Prohibition, Homicide, and GDP

Ari Etin

2025-08-26

Claim Evaluation: Prohibition Caused Murder Rate Increase

Summary
Evaluating Szalavitz’s causal speculation regarding Prohibition and murder rates, I find some evidence of
statistically significant correlation between Prohibition and murder rates in a fixed effects model. Over
this same period, Real GDP has a negative significant correlation with homicide rates. However, neither
relationship holds when strictly evaluating the upward trend in homicide rates leading up to and through
Prohibition.

Bakcground
In chapter 2, Szalavitz writes “the murder rate went from 6.5 per 100,000 in 1918 before Prohibition to 9.7
per 100,000 in 1933, the year of repeal, nearly a 50% rise. Suggesting that the relationship was likely to be
causal, this rate then fell back down under 6 per 100,000 by 1942.” A causal claim from three data points
seems like a stretch, so I decided to explore this some more.
Data from:
• Homicides: fraser.stlouisfed.org, page 414 (columns 972 and 980)
• Historical Real GDP Per Capita : Maddison Project Database 2023

Load and Describe Data

Below is the top few rows from the cleaned dataset.
## Warning: package 'knitr' was built under R version 4.4.3
## Warning: package 'kableExtra' was built under R version 4.4.3

Year Homicides Suicides Prohibition GDP GDPGrowth

1901 1.2 10.4 FALSE 8770.363 9.1170825
1902 1.2 10.3 FALSE 8684.228 -0.9821167
1903 1.1 11.3 FALSE 8941.348 2.9607698
1904 1.3 12.2 FALSE 8663.658 -3.1056794
1905 2.1 13.5 FALSE 9121.332 5.2826829
1906 3.9 12.8 FALSE 9980.113 9.4150810

## Warning: package 'psych' was built under R version 4.4.3

## Warning in FUN(newX[, i], ...): no non-missing arguments to min; returning Inf
## Warning in FUN(newX[, i], ...): no non-missing arguments to max; returning -Inf

1
 vars n mean sd median trimmed mad min max range skew kurtosis se
Year 1 50 1925.5 14.6 1925.5 1925.5 18.5 1901.0 1950.0 0.049.0 -1.3 2.1
Homicides 2 50 6.3 2.2 6.3 6.5 2.1 1.1 9.7 -8.6 0.1 0.3
0.8
Suicides 3 50 13.3 2.0 13.1 13.3 2.4 10.0 17.4 7.4 0.2 -1.2 0.3
Prohibition 4 50 NaN NA NA NaN NA Inf -Inf -Inf NA NA NA
GDP 5 50 11070.2 2292.6 10336.1 10785.7 1695.3 8048.2 16999.3 8951.1 1.1 0.1 324.2
GDPGrowth 6 50 1.5 6.8 1.2 1.9 7.5 -15.6 12.9 28.5 - -0.6 1.0
0.4

## 'data.frame': 50 obs. of 6 variables:

## $ Year : int 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 ...
## $ Homicides : num 1.2 1.2 1.1 1.3 2.1 3.9 4.9 4.8 4.2 4.6 ...
## $ Suicides : num 10.4 10.3 11.3 12.2 13.5 12.8 14.5 16.8 16 15.3 ...
## $ Prohibition: logi FALSE FALSE FALSE FALSE FALSE FALSE ...
## $ GDP : num 8770 8684 8941 8664 9121 ...
## $ GDPGrowth : num 9.117 -0.982 2.961 -3.106 5.283 ...
Data loaded in as expected, without any concerning outliers. Homicides is slightly left-skewed while suicides,
GDP, and GDP growth are all slightly right skewed. Below are the data for the years Szalavitz mentions, to
confirm that we are looking at comparable data:

Year Homicides Suicides Prohibition GDP GDPGrowth

18 1918 6.5 12.3 FALSE 10471.306 7.188271
33 1933 9.7 15.9 TRUE 8048.225 -3.964915
42 1942 5.9 12.0 FALSE 14869.861 9.713130

2
Visually Explore Data

Homicide and Real GDPPC by year

10

16000
8
Homicide Rate

Real GDPPC
14000
6

12000
4

10000
2

8000

1900 1910 1920 1930 1940 1950

Year
Blue dots for years during prohibition, green lines for years listed
We see that the prohibition years do indeed seem to have higher homicide rates, but they may align with an
overall trend of increased homicides starting from around 1910 and ending around 1934. There may also be
some correlation with GDP. Both of these ideas will be evaluated. It is interesting that Szalavitz chooses
dates close to, but not quite, the start and end of prohibition. These dates happen to correspond with local
minima.

