lester Pearson and the Liberal Party in the 1963 federal election, and through the successive governments of Pierre Trudeau, the federal cabinet commuted all death sentences as a matter of policy. Hence, the de facto abolition of the death penalty in Canada occurred in 1963, with legal abolition a formality"
9 J/ D) c& u: R# L) S3 B
The de jure abolition of death penalty in Canada came in 1976 and 1998.

Donahue and Wolfers also mentioned studies using the Illinois death penalty moratorium as a natural experiment (page 819). In January 2003, Outgoing Illinois governor George Ryan, pardoned four death-row inmates and commuted the sentences of the remaining 167 to life in prison without the possibility of parole. The authors found "the relationship between homicides in Illinois and the rest of the country is roughly unchanged since the moratorium. If anything, the bars appear persistently negative, suggesting that Illinois experienced about three fewer homicides per month than one would
have expected based upon its previous relationship with the rest of the country"
; a3 Z) c3 K/ \7 C5 g3 v# v# B% i
5 [; B2 y# Z: SI think these answers your question.

Dracula

方恨少 发表于 2013-3-6 22:27
就说这个激情杀人，临时起意的杀人也是杀人，要不要预防呢，死刑对激情杀人也许有震撼作用，也许没有，废 ...

2 J# n) _' y# j) S6 B* bI am not sure of the meaning of 激情杀人. It must be a new word. I never heard of the term when I was in China. There have always been differentiation between murder and manslaughter in English common law. And I think it is a fair and reasonable concept.
  }" i. W& h, k% a. ~$ W( J
0 Y5 i& k' i5 M& b3 {For example, in US, if a person kills someone after provocation, that is an event which would cause a reasonable person to lose self-control, he or she is less culpable to the killing and is guilty of manslaughter instead of murder. The sentence for first-degree manslaughter is from 5-25 years, compared to 25 years which is the minimum sentence for second-degree murder. One example of provocation is a husband going back home, unintentionally catching his wife in bed with another man, losing control and killing his wife or her lover. However, if there is a cooling-off period, for example, the husband withdraws from the house after catching his wife having an affair, walking around for an hour, then decides to kill his wife. This is pre-meditation and will be punished as second-degree murder.
% m# D" z( c2 ?0 U& q3 y6 T% `
$ T" F8 r, q8 ~% j0 d$ b( q# OFor this particular case, I cannot think of any mitigating circumstance in killing the infant, so manslaughter defense does not apply. It can be regarded as extremely heinous, and if he were in US, he would probably be convicted of first-degree murder, receiving death penalty in death penalty states.

: r$ k! U" _# E! h- s0 k

Dracula

- p/ C/ n. x& U  }/ V+ Z& O

方恨少 发表于 2013-3-6 23:46
激情杀人专业的定义是啥我也不知道，我理解就是受到刺激，一时失去理智杀人。这个词是药家鑫案后开始流行 ...

There is pre-meditation in the third scenario you are referring to. If the prosecution can prove this is the case, manslaughter no longer applies. It would be second degree murder.

- R0 p1 F, F! sThis is the same case with self-defense. The belief you are defending yourself against great harm must be reasonable. So in many cases, the prosecution and the defense's arguments hinge on how reasonable these beliefs are.
2 `1 S+ I4 K! X5 v+ l) R

Dracula

code_abc 发表于 2013-3-7 09:20
$ m5 `4 Q4 A( U: i3 C3 ^5 b事实上，整篇报告给我的感觉是模型不正确。从谋杀率和死刑执行率的关系来检查死刑威慑力的模型过于简单。 ...

