Abstract
The
difference in the infant mortality rate between states in the United States is
immense. No better is this demonstrated than in comparing the American South
with the other sub regions of the United States. The South exhibits the lowest
average median household income and the highest infant mortality rates in the
country. Testing shows a statistically significant negative relationship
between median household income and the infant mortality rate in the South.
Similar tests were done for the Gini coefficient and the number of physicians
per capita but results were inconclusive.
The
infant mortality rate is one of the most effective indicators of quality of
life in a country. Factors such as ease
of access to healthcare, quality of healthcare, and income are captured in this
statistic(MacDorman, 2008). The United States ranks as the country with the 57th
lowest infant mortality rate in the world at 5.8, ranked between Latvia and
Bosnia and Herzegovina (CIA). The infant mortality rate is calculated as the
amount of infant deaths under the age of one for every 1000 live births. A ranking
this low is particularly surprising for a country as wealthy as the United
States. However, upon further review of the data, there is a large amount of
variance in infant mortality rates between states. This is not necessarily
surprising, given that there is considerable economic and political diversity
within the United States itself. What is surprising are the trends that become
immediately apparent, most notably the sharp difference between the traditional
American South and the rest of the country. The South, a region as defined by
The United States Census Bureau as shown in figure 1, has an average infant mortality
rate of 7.22, landing the region between Kuwait and Puerto Rico (CIA). In this
essay I will attempt to explain the gap using statistical analysis of quality
of life databases to further understand the causes of this disparity.
Figure 1 Census Regions (Census Bureau
Regions and Divisions)
Literature
Review
There is a wide
variety of research focused on income inequality, especially in recent decades
as inequality has continued to rise (Lawrence). By in large, researchers have
focused on various indices that attempt to model the quality of life of
citizens. By synthesizing the contributions
of previous researchers, I hope to make a connection between these factors and
the unusually high Infant Mortality Rate in the United States.
In “Income Distribution and Mortality: Cross Sectional
Ecological Study of the Robin Hood Index in the United States” Kennedy,
Kawachi, and Prothrow-Stith attempt to determine the effects that standard
measures of income inequality such as the Robin Hood Index (also known as the
Hoover or Schutz index) and the Gini coefficient have on general mortality in the
United States. The Robin Hood Index is an estimation of the redistribution of
wealth from citizens about the mean to below the mean (Kennedy, Kawachi, and
Prothrow-Stith 1996, p. 1004). Similarly, the Gini coefficient also measures
inequality. The results of the researchers’ tests suggested a high correlation
between mortality and the Robin Hood Index, while the Gini coefficient
expressed very little correlation(Kennedy et al., 1996, p. 1005). Although, Kennedy does mention a
study done by the OECD that showed a high correlation between the Gini
coefficient and average life expectancy.
“The Wealth of Nations Revisited: Income and Quality of
Life” by Ed Diener and Carol Diener sets out to answer the question, “Does
economic prosperity enhance the quality of human life.” (Diener & Diener 275).
To answer this question, Diener analyzed 32 indices that constituted a
“diverse, representative set of human values.” (Diener & Diener p. 276).
The physician per capita metric is one of these indices that is particularly
relevant to the topic of this paper. The number of physicians per capita
doesn’t necessarily signal the quality of healthcare, but it may represent the
ease of access to health facilities. Diener also makes another relevant
statement: “Wealthy nations have faults that must not be understated; the high
crime rate of the U.S.A. comes readily to mind.” (Diener & Diener p. 284).
Although crime rates are not explicitly relevant to the topic, the sentiment
rings true that wealth does not solve all problems.
Data
Data was
pulled from government databases, and nearly all for the year 2016, I will touch
on the two exceptions shortly. For the infant mortality rate, the CDC’s
National Center for Health Statistics was used. However, the CDC’s data is
strangely incorrect for Vermont. Vermont is listed as 0.0 infant mortality
rate. To amend these discrepancies, I used the CDC’s 2015 data for Vermont. I
considered using a different source for Vermont, but I decided against it due
to possible differences in how the two sources would measure the rate. Median
household income data was pulled from the United States Census Bureau’s
“Household Income: 2016” paper, Gini coefficient data was also pulled from the
USCB. The physician data comes from the “2017 State Physician Workforce Data
Report” published by the Association of American Medical Colleges. Below are
the summary statistics for the mentioned variables. Although, this doesn’t
become relevant until the results section, obesity data was taken from State of
Obesity.
Variables
|
Mean
|
Std. Dev.
|
Min
|
Max
|
Infant
Mortality Rate
|
6.09
|
1.25
|
3.7
|
9.1
|
Median
Household Income
|
58,143
|
9,819.79
|
41,754
|
78,945
|
Gini
Coefficient
|
0.47
|
0.02
|
0.41
|
0.54
|
Physicians
per 100,000
|
277.83
|
99.3
|
186.1
|
866.3
|
Theoretical Model:
Predicted IMRi = β0 (+/-) β1Median HH Incomei (+/-) β2GiniCoefficienti (+/-) β3PhysiciansPer100ki (+/-) β4South
IMR
= The infant mortality rate per 1000 live births.
Median
HH Income = Median HH Income of each state.
Gini
Coefficient = The Gini coefficient value that each state possesses.
PhysiciansPer100k
= The number of physicians in each state, adjusted for population.
