Wolf’s, Jakubowski’s, Haveman’s, and Courey’s research article, “The Income and Health Effects of Tribal Casino Gaming on American Indians,” is an expansive study about the effects of casino gaming among American Indians’ income, health status, access to healthcare, and health related behaviors (499). These researchers (Wolf, Jakubowski, Haveman, and Courey) found a positive correlation between Indian gaming and American Indian’s income, health status (health related behaviors and access to healthcare) using a difference-in-difference framework and before-after comparisons (500).

Introduction:

Wolf, Jakubowski, Haveman, and Courey presented a broad scope of methods to implement Indian gaming casinos and American Indian Health using statistical data and model estimations to show positive results. Most data was gathered from a study on an income-health gradient that focused on children (500). This case study showed a positive health-income gradient among children and it also showed that as the children got older, the health-income gradient became steeper (500). This case study was followed by a series of studies that investigated the income-health gradient among children and its “change over the life course” (500).
For example, in the Currie and Stabile (2003) study, the researchers (Wolf, Jakubowski, Haveman, and Courey) found that, “income-health gradient among Canadian children was strong and healthy despite access to universal health coverage” (500). Another study soon followed called the Condliffe and Link (2008) case study that found the “determinants of gradient steepening” by looking at the “health shocks” or any series of disadvantages as those children got older (Murasko, 2008) (500). The researchers also found another case study by Murasko (2008) who “attempted to differentiate the influence of permanent from current income,” and another study by Khanama et al. (2009) who looked at the extent factors like parental health and education that “reduced the income gradient among children” (500).

Background:

The researchers (Wolf, Jakubowski, Haveman, and Courey) discussed the health disparities in which they found that American Indians have the lowest levels of income and poverty is double compares on a national level (501). The researchers found that American Indians had the lowest education and about 25% of American Indians who were at least 25 years old and older didn’t have a high school diploma (501).

American Indian Economic Development Efforts:

In the early 1980’s, the U.S. government gave American Indian tribes permission to initiate gaming enterprises Class I level (social gaming) and Class II level (bingo gaming) (501). In 1988, the Indian Gaming Regulatory Act (IGRA) was passed and allowed development of Class III level gaming (501).
The researchers found a 2005 statistics on American Indian gaming statistics that showed that there were about 360 American Indian gaming enterprises (501). There were about 220 federally recognized tribes who operated Class I, Class II, and Class III casinos (501). The researchers found that the National Indian Gaming Commission (NICG), the federal regulatory agency (oversees tribal gaming), estimated that all tribal gaming operations revenue increased from $9.8 billion in 1999 to $25.1 billion in 2006 (501). The researchers also found that the upper Midwest had the most casinos that have been operating for more than 10 years and the new casinos that have been operating for a few years are located mainly in the West (501).

Additional Findings:

The researchers (Wolf, Jakubowski, Haveman, and Courey) found that tribal budgets increased their spending on social services and direct income transfers to members because of the increase in net revenue from gaming enterprises (Gonzales 2003) (503). They found a study that showed a 35% increase in real median income on non-Navajo American Indian reservations with gaming between 1990 to 2000 compared to the 14% on those without gaming on American Indian Reservations (Taylor and Kalt 2005) (503).
In 2000, about 150,000 American Indians were employed in gaming, which was 7% of the total American Indian work force (Taylor and Kalt 2005) (503). By 2008, the National Indian Gaming Association reported that American Indian gaming created 636,000 jobs, but most are filled by non–American Indians (503).
In 2002, the Evans and Topoleski study used a national county-based self-collected casino data and found that when a tribal casino has been opened for more than 4 years, mortality rates significantly decreased by about 2% in counties with tribal casinos and about 1% in counties that were less than 50 miles from a casino (pg. 502).
The researchers found a study that reported improved psychiatric outcomes in a cohort of American Indian children in North Carolina following the opening of a nearby tribal casino, Costello et al. (2003, 2010) (503). In 2008, in the Akee et al. study and with the same cohort of American Indian children in North Carolina, they used a difference-in-differences approach and found that the American Indian adolescents who experienced casino gaming had higher educational outcomes (503).

