不為任何立場，只是想正確的把事情弄清楚。
我把你給的參考資料都看了，
也看了一些參考資料引用的原始論文，
首先看原文:
※ 引述《jeanvanjohn (尚市長)》之銘言：
: ※ 引述《hoshi1992 (YAYAUUU)》之銘言：
: : 為了避免那99%的真正的殺人犯被放出來再作案
: : (假釋or之類的)
: : 1%冤死的是能接受的代價
: : 阿可是阿
: : 幹這1%又不是你，憑什麼決定他們要被犧牲阿
: : 完全是一種別人的孩子死不完的概念
: : 這就是反廢死最機掰的地方啦
: 那反過來說，殺人犯被放出來有3%機率再殺人，那3%冤死是不是就是可接受的代價?
^^^^^^^^^^^^
這個3%的來源是我想探究的地方，因為我以前從沒看過這麼篤定的數字
: 不好意思，廢死聯盟的人認為，"和誤判相比，這樣的3%是可以接受的":
引用1: http://www.fhl.net/2000/death_issue/issue3-5.htm
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
這個引用是陳文珊的文章，文章中跟 3% 這個數字有關的文字如下:
1972年533位死囚經富爾曼(Furman)最高法院改判後，監禁或假釋出獄的人中只有六人再
次 犯下謀殺案。
也就是說 再犯率應為 6/533 = 1.多%，但後來原PO說陳文珊的文章有欺騙之嫌，
所以又在推文中補了兩個參考資料。
================================================================
引用2.The Case Against The Death Penalty by Hugo Adam Bedau
http://users.rcn.com/mwood/deathpen.html
姑且不論該文章引用各種資料和文獻後，
最後的結論是支持廢死，我們來看看關於1972年這個案件:
這個文章在Deterrence單元中的第三個小單元中提到:
A recent study examined the prison and post-release records of 533 prisoners
on death row in 1972 whose sentences were reduced to life by the Supreme
Court's ruling in Furman. The research showed that 6 had committed another
murder. But the same study showed that in 4 other cases, an innocent man had
been sentenced to death.(16)
簡單來說就是 再犯率 6/533 和陳文珊的文章內容是一樣的，
另外後面的意思好像是指說有4個犯人是被無辜判死。(有錯歡迎更正)
引用3.A National Study of the Furman-Commuted Inmates: Assessing the Threat to Society from
Capital Offenders by James W. Marquart & Jonathan R. Sorensen
http://tinyurl.com/obn22r5
好的，這篇就是引用2中上面那段最後(16)的那篇論文，
在論文最後conclusions中提到:
While there is some ambiguity as to just how many death row prisoners were actually affected by Furman, the research reported here accounts
for 558 inmates scattered across twenty-nine states and the District of Columbia.91 We tracked these 558 prisoners' institutional and
release behavior for nearly fifteen years. In the prison setting, these prisoners committed six murders-killing four prisoners and two correctional
officers. However, the majority of the former death row prisoners served out their sentences with few instances of serious
institutional misconduct. A minority were responsible for the bulk of disciplinary
infractions.
Over the course of fifteen years, 239 Furman-commuted prisoners were released to the free community. These parolees have spent an average
of five years in society. Twenty-one percent recidivated and were returned to prison, 12% committing new felonies.94 Only one parolee
from Texas committed a second homicide. On the other hand, nearly 80% of those released to the free society have not, at least officially,
committed additional crimes.
也就是說，558人中，在監獄中殺了4個囚犯和2個懲教官員，
而真正放出來的犯人239名只有一個假釋者犯下第二次殺人罪行，
以下是原PO貼了以上引用之後的敘述:
重點是，這份報告是廢盟也認可的報告。
1972年，有558名被判死刑的犯人因為美國最高法院對死刑的判定標準改變而免死，
其中315人(含死在獄中以及逃獄被抓者)未出監，實際出監者243人。
在243人中，再犯各種罪行的是52人(21%)，其中再犯謀殺的是7人，
所以確切的數字是2.8%。
也就是說，死刑犯被假釋出獄後，有3%的機率再犯殺人罪。
這是美國以及廢死聯盟也認可的數據。
===============================================================
怎麼算都不是這個數據，此外這篇論文最後來提到:
H.L.A. Hart asked: "What is the weight and character of the evidence that the
death penalty is required for the protection of society?" This question, in
our opinion, is the most salient one in any discussion of the utility of
capital punishment. Seven (1.3%) Furman-commuted prisoners were responsible
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
for seven additional murders. Certainly execution of all 558 prisoners would
have prevented these killings. However, such a "preemptive strike" would not
have greatly protected society. In addition, four innocent prisoners would
have been put to death. The question then becomes whether saving the lives of
the seven victims was worth the execution of four innocent inmates.
The data in this paper suggest that these prisoners did not represent a
significant threat to society. Most have performed well in the prison; those
few who have committed additional violent acts are indistinguishable from
those who have not. Therefore, over-prediction of secondary violence is
indicated. More than two-thirds of the Furman inmates, using a very liberal
definition of violence, were false positives-predicted to be violent but were
not. We cannot conclude from these data that their execution would have protected or benefitted society.
也就是說588人最後再犯下殺人案件的有7人，然後被誤判的有4人，
論文最後的觀點是對死刑抱持懷疑的態度，認為死刑未必能夠作為保護或使社會受益。
排版不好請見諒 (_._)

簡單來說 就是尚市長不是算錯就是偽造數據呵