Research Question

Is the mean poverty rate different among U.S. counties with no smoking ban, a partial smoking ban, and a comprehensive smoking ban?

Introduction

For this project, I am using the OpenIntro county_complete dataset. The dataset contains 3,142 U.S. counties and 188 variables. Each row represents one county. The dataset includes information about population, income, education, employment, poverty, health, housing, and other county-level characteristics.

For my research question, I will focus on two variables. poverty_2017 is a quantitative variable that represents the percent of the county population living below the poverty level in 2017. smoking_ban_2010 is a categorical variable that identifies whether a county had no smoking ban, a partial smoking ban, or a comprehensive smoking ban in 2010. The dataset comes from OpenIntro and can be accessed at https://www.openintro.org/data/index.php?data=county_complete.

Data Analysis

For the data analysis, I will first import and inspect the dataset. Then I will select the two variables needed for my question, remove missing values, and convert the smoking-ban variable into a factor. I will use summary statistics to compare the poverty rates across the three groups. I will also create a histogram to look at the overall poverty-rate distribution and a boxplot to compare the three smoking-ban groups.

county <- read.csv("https://www.openintro.org/data/csv/county_complete.csv")

dim(county)
## [1] 3142  188
head(county[, c("poverty_2017", "smoking_ban_2010")])
##   poverty_2017 smoking_ban_2010
## 1         13.7             none
## 2         11.8             none
## 3         27.2          partial
## 4         15.2             none
## 5         15.6             none
## 6         28.5             none
df <- county |>
  select(poverty_2017, smoking_ban_2010) |>
  filter(!is.na(poverty_2017), !is.na(smoking_ban_2010)) |>
  mutate(smoking_ban_2010 = factor(smoking_ban_2010))

dim(df)
## [1] 3114    2
summary(df)
##   poverty_2017        smoking_ban_2010
##  Min.   : 2.40   comprehensive: 554   
##  1st Qu.:11.40   none         :1925   
##  Median :15.20   partial      : 635   
##  Mean   :15.99                        
##  3rd Qu.:19.40                        
##  Max.   :52.00
table(df$smoking_ban_2010)
## 
## comprehensive          none       partial 
##           554          1925           635
group_summary <- df |>
  group_by(smoking_ban_2010) |>
  summarise(
    counties = n(),
    mean_poverty = mean(poverty_2017),
    median_poverty = median(poverty_2017),
    sd_poverty = sd(poverty_2017)
  )

group_summary
## # A tibble: 3 × 5
##   smoking_ban_2010 counties mean_poverty median_poverty sd_poverty
##   <fct>               <int>        <dbl>          <dbl>      <dbl>
## 1 comprehensive         554         14.6           13.6       6.14
## 2 none                 1925         16.6           16.1       6.46
## 3 partial               635         15.3           14.1       6.71
ggplot(df, aes(x = poverty_2017)) +
  geom_histogram(binwidth = 2, color = "black", fill = "lightblue") +
  labs(
    title = "Distribution of County Poverty Rates",
    x = "Poverty Rate (%)",
    y = "Number of Counties"
  )

ggplot(df, aes(x = smoking_ban_2010, y = poverty_2017)) +
  geom_boxplot(fill = "lightblue") +
  labs(
    title = "Poverty Rate by County Smoking Ban Type",
    x = "Smoking Ban Type",
    y = "Poverty Rate (%)"
  )

Justification of Approach

The outcome variable, poverty_2017, is quantitative, while smoking_ban_2010 is categorical with three independent groups. Because I am comparing the mean poverty rate across three groups, a one-way ANOVA is an appropriate statistical test. The summary table helps compare the average poverty rates and variability in each group. The histogram shows the overall distribution of poverty rates, and the boxplot makes it easier to visually compare the three smoking-ban groups.

Statistical Analysis

The null hypothesis states that the mean poverty rate is the same for all three smoking-ban groups. The alternative hypothesis states that at least one group has a different mean poverty rate.

H₀: μₙₒₙₑ = μₚₐᵣₜᵢₐₗ = μcₒₘₚᵣₑₕₑₙₛᵢᵥₑ

Hₐ: At least one of the mean poverty rates is different.

I will use a significance level of α = 0.05.

anova_model <- aov(poverty_2017 ~ smoking_ban_2010, data = df)

anova_results <- summary(anova_model)

anova_results
##                    Df Sum Sq Mean Sq F value   Pr(>F)    
## smoking_ban_2010    2   2018  1008.9   24.21 3.67e-11 ***
## Residuals        3111 129624    41.7                     
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
p_value <- anova_results[[1]][["Pr(>F)"]][1]

p_value
## [1] 3.673333e-11

The ANOVA p-value is 3.673^{-11}. Since the p-value is less than 0.05, I reject the null hypothesis.

This means there is enough evidence to conclude that at least one smoking-ban group has a different mean poverty rate.

Discussion of Results

The summary statistics and boxplot allow the poverty rates of the three smoking-ban groups to be compared before running the hypothesis test. The ANOVA then tests whether the differences among the sample means are statistically significant. Based on the p-value and an alpha level of 0.05, the test determines whether the null hypothesis should be rejected.

The results should be interpreted as an association, not as proof that smoking-ban policies cause poverty rates to change. Counties can differ in many other ways, such as population size, income, education, employment, and location. These other factors may also be related to poverty and smoking-ban policies.

Conclusion and Future Directions

This project examined whether mean poverty rates differ among U.S. counties with no smoking ban, a partial smoking ban, and a comprehensive smoking ban. I cleaned the dataset, created summary statistics, used visualizations to compare the groups, and performed a one-way ANOVA. The hypothesis test gives a statistical answer to the research question using a significance level of 0.05.

For future research, I could include other county-level variables such as median household income, unemployment rate, education level, or metropolitan status. I could also compare data from different years to see whether the relationship between poverty and smoking-ban type changes over time.

References

OpenIntro. United States Counties (county_complete).
https://www.openintro.org/data/index.php?data=county_complete

OpenIntro CSV dataset.
https://www.openintro.org/data/csv/county_complete.csv