3
 Homicide and Real GDPPC Growth by year
10

10
8

5
Homicide Rate

Real GDPPC
6

−5
4

−10
2

−15

1900 1910 1920 1930 1940 1950

Year
Blue dots for years during prohibition, green lines for years listed
The GDP Growth data seems like a mess. For completeness, its value will also be evaluated.

Hypotheses
I propose six hypotheses. The first three regard the entire range of data (1901-1950) and the second regard
just the date range highlighted earlier (1910-1934).
1. H0 : When accounting for fixed effects, Prohibition does not predict homicide rates.
2. H0 : When accounting for fixed effects, Real GDP per Capita does not predict homicide rates.
3. H0 : When accounting for fixed effects, Real GDP per Capita Growth Rates do not predict homicide
rates.
4. H0 : When accounting for fixed effects, and in the date range 1910-1934, Prohibition does not predict
homicide rates.
5. H0 : When accounting for fixed effects, and in the date range 1910-1934, Real GDP per Capita does not
predict homicide rates.
6. H0 : When accounting for fixed effects, and in the date range 1910-1934, Real GDP per Capita Growth
Rates do not predict homicide rates.

Build Regressions
All data (1901-1950)
Since there appears to be an overall upwards trend (regardless of prohibition or GDP) since 1910, I decide to
implement a fixed effects model.

4
##
## Call:
## glm(formula = Homicides ~ Year, data = data)
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -133.13018 36.57305 -3.640 0.000666 ***
## Year 0.07239 0.01899 3.811 0.000393 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for gaussian family taken to be 3.75635)
##
## Null deviance: 234.88 on 49 degrees of freedom
## Residual deviance: 180.30 on 48 degrees of freedom
## AIC: 212.03
##
## Number of Fisher Scoring iterations: 2

Std. Deviance resid.

Pearson Residuals

Residuals vs Fitted 0.0 1.0 2.0 Q−Q Residuals

2 4 3
0
−4

234

4.5 5.5 6.5 7.5 0.0 0.5 1.0 1.5 2.0 2.5

Predicted values Theoretical Quantiles

Std. Pearson resid.

Std. Pearson resid.

Scale−Location Residuals vs Leverage

234
0.8

Cook's distance
−2
0.0

3 2 1

4.5 5.5 6.5 7.5 0.00 0.02 0.04 0.06 0.08

Predicted values Leverage

Add in Real GDP Per Capita (henceforth called GDP):

##
## Call:
## glm(formula = Homicides ~ Year + GDP, data = data)
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)

5
## (Intercept) -2.963e+02 4.488e+01 -6.603 3.28e-08 ***
## Year 1.614e-01 2.396e-02 6.737 2.05e-08 ***
## GDP -7.465e-04 1.524e-04 -4.899 1.18e-05 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for gaussian family taken to be 2.539532)
##
## Null deviance: 234.88 on 49 degrees of freedom
## Residual deviance: 119.36 on 47 degrees of freedom
## AIC: 193.4
##
## Number of Fisher Scoring iterations: 2

Std. Deviance resid.

Pearson Residuals

Residuals vs Fitted Q−Q Residuals

2 3 4
2

1.5
−1

0.0
32 4
−4

4 5 6 7 8 9 0.0 0.5 1.0 1.5 2.0 2.5

Predicted values Theoretical Quantiles

Std. Pearson resid.

Std. Pearson resid.

