Jessica Russell
Quantitative Reasoning
Professor Schwartz
20 Dec 2002
Race and Education in Massachusetts
Introduction
Despite perceptions otherwise, segregation is still reproduced in the United Stated, and more locally, segregation surrounds us, in regions of Massachusetts. With maps we can see that racial and ethnic groups are clearly not uniformly spread throughout the state. Rather, in Maps 1 and 3, we can see that in Massachusetts there are pockets of tracts where minority groups are concentrated and over represented, while being underrepresented throughout the rest of the state. A look at the racial and ethnic distribution of Hampden County reveals the acute differences of racial composition between neighborhoods. Map 2 shows the concentration of African Americans in the tracts of Hampden County. The highest concentrations are grouped in the north Springfield area. Map 4 shows the concentration of Hispanics in the tracts of Hampden County, where the highest concentrations are west and north of Springfield, and in Holyoke. As Prof. Smith asserted in his 1995 paper drawing on data from the 1990 census, “If you look at the distribution of racial groups in the region, it is clear that Springfield and Holyoke contain a disproportionate number of blacks and Puerto Ricans.” (1995, 2)
The implications of such disproportions are still disputed. However, many agree that sociologically, racial concentration has been linked to adverse social outcomes. In their book, American Apartheid, Massey and Denton argue that,
“If racial segregation concentrates poverty in space, it also focuses and amplifies any change in the economic situation of blacks…In a racially segregated city, any increase in black poverty is confined to a small number of black neighborhoods; and the greater the segregation, the smaller the number of neighborhoods absorbing the shock and the more severe the resulting concentration of poverty. If neighborhoods are also segregated by class, not only is the additional poverty restricted to black neighborhoods, it is confined primarily to poor black neighborhoods.” (1993, 126)
By looking at charts A and B, we can see that the concentration of race in tracts is reflected in school enrollment: in 2000, Springfield schools reported that 44% of their students were Hispanic, 29% of students were African American, and 25% were non-Hispanic Caucasian.
Massey and Denton argue that “segregation also concentrates educational disadvantage” (1993, 141). Drawing on their findings, I venture that such concentrations of race and poverty also affect the level of school achievement among the residents. I will look for patterns of educational achievement in tracts with high minority concentration. Keeping in mind the segregation of the Western Massachusetts cities of Springfield and Holyoke within Hampton County, I will analyze Massachusetts school achievement data within the African American and Hispanic communities, as those groups are most over represented in the county.
Data & Variable Description
I will analyze at the level of tracts, using data drawn from the U.S. Census 2000 for Massachusetts. I will use crosstabs and chi-square tests in SPSS software to analyze information from the data set for Massachusetts, consisting of 1356 tracts.
My variables will include percent black of population, percent Hispanic of the population, and the educational attainment data. (This data includes: number of people with less than 9th grade education, 9th to 12th grade education, high school education, some college education, Associate’s Degree, Bachelor’s Degree, and Graduate’s Degree per tract.)
According to U.S. Census definitions, “’Black or African American’ refers to people having origins in any of the Black racial groups of Africa. It includes people who indicated their race or races as ‘Black, African Am., or Negro’ or wrote in entries such as African American, Afro American, Nigerian, or Haitian.” Those identifying as Hispanic, responded positively as “Spanish, Hispanic, or Latino.” (2001, 1-2) The variables show the frequency with which people identified with these and other groups when questioned by Census 2000.
Recoding of the variables was necessary in order to work with the data manageably in crosstabulation. I chose breaks to give meaningful representation to each category (Low, Medium, and High). After examining the histogram, I spread the range for useful distribution.
My new categories for race and ethnicity are as follows:
For percent black of the population: 0-5% is Low, 5-15% is Medium, and 15% and higher is considered High percent black of population.
For percent Hispanic of population: 0-5% is Low, 5-20% is Medium, and 20% and higher is considered High percent Hispanic of population.
To recode the educational achievement variables, it was necessary to first create a mean with all of them, which could then be recoded into manageable figures. For the purpose of my investigation, I recoded the mean into the following categories: some high school or less, high school, some college, and Associates degree or higher.