South
= 0 if the state does not fall in the region, 1 if it does
Using an ordinary least squares model, I attempt to
predict the rate of infant deaths based on a few key variables from the
literature that I believed would best describe the dependent variable. The first these regressors is the Median
Household Income in each state. Household income is predicted to have a
negative effect on the infant mortality rate, increases in income decrease the
rate of infant mortality. One thing to note is the decision to keep the
variable in its standard form, it would be reasonable to suggest that increased
household income would only decrease the infant mortality rate up to a certain
point, where the association levels off. This makes sense in theory, for
example, say a state had a median income of $1,000,000.00 dollars, affording
medical care would no longer be an issue, yet the state would still have infant
mortalities due to unavoidable circumstances. The problem in applying this to
the data is that the leveling off point is unknown, it may be outside our range
of values. Ultimately, I elected towards not potentially biasing my model with
a subjective decision.
The
Gini coefficient is expected to have a positive effect on the infant mortality
rate. This is due to the nature of the Gini coefficient. The value of the Gini
coefficient is between 0 and 1, 0 representing perfect inequality and 1
representing complete inequality (Household Income p.1). As income inequality
increases, the infant mortality rate is expected to rise. As for physicians,
the sign is expected to be negative, for the reasons mentioned in the
literature review. Finally, a dummy variable is used to account for the states
in the South. The expected association should be positive if the reasoning is
correct. This variable will be critical in determining if there is a
statistically significant difference in mortality rates in the South.
Results
The results of the model are mixed. Both the Gini
coefficient and physicians per capita exhibit the opposite signs that were
predicted as shown in table 1 below. The Gini coefficient was shown to have a
negative relationship with infant mortality, while physicians per capita has a
positive correlation. This outcome suggests that some of the income inequality
measure is already captured inside the differences in household income. When making this model, I used the assumption
that the United States is politically and economically diverse. While that is
true, it seems that diversity is overstated.
The physician result is interesting. The very slight
positive relationship indicates there may be an omitted variable that is
potentially skewing the results. To address this, I added in a variable
representing the obesity rate in each state. The intent of this is to more
accurately represent the healthcare system in America, where obesity is a major
health concern. States with higher levels of obesity may serve as a signal for
increased healthcare resources. In theory this might make sense, however in
reality the results were even more muddled, table 3 shows this issue. While the
obesity variable was shown to have an adverse impact on infant mortality;
ultimately, this model simply does not do an accurate job of modeling the
health care system’s effects on the infant mortality rate.
The model is not a complete disappointment however, the
median household income expressed our predicted sign, and had a t-score of -4.09,
a statistically significant value. This allows us to conclude that changes in
household income do have a meaningful impact on the infant mortality rate. Similarly,
the South dummy variable has a t-score of 4.88, suggesting that states in the
South have a uniformly lower infant mortality rate than the rest of the United
States.
Figure 2 Blue series represents
observed data while orange represents values predicted by the model
Conclusion
The
goal of this paper was to gather a greater understanding of the factors that
affect the infant mortality rate and why there is so much variance in the rate
within the United States, particularly in the South. Although the model was not
a complete success, the effect that the health care system has remains unclear.
A greater understanding of the role that a household’s income has on the
well-being of the household has been reached.
References
Association of American Medical colleges. 2017 State
Physician Workforce Data Report [Data file]. Retrieved from https://members.aamc.org/eweb/upload/2017%20State%20Physician%20Workforce%20Data%20Report.pdf
Diener, E., & Diener, C. (1995). The Wealth of
Nations Revisited: Income and Quality of Life. Social Indicators
Research, 36, 275-286. Retrieved April 20, 2018, from
https://www.jstor.org/stable/27522878?seq=1#page_scan_tab_contents.
Kennedy, B. P., Kawachi, I., & Prothrow-Stith, D.
(1996). Income Distribution And Mortality: Cross Sectional Ecological Study Of
The Robin Hood Index In The United States. BMJ: British Medical
Journal, 312, 1004-1007. Retrieved April 20, 2018, from
https://www.jstor.org/stable/29731386?seq=2#page_scan_tab_contents.
Lawrence, Robert Z. [PetersonInstitute]. (2015, July
31). Has Productivity Outstripped Wage
Growth? [Video file]. Retrieved from https://www.youtube.com/watch?v=8sf3kt1KduY&index=1&list=PLMTpY1ixcT_Cz37OI96p-5i4gjG8kqgt5&t=0s
MacDorman, M. F., & Mathews, T. (2008, October 9).
Recent Trends in Infant Mortality in the United States. Retrieved from https://www.cdc.gov/nchs/products/databriefs/db09.htm
National Center for Health Statistics. (2016). Infant Mortality Rates by State [Data
file]. Retrieved from https://www.cdc.gov/nchs/pressroom/sosmap/infant_mortality_rates/infant_mortality.htm
National Center for Health Statistics. (2017). Infant Mortality Rates by State [Data
file]. Retrieved from https://www.cdc.gov/nchs/pressroom/sosmap/infant_mortality_rates/infant_mortality.htm
The
State of Obesity. (2017). Adult Obesity
in the United States [Data file]. Retrieved from https://stateofobesity.org/adult-obesity/
The World Factbook. (n.d.). Retrieved from https://www.cia.gov/library/publications/the-world-factbook/rankorder/2091rank.html
United States Census Bureau. (2017). Household Income: 2016 [Data file].
Retrieved from https://www.census.gov/content/dam/Census/library/publications/2017/acs/acsbr16-02.pdf
United States Census Bureau FactFinder. Gini
coefficient [Data file] . Available from https://factfinder.census.gov/faces/tableservices/jsf/pages/productview.xhtml?pid=ACS_10_1YR_B19083&prodType=table