你提到的几点都对，除死刑外，还有很多其它因素影响杀人率，死刑和杀人率的关系可能是reverse-causality，因此美国没有死刑州的杀人率低于有死刑的州，加拿大杀人率低于美国并不能证明废除死刑会降低杀人率，同样简单的用时间数据将杀人率变化和是否有死刑同样是不可以。但是这篇文章用加拿大做美国的比较对象，用没有死刑的州作有死刑州的比较对象，用Illinois州长George Ryan离任前中止Illinois死刑实行做natural experiment等等就是就是为了解决你提到的这些统计问题。我觉得你读这篇文章之前心里的成见已经比较大，有点对这篇文章统计方法的逻辑视而不见。
1 n, T' X9 e. {9 I+ V1 V6 B
5 a. @0 B  u6 q) B. X/ N  t假设杀人率m=a*d+b*x+e，d是个dummy variable，如果有死刑d=1，否则d=0，x是其它因素，e是noise。如果不控制x的话，即使我们观测到d和m的关系，比如废除死刑的第一年，杀人率比前一年上升，我们也不清楚到底是废除死刑造成的呢，还是其他因素变化造成的，没法作出结论。就具体这个问题来说，重要的因素比如60年代，70年代，美国加拿大baby boom一代成长为青年，青少年在人口的大大比例加大，我们知道青少年犯罪的比例远大于其他年龄组，这一点就会使犯罪率大大增加，其它的包括60年代美国毒品使用开始泛滥，同样会大大提高犯罪率，因此我们观测到加拿大1963年废除死刑后，杀人率升高并不能得出结论废除死刑提高杀人率，类似的80年代以来美国一些州废除死刑，比如1984年Massachusetts，2004年纽约州等，废除后犯罪率大幅度下降，我们也不清楚到底是废除死刑真的降低犯罪率呢，还是这些其他因素变化比如青年人口比例减少导致犯罪降低。因此我们需要comparison group，比如美国和加拿大来解决这个问题。美国加拿大经济文化，人口组成等都非常类似，因此这些因素的变化应该也是类似的，是很好的control group。1963年加拿大废除死刑而美国保持不变，我们不应该只是看加拿大在1962到1963年的变化，而是应该看加拿大1962-1963年变化同美国1962-1963年变化的差别，如果这个差别是正的，我们可以比较确信的得出结论，死刑确实减少杀人率，如果差别几乎是0的话，即使死刑的影响真的是正的话，死刑的影响也不会大，同0会很接近。类似的，当各州改变政策，我们用没有改变政策的州作control group，看两者变化的差别。这种方法在统计学上这叫diff-in-diff，是解决endogeneity问题的重要方法之一，我在西西河写过文章介绍这种方法，有时间我可能会贴到这里来。如果你看文章的那两张图，以及统计分析结果就会发现死刑对变化的差别没有显著的影响，因此即使真能降低杀人率，效果也很有限，而且数据也不能让我们排除废除死刑会降低杀人率这个假设的可能。

你提到的可能是犯罪率变化影响死刑政策，reverse causality的问题，diff-in-diff能够部分解决，但是总还有疑问。文章用Illinois州长George Ryan在2000至2003年终止死刑作为natural experiment就是用来解决这个问题。George Ryan当时已经被调查腐败(离任后被判刑5年)，他知道再次当选无望，因此他的决定并不是出于选民要求废除死刑，相当程度是看到大量的无辜的人被判死刑，良心的驱使，因此比较Illinois前后杀人率的变化可以解决这个reverse causality的问题。分析结果我前面引了，没有任何数据证明Illinois因此杀人率提高，而且有一些证据说明Illinois杀人率反而降低。因此这些以及其他证据加在一起，我的看法是即使死刑能降低杀人率，效果也极有限。
+ y$ x3 d* |8 X+ [! }0 S3 n
: T' k1 y, w; Y( F; }4 J
2 I2 u  w1 }' W# {' _/ _
( `9 [( {4 d0 r# v2 L0 ^

Dracula

code_abc 发表于 2013-3-7 16:31
- H2 x6 D( S: w说实话，成见有点，但不是很大，而且我也觉得死刑和杀人率关系不算很大，主要观点我都在第一帖写了。后面 ...

文章的模型有延迟，象表1，表9都是，我在网上打字不方便，选了一个最简单的模型。文中用Illinois, Texas 作natural experiment，就是为了解决反馈的问题。第五部分专门提到用instrumental variable方法解决reverse causality的问题。我还是觉得你的成见很深，没看文章就批评别人的统计方法。

' h7 R) }+ n: h, p. P# q2 t7 s