Our analysis relies on combining public health surveillance data with external information about trends in sports gambling.
Our project utilizes two main sources spanning the period 2017 through 2024:
Primary Source (Health Outcome / Demographic Data): The Behavioral Risk Factor Surveillance System (BRFSS). This is an annual, state-based, nationwide telephone survey conducted by the CDC, providing detailed self-reported health and behavioral data. We analyzed 2.75 million completed interviews across all available years. We chose this period as it spans the period before the 2018 Supreme Court ruling (PASPA repeal), which allowed many states to legalize sports betting.
Secondary Sources (Sports Gambling Policy and Market Activity Data): We used data scraped from Legal Sports Report for market handle and manually compiled state legalization dates from the American Gaming Association.
Due to variations in the BRFSS survey over the years, we standardized
the data. This process was handled by a dedicated source/
code folder, with the following functions ensuring data quality and
analytical readiness:
HLTHPL1 and HLTHPL2).
Seeing that the variables differed in name only (the responses were
coded in the same way), this was handled by renaming different versions
to the single standardized name.
(e.g. hlthpl_standard).standardize_brfss_variable = function(raw_data){
column_names = colnames(raw_data)
raw_data =
raw_data |>
rename_with(.fn = ~ "hlthpl_standard", .cols = starts_with("hlthpl", ignore.case = FALSE)) |>
rename_with(.fn = ~ "rfdrh_standard", .cols = starts_with("rfdrhv", ignore.case = FALSE)) |>
rename_with(.fn = ~ "rfbing_standard", .cols = starts_with("rfbing", ignore.case = FALSE)) |>
rename_with(.fn = ~ "income_standard", .cols = starts_with("income", ignore.case = FALSE)) |>
rename_with(.fn = ~ "medcost_standard", .cols = starts_with("medcost", ignore.case = FALSE))|>
rename_with(.fn = ~ "addepe_standard", .cols = starts_with("addepe", ignore.case = FALSE))
if (!("sdlonely" %in% column_names)) {
raw_data = raw_data |> mutate(sdlonely = NA_integer_)
}
if (!("lsatisfy" %in% column_names)) {
raw_data = raw_data |> mutate(lsatisfy = NA_integer_)
}
if (!("emtsuprt" %in% column_names)) {
raw_data = raw_data |> mutate(emtsuprt = NA_integer_)
}
if (!("sdhemply" %in% column_names)) {
raw_data = raw_data |> mutate(sdhemply = NA_integer_)
}
if (!("sdhbills" %in% column_names)) {
raw_data = raw_data |> mutate(sdhbills = NA_integer_)
}
if (!("sdhutils" %in% column_names)) {
raw_data = raw_data |> mutate(sdhutils = NA_integer_)
}
if ("sex1" %in% column_names) {
raw_data = raw_data |> mutate(sex = sex1)
}
if (!("urbstat" %in% column_names)) {
raw_data = raw_data |> mutate(urbstat = NA_integer_)
}
raw_data
}dispcode = 1100) and converts the date of the interview to
a Date type variable.filter_brfss_data = function(brfss_data_raw, data_year = 2024){
column_names = colnames(brfss_data_raw)
brfss_data_clean =
brfss_data_raw |>
select(qstver, dispcode, state, seqno, iyear, imonth, iday,
sex, marital, educag, employ1, children, income_standard, urbstat, imprace, ageg5yr, hlthpl_standard,
genhlth, rfhlth, physhlth, phys14d, menthlth, ment14d, poorhlth, medcost_standard, totinda, michd, addepe_standard,
decide, diffalon, lsatisfy, emtsuprt, sdlonely, sdhemply, sdhbills, sdhutils, rfbing_standard, rfdrh_standard
) |>
mutate(
date = as.Date(paste(iyear, imonth, iday, sep = "-"))
) |>
filter(
dispcode == 1100,
year(date) == data_year
) # only selecting completed interviews in the right year
brfss_data_clean
}1 ~ "Male"), and connects the date of each
interview to the status of legalization in that state. This