Scale−Location Residuals vs Leverage

2

32 4
1.0

0
−2

Cook's distance
0.0

4 32 0.5

4 5 6 7 8 9 0.00 0.05 0.10 0.15

Predicted values Leverage

Swap GDP with whether a year was during the Prohibition:

##
## Call:
## glm(formula = Homicides ~ Year + Prohibition, data = data)
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -126.50019 27.17764 -4.655 2.68e-05 ***
## Year 0.06853 0.01412 4.855 1.37e-05 ***
## ProhibitionTRUE 2.87148 0.45372 6.329 8.54e-08 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

6
##
## (Dispersion parameter for gaussian family taken to be 2.071197)
##
## Null deviance: 234.875 on 49 degrees of freedom
## Residual deviance: 97.346 on 47 degrees of freedom
## AIC: 183.21
##
## Number of Fisher Scoring iterations: 2

Std. Deviance resid.

Pearson Residuals

Residuals vs Fitted Q−Q Residuals

34 34
4 3
2

1.5
−2

0.0
34

4 5 6 7 8 9 0.0 0.5 1.0 1.5 2.0 2.5

Predicted values Theoretical Quantiles

Std. Pearson resid.

Std. Pearson resid.

Scale−Location Residuals vs Leverage

34
34
2
1.0

0
−2

Cook's distance
0.0

3 2 1

4 5 6 7 8 9 0.00 0.02 0.04 0.06 0.08

Predicted values Leverage

Add GDP back in:

##
## Call:
## glm(formula = Homicides ~ Year + GDP + Prohibition, data = data)
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -2.448e+02 3.555e+01 -6.886 1.35e-08 ***
## Year 1.331e-01 1.902e-02 7.000 9.11e-09 ***
## GDP -5.355e-04 1.224e-04 -4.375 6.90e-05 ***
## ProhibitionTRUE 2.348e+00 4.035e-01 5.820 5.38e-07 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for gaussian family taken to be 1.494306)
##
## Null deviance: 234.875 on 49 degrees of freedom

7
## Residual deviance: 68.738 on 46 degrees of freedom
## AIC: 167.81
##
## Number of Fisher Scoring iterations: 2

Std. Deviance resid.

Pearson Residuals

Residuals vs Fitted Q−Q Residuals

3 4
2

1.5
0

0.0
−3

24
3

4 5 6 7 8 9 10 0.0 0.5 1.0 1.5 2.0 2.5

Predicted values Theoretical Quantiles

Std. Pearson resid.

Std. Pearson resid.

Scale−Location Residuals vs Leverage
24
3 2
1.0

0
−2

Cook's distance
0.0

4 32

4 5 6 7 8 9 10 0.00 0.05 0.10 0.15

Predicted values Leverage

Add growth rate of GDP:

##
## Call:
## glm(formula = Homicides ~ Year + GDP + GDPGrowth + Prohibition,
## data = data)
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -2.474e+02 3.590e+01 -6.892 1.47e-08 ***
## Year 1.345e-01 1.921e-02 7.005 1.00e-08 ***
## GDP -5.509e-04 1.247e-04 -4.417 6.23e-05 ***
## GDPGrowth 2.042e-02 2.755e-02 0.741 0.462
## ProhibitionTRUE 2.423e+00 4.179e-01 5.799 6.20e-07 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for gaussian family taken to be 1.509095)
##
## Null deviance: 234.875 on 49 degrees of freedom
## Residual deviance: 67.909 on 45 degrees of freedom
## AIC: 169.2

8
##
## Number of Fisher Scoring iterations: 2

Std. Deviance resid.

Pearson Residuals

Residuals vs Fitted Q−Q Residuals

1 4 3
2

1.5
0

0.0
−3

134

4 5 6 7 8 9 10 0.0 0.5 1.0 1.5 2.0 2.5

Predicted values Theoretical Quantiles

Std. Pearson resid.

Std. Pearson resid.