Hypotheses
To examine this theory I will test two sets of hypotheses. The Research Hypothesis for Table I is: I hypothesize that there is an inverse relationship between the concentration of minorities in tracts and the education levels achieved within those tracts. That is, I expect to find that tracts with a higher proportion of African Americans report lower levels of education achieved.
The Null Hypothesis for Table I is: There is no association between the proportion of African Americans and the education levels in MA census tracts in 2000; any observed difference in education levels between high and low minority tracts is due to chance.
The Research Hypothesis for Table II is: I hypothesize that there is an inverse relationship between the minority composition of a census tract and the levels of education achieved in that tract. That is, I expect to find that tracts with a higher proportion of Hispanics to report lower levels of education achieved.
The Null Hypothesis for Table II is: There is no association between the proportion of Hispanics and the level of education achieved in MA census tracts in 2000; any observed difference in educational achievement between high and low minority tracts is due to chance.
Analysis
By using a chi-square test, we find out how unlikely it is that the null hypothesis is true. The greater the difference between the calculated value and the expected value (as calculated by the chi-square) the less likely it is that the pattern we observe has occurred by chance. The probability number calculated by the chi-square is a tool for rejecting or accepting a null hypothesis. For instance, a chi-square with a probability of .01 indicates that if the null hypothesis were correct, there is a 1 in 100 chance of calculating (within the same dimensions/ Degrees of Freedom) a chi-square value as large or larger than the one in question. For my analysis I have set the level of statistical significance at .05. Thus any chi-square probability calculated as less than .05, would be sufficient ground for rejecting the null hypothesis, in which case, the research hypothesis would stand.
Each of the computed chi-square tables comparing racial background and achieved education has six degrees of freedom. Thus, using a “Distribution of Chi-square” table, I have determined that at .05 significance level, a chi-square value larger than 12.592 will be sufficient to rule out chance as a cause for the patterns observed.
Table Descriptions
In Table I, a High percent African American population is defined as 15 percent or more African American population in a tract, and according to this classification, nearly 10 percent of all MA tracts reported a high rate of African American residents to the census in 2000. Here a Low percent African American population is defined as 5 percent or less African American residence in a tract and this describes more than 75 percent of MA tracts as measured by Census 2000. The crosstabulation calculated the correspondences between Low, Medium, and High percent African American percent population in tracts, and the meaned education achievement as recoded above. Table I shows a generally negative relationship between the proportion of African Americans in tracts in Massachusetts and the mean levels of educational achievement across those tracts in 2000. That is, as the proportion of African Americans increases, the levels of achieved education decrease.
Comparing tracts with meaned education levels at “Some College,” we can see that nearly 50 percent of tracts reporting a low African American residence had a meaned educational achievement of some college, while just over 15 percent of high percentage African American tracts reported the same meaned achievement; less than half as much. Just 4.6 percent of high percent African American tracts reported a mean completion of an associate’s degree or more, while 11 percent of low African American tracts fell into the same category; again that is more than twice as much. However, while 35.5 percent of low African American tracts show a mean education level at high school completion, over 51 percent of high African American tracts display the same mean. Furthermore, while only 4.1 percent of low African American tracts fall into the category of mean academic achievement at some high school, 28.5 percent of high African American tracts fall into that category.
The observed value of the chi-square test corresponding to Table I is 170.975. This number greatly exceeds the expected test statistic of 12.592 for a table with 6 degrees of freedom at .05 level. Moreover, the probability associated with attaining a calculated chi-square statistic so high is less than one thousandth of one percent.
In Table II a High percent Hispanic population is defined as 20 percent or higher rate of Hispanic residence in a tract, and according to this classification over 10 percent of all tracts in Massachusetts reported a rate of high Hispanic residence. Low Hispanic residence is defined as a 5 percent or lower rate of Hispanic residence in a tract, and 67 percent of Massachusetts tracts fell into this category. The crosstabulation calculated the correspondences between Low, Medium, and High percent Hispanic population in tracts, and the meaned educational achievement corresponding to the tracts. Table II shows a generally negative relationship between the proportion of Hispanic residence in tracts in Massachusetts and the mean levels of educational achievement across those tracts in 2000. So, as the proportion of Hispanics increases, the levels of achieved education decrease.