sb_legal treatment variable is added by merging with state_legalization_dates.csv.load_clean_brfss = function(filepath){
#get date when sports betting became legal by state
sb_legal_dates =
read_csv(here::here("data", "legal_sports_report","state_legalization_dates.csv")) |>
select(first_start, abbr, fips)
cleaned_df =
read_csv(filepath) |>
mutate(
id = paste0(state, seqno), # concatenating state and seqno to get a unique row id for each interview
sex = case_match(sex, 1 ~ "Male", 2 ~ "Female", .default = NA) |> as.factor(),
marital_status =
case_match(
marital,
1 ~ "Married",
2 ~ "Divorced",
3 ~ "Widowed",
4 ~ "Separated",
5 ~ "Never Married",
6 ~ "Member of an unmarried couple (Partner)",
.default = NA) |> as.factor(),
education_status =
case_match(
educag,
1 ~ "Did not graduate High School",
2 ~ "Graduated High School",
3 ~ "Attended College or Technical School",
4 ~ "Graduated from College or Technical School",
.default = NA) |> as.factor(),
employment_status =
case_match(
employ1,
1 ~ "Employed for wages",
2 ~ "Self-employed",
3 ~ "Out of work (<1 year)",
4 ~ "Out of work (>1 year)",
5 ~ "Homemaker",
6 ~ "Student",
7 ~ "Retired",
8 ~ "Unable to work (Disabled)",
.default = NA) |> as.factor(),
children_in_household = case_match(children, 88 ~ 0, 99 ~ NA, .default = children),
income_level =
case_match(
income_standard,
1 ~ "Less than $10,000",
2 ~ "$10,000 to < $15,000",
3 ~ "$15,000 to < $20,000",
4 ~ "$20,000 to < $25,000",
5 ~ "$25,000 to < $35,000",
6 ~ "$35,000 to < $50,000",
7 ~ "$50,000 to < $75,000",
c(8, 9, 10, 11) ~ "$75,000 or more",
.default = NA
) |> as.factor(),
insurance_coverage =
case_match(
hlthpl_standard,
1 ~ "Have some form of insurance",
2 ~ "Do not have some form of health insurance",
9 ~ NA,
.default = NA) |> as.factor(),
race =
case_match(
imprace,
1 ~ "White, Non-Hispanic",
2 ~ "Black, Non-Hispanic",
3 ~ "Asian, Non-Hispanic",
4 ~ "American Indian, Alaska Native, Non-Hispanic",
5 ~ "Hispanic",
6 ~ "Other/Multiracial, Non-Hispanic",
.default = NA) |> as.factor(),
age_group_5yr =
case_match(
ageg5yr,
1 ~ "18-24",
2 ~ "25-29",
3 ~ "30-34",
4 ~ "35-39",
5 ~ "40-44",
6 ~ "45-49",
7 ~ "50-54",
8 ~ "55-59",
9 ~ "60-64",
10 ~ "65-69",
11 ~ "70-74",
12 ~ "75-79",
13 ~ "80 or older",
.default = NA) |> as.factor(),
urban_status =
case_match(
urbstat,
1 ~ "Urban counties",
2 ~ "Rural counties",
.default = NA) |> as.factor(),
general_health =
case_match(
genhlth,
1 ~ "Excellent",
2 ~ "Very Good",
3 ~ "Good",
4 ~ "Fair",
5 ~ "Poor",
.default = NA) |> as.factor(),
general_health_refactored =
case_match(
rfhlth,
1 ~ "Good or Better Health",
2 ~ "Fair or Poor Health",
.default = NA) |> as.factor(),
michd = case_match(michd, 1 ~ "Yes", 2 ~ "No", .default = NA) |> as.factor(),
physical_health = case_match(physhlth, 88 ~ 0, 77 ~ NA, 99 ~ NA, .default = physhlth),
physical_health_not_good_days = case_match(phys14d, 1 ~ "0", 2 ~ "1-13", 3 ~ "14+", 9 ~ NA, .default = NA) |> as.factor(),
leisure_physical_activity_last_30_days = case_match(totinda, 1 ~ "Yes", 2 ~ "No", 9 ~ NA, .default = NA) |> as.factor(),
mental_health = case_match(menthlth, 88 ~ 0, 77 ~ NA, 99 ~ NA, .default = menthlth),
mental_health_not_good_days = case_match(ment14d, 1 ~ "0", 2 ~ "1-13", 3 ~ "14+", 9 ~ NA, .default = NA) |> as.factor(),
depressive_disorder = case_match(addepe_standard, 1 ~ "Yes", 2 ~ "No", .default = NA) |> as.factor(),
poor_health = case_match(poorhlth, 88 ~ 0, 77 ~ NA, 99 ~ NA, .default = poorhlth),
binge_drink = case_match(rfbing_standard, 1 ~ "No", 2 ~ "Yes", 9 ~ NA, .default = NA) |> as.factor(),
heavy_drink = case_match(rfdrh_standard, 1 ~ "No", 2 ~ "Yes", 9 ~ NA, .default = NA) |> as.factor(),