Scale−Location Residuals vs Leverage
134

2
1.0

0
−2

Cook's distance
0.0

2
4 1

4 5 6 7 8 9 10 0.00 0.05 0.10 0.15 0.20

Predicted values Leverage

During the rise (1910-1934)

Since we noticed that curious trend of consistent increase in murder rates over the course of 1910-1935, let’s
see if the models change when we zoom into it.
##
## Call:
## glm(formula = Homicides ~ Year, data = data[(data$Year <= 1934 &
## data$Year >= 1910), ])
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -3.445e+02 1.828e+01 -18.84 1.77e-15 ***
## Year 1.832e-01 9.512e-03 19.25 1.11e-15 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for gaussian family taken to be 0.1176291)
##
## Null deviance: 46.3144 on 24 degrees of freedom
## Residual deviance: 2.7055 on 23 degrees of freedom
## AIC: 21.357
##
## Number of Fisher Scoring iterations: 2

9
 Std. Deviance resid.
Pearson Residuals

Residuals vs Fitted Q−Q Residuals

21 21 10
0.5

22

1.5
16
−0.5

0.0
10

6 7 8 9 0.0 0.5 1.0 1.5 2.0

Predicted values Theoretical Quantiles

Std. Pearson resid.

Std. Pearson resid.

Scale−Location Residuals vs Leverage
10 21 21 0.5

2
16
1.0

−2 0
32
Cook's distance
0.0

10 0.5

6 7 8 9 0.00 0.05 0.10 0.15

Predicted values Leverage

As the graph suggests, year is a significant predictor.

Now add in Real GDP Per Capita (called GDP as above):
##
## Call:
## glm(formula = Homicides ~ Year + GDP, data = data[(data$Year <=
## 1934 & data$Year >= 1910), ])
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -3.444e+02 1.866e+01 -18.451 7.14e-15 ***
## Year 1.830e-01 9.726e-03 18.814 4.76e-15 ***
## GDP 1.980e-05 6.996e-05 0.283 0.78
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for gaussian family taken to be 0.1225296)
##
## Null deviance: 46.3144 on 24 degrees of freedom
## Residual deviance: 2.6957 on 22 degrees of freedom
## AIC: 23.266
##
## Number of Fisher Scoring iterations: 2

10
 Std. Deviance resid.
Pearson Residuals

Residuals vs Fitted Q−Q Residuals

21 21 10
0.5

22

1.5
29
−0.5

0.0
10

6 7 8 9 0.0 0.5 1.0 1.5 2.0

Predicted values Theoretical Quantiles

Std. Pearson resid.

Std. Pearson resid.

Scale−Location Residuals vs Leverage
10 21 1

2
29 0.5
1.0

0
29 32

−2
0.5
Cook's distance
0.0

10 1

6 7 8 9 0.00 0.10 0.20 0.30

Predicted values Leverage

GDP is not a significant predictor of homicide rate in this period.

Swap GDP with whether a year was during the Prohibition:
##
## Call:
## glm(formula = Homicides ~ Year + Prohibition, data = data[(data$Year <=
## 1934 & data$Year >= 1910), ])
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -326.21376 25.64974 -12.718 1.29e-11 ***
## Year 0.17359 0.01339 12.969 8.83e-12 ***
## ProhibitionTRUE 0.19740 0.19445 1.015 0.321
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for gaussian family taken to be 0.1174728)
##
## Null deviance: 46.3144 on 24 degrees of freedom
## Residual deviance: 2.5844 on 22 degrees of freedom
## AIC: 22.212
##
## Number of Fisher Scoring iterations: 2

11
 Std. Deviance resid.
Pearson Residuals

Residuals vs Fitted Q−Q Residuals

21 10
0.5

16 21

1.5
16
−0.5

0.0
10

6 7 8 9 0.0 0.5 1.0 1.5 2.0

Predicted values Theoretical Quantiles

Std. Pearson resid.

Std. Pearson resid.

Scale−Location Residuals vs Leverage
10 21 21

2
16
1.0

0.5

0
20 1

−2
Cook's
10 distance
0.0

6 7 8 9 0.0 0.1 0.2 0.3 0.4 0.5

Predicted values Leverage

Prohibition is not a significant predictor of homicide rate in this period.