How Researchers Organized Their Data:

Wolf, Jakubowski, Haveman, and Courey assembled a multilevel and multisource data system to support their estimates of the possible effects of casino gaming on American Indian’s income and health-related behaviors (503-504). Such data included health, socioeconomic, and demographic characteristics, which researchers believe are linked to American Indian tribal gaming operations and community health resources (503-504).
Wolf, Jakubowski, Haveman, and Courey assembled a multilevel and multisource data system to support their estimates of the possible effects of casino gaming on American Indian’s income and health-related behaviors (503-504). Such data included health, socioeconomic, and demographic characteristics, which researchers believe are linked to American Indian tribal gaming operations and community health resources (503-504).
The first system was an Individual-level data that compiled a large sample of information on American Indians (N = 24,079) from the Behavioral Risk Factors Surveillance System (BRFSS) (CDC 1988–2003) (503). This information was based on many variables that include income, health-related issues, residence (in counties only), and basic socio-demographic and economic characteristics (503). With the residence information, the researchers were able to find out who were tribal residents on and off Indian reservation and/or where tribal members were located and whether or not tribes had land reservations or not (503). The second system was a Tribal-level data on American Indian tribes with Class III gaming (collected by Evans and Topoleski , 2002) (503).
Researchers then supplemented all this compiled both data systems with a comprehensive data of tribally owned casinos that were open (open date through 2005), tribally affiliated, by county, and whether or not it was a Class III gaming compact and/or casino-style gaming (504). Researchers also use the U.S. Census Bureau (2008) for information on the geographic location of all tribal reservation lands (504). Contextual data was gathered from the Area Resource File (ARF 2008) because it had pertinent information on the “availability and aggregate utilization of health resources and facilities, population, and economic data for each county” (504). Most of all county-level data for 1989 was collected in 1990 (504).
Researchers used the data to estimate the “exogenous relationship between American Indian gaming and American Indian income,” in which the “casino-generated income” also affected the health of American Indians in many ways (504). These data have several distinct measured the individual-level collected using a “random-digit-dial sampling procedure” that grabs both American Indians who live on Indian reservations and/or tribal land and American Indians who do not live on Indian reservation and/or tribal land (504). The data also included other individual-level variables collected at many intervals and every year was aggregated to build a “dynamic data analysis” (504).

The Researcher’s Model Estimation in 2 Equations:

The researcher’s identified the effects that the Class III Indian gaming casinos had on the income and the health-related behaviors of American Indians by using a “structural two-stage multiple regression model” (504). This model predicted that the Indian casino gaming had an effect on health-related behaviors that also correlated with the increase (exogenous) in income (504). And that income was also connected with an operating Indian gaming casino which researchers predicted that it had no effect on the health-related issues of American Indians (504). The researchers found the connection between gaming and income and then additional connections on Indian gaming casinos that were making more money to more health measures (504). The researcher used this first equation: Y = Y (I, X, U), which shows how the individual income of American Indians is correlated to an Indian gaming casino with a set of separate (exogenous) variables (I, X, and U) (504).
The researchers hypothesized that the household income (Y) is positively correlated to the presence of an Indian gaming casino (I) (504). The (X) is a “vector of exogenous variables” that affects (Y). In other words (Y) is accountable for the “individual, county, and year” and the (U) is an “unmeasured third factor,” which is not associated with (X) and (I) (504).
The second equation is: H = H (Ŷ, X, U), which is a “structural equation” on the health-related variables where (H) is the predicted value of household income from (Ŷ), and the exogenous variables (505). This model rules for both Individual-level data and County-level data that are connected with income and health (505). The researchers also added a “year-specific dummy variables” to look for any past trends related with income or health behaviors in the American Indian population (505). The estimation of both equations is based on American Indians (a full sample) with and without an Indian gaming casino(s) (505). The researchers also estimated a number of other models as “sensitivity tests” (505).
Again the researchers predicted that tribal casino gaming was positively correlated with household income of American Indians and that the income increased because of the connection of a Indian gaming casino (505). In addition to the presence of an Indian casino gaming, was also better health and less negative health behaviors, and easier access to health care for American Indians (505).
The researchers believe that this was because of either a “direct change in household income through payments by tribes to members” or “improvements in tribal health infrastructure” (505). The tribal health infrastructure includes, “health facilities, nutrition programs, and community health workers” (505). The researchers found no correlation of increased employment with Indian gaming casinos and so the earnings of tribal members were not used (505).

