Description Usage Format Source References Examples
Lawyers collected data on convicted black murderers in the state of Georgia to see whether convicted black murderers whose victim was white were more likely to receive the death penalty than those whose victim was black, after accounting for aggravation level of the murder. They categorized murders into 6 progressively more serious types. Category 1 comprises barroom brawls, liquor-induced arguments lovers' quarrels, and similar crimes. Category 6 includes the most vicious, cruel, cold=blooded, unprovoked crimes.
1 |
A data frame with 12 observations on the following 4 variables.
Aggravation
the aggravation level of the crime, a
factor with levels "1"
, "2"
, "3"
, "4"
,
"5"
and "6"
Victim
a factor indicating race of murder victim, with
levels "White"
and "Black"
Death
number in the aggravation and victim category who received the death penalty
Nodeath
number in the aggravation and victim category who did not receive the death penalty
Ramsey, F.L. and Schafer, D.W. (2002). The Statistical Sleuth: A Course in Methods of Data Analysis (2nd ed), Duxbury.
Woodworth, G.C. (1989). Statistics and the Death Penalty, Stats 2: 9–12.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 | str(case1902)
# Add smidgeon to denominator because of zeros
empiricalodds <- with(case1902, Death/(Nodeath + .5))
plot(empiricalodds ~ as.numeric(Aggravation), case1902, log="y",
pch=ifelse(Victim=="White", 21, 19),
xlab="Aggravation Level of the Murder", ylab="Odds of Death Penalty")
legend(3.8,.02,legend=c("White Victim Murderers","Black Victim Murderers"),pch=c(21,19))
fitbig <- glm(cbind(Death,Nodeath) ~ Aggravation*Victim, case1902, family=binomial)
# No evidence of overdispersion; no statistically significant evidence
# of interactive effect
anova(fitbig, test="Chisq")
fitlinear <- glm(cbind(Death,Nodeath) ~ Aggravation + Victim, case1902, family=binomial)
summary(fitlinear)
# Mantel Haenszel Test, as an alternative
table1902 <- with(case1902, rbind(Death,Nodeath))
dim(table1902) <- c(2,2,6)
mantelhaen.test(table1902)
|
'data.frame': 12 obs. of 4 variables:
$ Aggravation: Factor w/ 6 levels "1","2","3","4",..: 1 1 2 2 3 3 4 4 5 5 ...
$ Victim : Factor w/ 2 levels "White","Black": 1 2 1 2 1 2 1 2 1 2 ...
$ Death : num 2 1 2 1 6 2 9 2 9 4 ...
$ Nodeath : num 60 181 15 21 7 9 3 4 0 3 ...
Analysis of Deviance Table
Model: binomial, link: logit
Response: cbind(Death, Nodeath)
Terms added sequentially (first to last)
Df Deviance Resid. Df Resid. Dev Pr(>Chi)
NULL 11 212.284
Aggravation 5 198.320 6 13.964 < 2.2e-16 ***
Victim 1 11.725 5 2.239 0.0006167 ***
Aggravation:Victim 5 2.239 0 0.000 0.8151731
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Call:
glm(formula = cbind(Death, Nodeath) ~ Aggravation + Victim, family = binomial,
data = case1902)
Deviance Residuals:
1 2 3 4 5 6 7 8
0.02705 -0.03705 -0.27695 0.46062 -0.22255 0.33222 0.02846 -0.03695
9 10 11 12
1.21437 -0.55797 0.00006 0.00007
Coefficients:
Estimate Std. Error z value Pr(>|z|)
(Intercept) -3.4207 0.6144 -5.567 2.59e-08 ***
Aggravation2 1.6090 0.8506 1.892 0.05855 .
Aggravation3 3.3902 0.7474 4.536 5.74e-06 ***
Aggravation4 4.5004 0.7858 5.727 1.02e-08 ***
Aggravation5 5.8814 0.9128 6.443 1.17e-10 ***
Aggravation6 26.2636 8772.8073 0.003 0.99761
VictimBlack -1.7409 0.5426 -3.208 0.00134 **
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
(Dispersion parameter for binomial family taken to be 1)
Null deviance: 212.2838 on 11 degrees of freedom
Residual deviance: 2.2391 on 5 degrees of freedom
AIC: 38.105
Number of Fisher Scoring iterations: 19
Mantel-Haenszel chi-squared test with continuity correction
data: table1902
Mantel-Haenszel X-squared = 9.6983, df = 1, p-value = 0.001844
alternative hypothesis: true common odds ratio is not equal to 1
95 percent confidence interval:
1.910687 15.789312
sample estimates:
common odds ratio
5.49258
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.