Add GDP back in:
##
## Call:
## glm(formula = Homicides ~ Year + GDP + Prohibition, data = data[(data$Year <=
## 1934 & data$Year >= 1910), ])
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -3.246e+02 2.769e+01 -11.721 1.12e-10 ***
## Year 1.728e-01 1.433e-02 12.061 6.63e-11 ***
## GDP -1.463e-05 7.854e-05 -0.186 0.854
## ProhibitionTRUE 2.162e-01 2.230e-01 0.970 0.343
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for gaussian family taken to be 0.1228636)
##
## Null deviance: 46.3144 on 24 degrees of freedom
## Residual deviance: 2.5801 on 21 degrees of freedom
## AIC: 24.171
##
## Number of Fisher Scoring iterations: 2

12
 Std. Deviance resid.
Pearson Residuals

Residuals vs Fitted Q−Q Residuals

21 10
0.5

16 21

1.5
16
−0.5

0.0
10

6 7 8 9 0.0 0.5 1.0 1.5 2.0

Predicted values Theoretical Quantiles

Std. Pearson resid.

Std. Pearson resid.

Scale−Location Residuals vs Leverage
10 21 21

2
16
0.5
1.0

0
32 1

−2
Cook's
10 distance
0.0

6 7 8 9 0.0 0.1 0.2 0.3 0.4 0.5

Predicted values Leverage

With both GDP and Prohibition included, neither is a significant predictor of homicide rate.
Add growth rate of GDP:
##
## Call:
## glm(formula = Homicides ~ Year + GDP + GDPGrowth + Prohibition,
## data = data[(data$Year <= 1934 & data$Year >= 1910), ])
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -3.208e+02 2.818e+01 -11.382 3.44e-10 ***
## Year 1.710e-01 1.455e-02 11.756 1.95e-10 ***
## GDP -5.730e-05 9.292e-05 -0.617 0.544
## GDPGrowth 1.184e-02 1.358e-02 0.872 0.394
## ProhibitionTRUE 3.017e-01 2.448e-01 1.233 0.232
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for gaussian family taken to be 0.1242826)
##
## Null deviance: 46.3144 on 24 degrees of freedom
## Residual deviance: 2.4857 on 20 degrees of freedom
## AIC: 25.238
##
## Number of Fisher Scoring iterations: 2

13
 Std. Deviance resid.
Pearson Residuals

Residuals vs Fitted Q−Q Residuals

21 10
0.5

21

1.5
20
−0.5

20

0.0
10

6 7 8 9 0.0 0.5 1.0 1.5 2.0

Predicted values Theoretical Quantiles

Std. Pearson resid.

Std. Pearson resid.

Scale−Location Residuals vs Leverage
10 21 21

2
20
1.0

0.5

0
32 1

−2
Cook's
10 distance
0.0

6 7 8 9 0.0 0.1 0.2 0.3 0.4 0.5 0.6

Predicted values Leverage

With GDP, Prohibition, and GDP growth rate included, none are a significant predictor of homicide rate.

Model Comparison
All data (1901-1950)
First I explore R2 values using McFadden’s R-squared:
## [1] "Year: 0.232337915649511"
## [1] "Year and GDP: 0.491823803582177"
## [1] "Year and Prohibition: 0.585540447355061"
## [1] "Year, GDP, and Prohibition: 0.707342066647572"
## [1] "Year, GDP, Prohibition, and GDP Growth: 0.71087076434528"
Including GDP and Prohibition seem to both increase model performance. GDP Growth rate did not seem
to have much of an effect.
Then I conduct an anova analysis:
anova(year, gdp, full, fullgrowth)

## Analysis of Deviance Table

##
## Model 1: Homicides ~ Year
## Model 2: Homicides ~ Year + GDP
## Model 3: Homicides ~ Year + GDP + Prohibition
## Model 4: Homicides ~ Year + GDP + GDPGrowth + Prohibition

14
## Resid. Df Resid. Dev Df Deviance F Pr(>F)
## 1 48 180.305
## 2 47 119.358 1 60.947 40.3863 9.263e-08 ***
## 3 46 68.738 1 50.620 33.5432 6.359e-07 ***
## 4 45 67.909 1 0.829 0.5492 0.4625
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Adding GDP to the model reduced about a third of the residual deviation. Prohibition reduced residual
deviation by about one half. Both of these changes occurred regardless of the order of analysis. GDP Growth
had no significant effect.