Variables Statistics:

The researchers used means and standard deviations to show statistics for each variable used in estimating both equations of the full sample of 24,079 American Indians from the BRFSS data between 1988 to 2003 (505). The researchers created other samples to show different variables in order to look at the means and standard deviations as well. The researchers used an indicator variable to differentiate who lived in a county with an Indian gaming casino between 1988 – 2003 to those who didn’t (505). All full samples showed tribes with and without gaming casinos. There are no significant “differences in socioeconomic characteristics between the Restricted Sample and the Full Sample,” (505).
The researchers had a Restricted Sample of 3,701 American Indians living in a county that had tribal gaming for 2+ years at the time of the study, which were labeled After Gaming Sample (509). The researchers stated that there were “15% of observations in the Full Sample and 41% of observations in the Restricted Sample” (509). The researchers used the variable (After Gaming Sample = 1) in an estimation to differentiate the observations in the Restricted Sample after a casino was established from observations in this sample “before a casino has been established” (509).

Household Income:

The researchers used the household income (taken from BRFSS between 1988-2003) is the dependent variable in Equation 1: Y = Y (I, X, U) of our estimation model (504, 509) is recorded in constant 2,000 dollars (509). The researchers found that the “pooled mean household incomes” for the Full Sample was $33,207, and the Restricted Sample was $31,819, and the After Gaming Sample was $33,377 (509). The difference in household income between those in the Restricted Sample and in the Full Sample was $1,356, which had no significance statistically (509). The difference in household income between those in the After Gaming Sample and the Restricted Sample observed was $2,659; which is “significantly different from zero (p < .05) (509).

Health Indicators:

The health-related variables are the dependent variables from equation 2: H = H (Ŷ, X, U), which were health-related indicators that the researchers used to estimate the effects of casino gaming on American Indians by looking at several indicators such as risky behavior, which included smoking and heavy drinking (509-510). The researchers used several more indicators such BMI, health, Diabetes, High cholesterol to name a few (510). The researchers found in the Full Sample that 36% smoked, 63% were overweight, and 24% in poor or fair health (510). In the Full Sample, researchers found that 75% (BRFSS respondents) having health insurance, but 17% didn’t (510).

Both Restricted Sample and After Gaming Samples:

The researchers stated that the “key independent variable is an indicator variable (= 1)” that was connected to the Restricted Sample (510). The Restricted Sample showed American Indians who lived in a county with Class III Indian gaming casino between 1988–2003 (510). Another indicator variable (= 1) connected to the After Gaming Sample, showed American Indians who lived in a county with a Class III gaming business that was established from “observations in a county before a casino is in place” (510-511). The researchers used “Individual control variables” from the BRFSS, which included “age, gender, education, marital status, and employment status” (511).
Additional indicator variables were distinguished by the education and marital status of American Indians (511). Lastly, the researchers distinguished 3 “binary employment statuses,” and collected county date from the Area Resource File (ARF) to control “possible tribal selectivity” in successful gaming compact negotiations (511). The researchers used various control variables that distinguished “poverty rate; unemployment rate; proportion of employment, rent; female headed households, and household proportions with a telephone (511). Plus, two health indicators for health, mortality rate, and doctors (511). The researchers measured all these variables in 1989 and 1990 to look for any differences in economic conditions and the possibility that medical care providers are linked to a casino (511).

Results of Tribal Gaming and Income of American Indians:

The researchers hypothesized that casino gaming in a county is positively correlated to household income of the American Indians living in the county (511). This link was estimated by a third indicators or variables called the Restricted Sample (= 1 variable) and the After Gaming Sample (= 1 variable) (511).
The researcher’s gaming variables creates a difference-in-differences variable (I in Equation 1) and captures both the “with-without effect and the before-after effect,” which is called that casino gaming indicator (511). The difference-in-differences are rough calculations that are used in economic analysis to examine the effects of a study when a “random assignment experiment” has not been utilized. This method is used to “minimize the possibility that the distribution of outcomes” in a “treated population” could be distinguished by those in the “control population” because of primary differences in the two populations (511).
In this case, the researchers can “isolate the influence of the treatment on the outcome(s)” (511). The difference-in-differences approach assumes that “differences in the outcomes” between the two populations could have possibly stayed constant without treatment so that observed differences before treatment (establishing a casino) can be used to justify any differences (511).

Bibliography:

Wolfe, Barbara, Jakubowski, Jessica, Haveman, Robert, Courey, Marissa. “The Income and Health Effects of Tribal Casino Gaming on American Indians.” Demography 49 (2012): 499–524. Accessed April 1, 2014. DOI 10.1007/s13524-012-0098-8