medical_cost_barrier = case_match(medcost_standard, 1 ~ "Yes", 2 ~ "No", 7 ~ NA, 9 ~ NA, .default = NA) |> as.factor(),
difficulty_self_care = case_match(diffalon, 1 ~ "Yes", 2 ~ "No", .default = NA) |> as.factor(),
life_satisfaction =
case_match(
lsatisfy,
1 ~ "Very satisfied",
2 ~ "Satisfied",
3 ~ "Dissatisfied",
4 ~ "Very dissatisfied",
.default = NA_character_
) |> as.factor(),
emotional_support =
case_match(
emtsuprt,
1 ~ "Always",
2 ~ "Usually",
3 ~ "Sometimes",
4 ~ "Rarely",
5 ~ "Never",
.default = NA_character_
) |> as.factor(),
loneliness =
case_match(
sdlonely,
1 ~ "Always",
2 ~ "Usually",
3 ~ "Sometimes",
4 ~ "Rarely",
5 ~ "Never",
.default = NA_character_
) |> as.factor(),
lost_reduced_employment = case_match(sdhemply, 1 ~ "Yes", 2 ~ "No", .default = NA) |> as.factor(),
financial_strain_bills = case_match(sdhbills, 1 ~ "Yes", 2 ~ "No", .default = NA) |> as.factor(),
financial_strain_utilities = case_match(sdhbills, 1 ~ "Yes", 2 ~ "No", .default = NA) |> as.factor(),
# Adding binary versions of outcomes for prop tests and regression
any_physical_health_not_good_days =
case_match(
physical_health_not_good_days,
"0" ~ FALSE,
"1-13" ~ TRUE,
"14+" ~ TRUE
),
any_mental_health_not_good_days =
case_match(
mental_health_not_good_days,
"0" ~ FALSE,
"1-13" ~ TRUE,
"14+" ~ TRUE
),
has_depressive_disorder =
case_match(
depressive_disorder,
"No" ~ FALSE,
"Yes" ~ TRUE
),
has_binge_drink =
case_match(
binge_drink,
"No" ~ FALSE,
"Yes" ~ TRUE
)
) |>
#add in whether sports betting was legal at the time of the interview
#NOTE: legalization dates are set to the first of the month so we will check that the interview was the following month (not day) by temporarily creating a new column month_of_interview
merge(sb_legal_dates, by.x = "state", by.y = "fips", all = TRUE) |>
mutate(
month_of_interview = floor_date(date, unit = "month"),
sb_legal = (as.numeric(month_of_interview > first_start))
) |>
# dropping any overwritten variables
select(
#Survey Identifiers
id, date, qstver,
# Demographic/Predictor Variables
state, urban_status, age_group_5yr, sex, marital_status, education_status, employment_status, children_in_household, income_level, race, insurance_coverage,
# Health/Financial Outcomes Variables
medical_cost_barrier, general_health, general_health_refactored, michd,
physical_health, physical_health_not_good_days, leisure_physical_activity_last_30_days,
mental_health, mental_health_not_good_days, poor_health, depressive_disorder, difficulty_self_care,
life_satisfaction, emotional_support, loneliness,
binge_drink, heavy_drink,
lost_reduced_employment, financial_strain_bills, financial_strain_utilities,
any_physical_health_not_good_days, any_mental_health_not_good_days,
has_depressive_disorder, has_binge_drink,
#Sports Betting Legal
sb_legal
)
cleaned_df
}Our analysis focuses on the following treatment variable (exposure), outcome variables (adverse health indicator), and demographic variables.
sb_legal: A binary flag indicating whether the
interview took place after sports betting was legalized
in that state (\(\text{1}\)) or before
(\(\text{0}\)).We primarily use binary (Yes/No) measures to capture the prevalence of these adverse health outcomes in the population:
any_mental_health_not_good_days is TRUE.any_physical_health_not_good_days is
TRUE.has_depressive_disorder is TRUE.has_binge_drink is TRUE.To control for social and economic factors influencing health, we include individual-level demographic predictors (e.g., sex, age group, income level, education, and employment status) in our regression models. These are important in our regression models.