During the rise (1910-1934)

As before, I first explore R2 values using McFadden’s R-squared:
## [1] "Year: 0.941584707331427"
## [1] "Year and GDP: 0.941796681036699"
## [1] "Year and Prohibition: 0.944198735604419"
## [1] "Year, GDP, and Prohibition: 0.9442908427371"
## [1] "Year, GDP, Prohibition, and GDP Growth: 0.946330910057844"
Including GDP, Prohibition, and GDP Growth had only limited effects.
Then I conduct an anova analysis:
anova(year_rise, gdp_rise, full_rise, fullgrowth_rise)

## Analysis of Deviance Table

##
## Model 1: Homicides ~ Year
## Model 2: Homicides ~ Year + GDP
## Model 3: Homicides ~ Year + GDP + Prohibition
## Model 4: Homicides ~ Year + GDP + GDPGrowth + Prohibition
## Resid. Df Resid. Dev Df Deviance F Pr(>F)
## 1 23 2.7055
## 2 22 2.6957 1 0.009817 0.0790 0.7816
## 3 21 2.5801 1 0.115516 0.9295 0.3465
## 4 20 2.4857 1 0.094484 0.7602 0.3936
As suggested by the R-squared values, the residual deviations do not decrease model to model by nearly as
much. No model reaches the p < 0.05 significance threshold.

Conclusions
Because the effect of including Prohibition in the model was significant, even when accounting for GDP,
we reject the null hypothesis, that Prohibition had no effect on homicide rates. We can also reject the null
hypothesis that GDP had no effect on homicide rates. We fail to reject the null hypothesis that GDP Growth
rates had no effect on homicide rates.
However, when examining the trend observed in the graphs of homicide rates — a steady increase from 1910
until 1934 — all predictors other than fixed effects lose significance. Thus, in this subset of years, we fail to
reject the null hypotheses that GDP, Prohibition, and GDP growth had no effect on homicide rates.
Further exploration may include lagging effects of GDP, effects of war, lagging effects of war (e.g. veterans
returning home with untreated psychological conditions), and other social shifts. Further expanding the
range of data may also provide valuable insights. Different types of homicides are not distinguished between

15
in these data. It is also possible that different operationalizations of GDP, homicides, etc. would result in
different outcomes.

Bibliography
Datasets
Bolt, J., & Van Zanden, J. L. (2024). Maddison Project Database 2023 (Version V1) [Dataset]. DataverseNL.
https://doi.org/10.34894/INZBF2
St. Louis Fed. (January 16, 2020). Suicide and homicide rate changes during Prohibition (1920-1933) in the
United States from 1900 to 1950 (rate per 100,000 people) [Graph]. In Statista. Retrieved August 26, 2025,
from https://www.statista.com/statistics/1088644/homicide-suicide-rate-during-prohibition/

R and R Packages

library(report)

## Warning: package 'report' was built under R version 4.4.3

cite_packages()

## - Makowski D, Lüdecke D, Patil I, Thériault R, Ben-Shachar M, Wiernik B (2023). "Automated Results

## - R Core Team (2024). _R: A Language and Environment for Statistical Computing_. R Foundation for S
## - Wickham H, Vaughan D, Girlich M (2024). _tidyr: Tidy Messy Data_. R package version 1.3.1, <https
## - William Revelle (2025). _psych: Procedures for Psychological, Psychometric, and Personality Resea
## - Xie Y (2025). _knitr: A General-Purpose Package for Dynamic Report Generation in R_. R package ve
## - Zhu H (2024). _kableExtra: Construct Complex Table with 'kable' and Pipe Syntax_. R package versi

16