Comparing tracts with meaned education at high school completion, we find that more than 40 percent of tracts with high Hispanic residence are counted, while 55.6 percent of tracts with medium -- that is, 5-20 percent Hispanic residence— Hispanic populations are counted and 31.9 percent of low Hispanic tracts fall into the same category. However, while 55.4 percent of low Hispanic tracts reported a meaned education achievement of some college, 4.8 percent of high Hispanic tracts reported the same. Furthermore, while 11.7 percent of high Hispanic tracts reported meaned education levels at and associate degree or more, only 0.7 percent of high Hispanic tracts show the same. Finally, 1 percent of low Hispanic tracts fell into the category of some high school as the meaned educational achievement, while 53.8 percent of high Hispanic tracts fell into the same category.
The observed test statistic for the chi-square corresponding with table II is 573.375. This value is also much higher than the expected test statistic of 12.592 for a table with 6 degrees of freedom at the .05 level. Again, the probability associated with attaining such a high observed chi-square statistic is less than one thousandth of one percent.
Summary & Conclusion
I assert that my research hypotheses stand in light of the analysis. I have found compelling evidence of inverse relationships based on Census 2000 data, which have been supported by chi-square assessments ruling out chance as a reason for the observed patterns. For these reasons, I reject the null hypothesis that there is no association between the proportion of African Americans and the education levels in MA census tracts in 2000. Instead, I find a statistically significant relationship between the concentration of African Americans in tracts and the education levels achieved within those tracts; the larger the proportion of African Americans, the lower the reported levels of education achieved are likely to be in that tract.
Furthermore, I reject the null hypothesis of no association between the proportion of Hispanics and the level of education achieved in MA census tracts in 2000. Instead, I find a statistically significant relationship between Hispanic concentration and education achieved at the tract level; the higher the proportion of Hispanic residents, the less likely it is that high levels of educational achievement will be reported within that tract.
I find it interesting that the crosstabulation of Percent Hispanic of Population and Education yielded results more dramatically compelling than the crosstabulation of Percent Black of Population. As my sources suggest that concentration may be stronger among the Hispanic communities than among African American communities; a possibility for further analysis would be to examine the relationship between magnitude of concentration and educational achievement. My findings reveal a strong probability that segregation adversely affects the education process. The reasons and implications of this statement seem overwhelming. Smith argues that this segregation is attributable to institutionalized and socially ignored racism. In agreement, Massey and Denton go further in a social argument to say that “…because poverty is associated with poor educational performance segregation also concentrates educational disadvantage…By concentrating low-achieving students in certain schools, segregation creates a social context within which poor performance is standard and low expectations predominate.” (1993, 141) A sociological study draws forth many questions. I worry about the present accessibility to multi-language schooling, the absence of which may alienate and discourage the children with less English skills to sail far in a potentially hostile environment. Most of all, I worry that the perceptions and stereotypes that draw their roots from such injustice will close more doors for those already disadvantaged. This is why analyses such as this are gravely important as tools for opening up communication and questions about the causes of the conditions we as people experience.
Works Cited
Grieco, Elizabeth M. and Rachel C. Cassidy. 2001. Overview of Race and Hispanic Origin. Census 2000 Brief. Washington DC: US Census Bureau.
Massey, Douglas S. and Nancy A. Denton. 1993. “The Missing Link” and “The Creation of Underclass Communities”. Pp. 9-16 and 115-142 in American
Apartheid. Segregation and the Making of the Underclass. Cambridge, MA: Harvard University Press.
Smith, Preston. 1995. “Hidden History of Racial Segregation in Springfield and Holyoke”. Paper presented to the Regional Approaches to Ending Housing
Segregation Conference. Holyoke, MA, April 1995.
Map
1
Map 2
Map
3
Map 4
Chart A: School enrollment in Springfield in
1990 and 2000.
Chart
B:
School enrollment in Springfield in 2000.
Data from the Lewis Mumford Center[1]
http://mumford1.dyndns.org/cen2000/SchoolPop/SDistSegdata/2511130sd.htm
Table I
Chi-Square Test for Table I
Table II
Chi-Square Test for Table II
|
|
|
|
|
|
|