Elissa Barratt

20 Dec, 2002

QR 3 Final Paper

Hispanic Concentration in Massachusetts

Race is one of the most discussed issues of our time. Our race and our ideas about race dictate much of our mental and physical lives. We have been physically divided into several different racial groups. This segregation began “during and after the Great Migration of southern blacks in the 1920’s” (Smith 4). This led to the creation of the ghettos. The ghettos greatly affected the black community, which remains segregated today. However, in Massachusetts, the Hispanic (and largely Puerto Rican) population is also ethnically concentrated. Unlike the African-Americans, most of the Hispanic residents did not migrate to Massachusetts until the 1970s and 80s (Smith 12). However, their populations also suffer much of the same concentration. Hispanics make up 30% of Holyoke and 17% of Springfield. Within these cities, 58% of Hispanics are concentrated in three neighborhoods of Springfield, and in Holyoke, 74% of Hispanics reside in four neighborhoods (Smith 2). These are clear signs of racial concentration among Hispanics. But this does not only exist at the city level. One-third of all Massachusetts residents who identified themselves as Hispanic on the 2000 Census live in Springfield, Boston, and Holyoke (this is statistic was calculated in table 1.0 which will be explained later).

This ethnic concentration negatively affects their residents in concentrated areas. Income for Hispanics tend to be lower than that for whites. Many Hispanics came to Massachusetts seeking employment in agriculture and manufacturing, traditionally low paying jobs. According to Massey and Denton, when this population becomes very concentrated, the poverty experienced within this community also becomes very concentrated. The poverty rates for that neighborhood are higher because it has more people with lower incomes than other areas. This region then becomes considered one of high poverty, and begins to shown signs of “social problems associated with income deprivation (e.g. crime, housing abandonment, . . . etc.)”(122). Therefore, areas of high ethnic concentration are associated with low incomes, and signs of instability.

In order to see if this hypothesis has some basis, many statistics must be collected. Theoretically, the areas of concentration would show signs of inequality, such as a low median family income and signs of instability, such as the amount of renter occupied housing. The data used for this was taken from the 2000 United States Census. The Census used a sample of one in six households in the United States to collect this data. The data is collected by census tract, making the unit of analysis the census tracts and not the residents themselves. The ethnic concentration of Hispanics is being studied, in Massachusetts, where 6.8% of the population reported that they were Hispanic or Latino, opposed to 12.5% nationally (US Census Bureau). This Hispanic population is concentrated within the state. To do this the percentages of Hispanics in three different cities was calculated out of the state total as can be seen in Table 1. These cities are known for their high Hispanic population.

However, the percent Hispanic within a census tract are being studied. As can be seen in Graph 1.0, the distribution of Hispanics in Massachusetts census tracts is skewed. That is, there are many census tracts with very low levels of Hispanics and very few with high levels of Hispanics. Therefore, the varying levels of Hispanics had to be recoded (separated) into different groups, partially to make the distribution somewhat more even. The percentages have been recoded into three different levels, this was done to allow for an easier comparison in the charts, the characteristics of the three levels are shown, and not that of each different percentage. This recoded distribution can be seen in Graph 1.1. Moreover, these levels help give designations of what is meant by ethnic concentration. When a tract falls into the high category, it will be considered an ethnically concentrated tract. The definitions for each of the different levels were chosen based on the distribution of Hispanics in census tracts. The new distribution was created so that there would be enough values at each different level to do calculations with. The percentages for low (0-5%), medium (5-20%), and high (+20%) are slightly lower than one might expect. This was done to compensate for the skewed graph, and allow more tracts to fall into the medium and high categories.

The reported median family income is also used as a variable. Median family income was used because it is an indication of the social class of a given census tract. If many high Hispanics tracts have low median incomes it can be a sign of social inequality. In Graph 2.0, the distribution for median family income is somewhat evenly distributed. Nevertheless, this variable also needed to be recoded. This again was done for clarity and the purpose of calculations and comparisons. The median family income was put into three groups, in order to shows the varying levels of income while not showing every possible income. The incomes were grouped into low medium and high categories this distribution can be seen in Graph 2.1. Low was determined to be $0-$40,000, medium to be $40,000-$70,000 and high to be above $70,000. These distributions can easily be seen in Graph 2.0 as natural breaking points. They also allow enough tracts to fall into each category that good calculations and comparisons can be done.

The amount of renter occupied housing was also used as an indication of the social status and stability of a tract. Tracts with higher levels of owner occupied housing are considered to be more wealthy and more stable. Low rates of home ownership could be considered a negative social outcome which arose from racial inequality. The distribution of renter occupied housing in Massachusetts is also somewhat skewed. In Graph 3.0, many tracts have low levels of renter occupied housing. Again this variable was recoded into three categories. This was also done to aid in calculations and establish what low, medium, and high levels of renter occupied housing are. These classes aid the analysis of varying levels of renter occupied housing, and simplify the distribution. Low is considered here to be 0-900 renter occupied housing units within a given tract, medium is 900-1750 and high is above 1750. When one looks at Graph 3.0 it is, to some degree, divided at the 1000, 2000, and 3000 houses mark. However, the graph is skewed, so to compensate for this, the levels of low, medium, and high are slightly lower than would seem natural for the graph. The distribution of Graph 3.1 does not really change the distribution and may look very skewed, but it ensures that there will be enough tracts in each category to do useful calculations with. This is very important because if very few tracts fall into one of these classes it is hard to reliably do calculations. This will be very important later on.

For the analysis of this data, graphs and crosstabs were used. Graphing showed the distribution of the different variables in Massachusetts, this establishes what one would expect to see. Crosstabs are also used, this allows the comparison of two variables, to observe how often they occur together. The chi-square value was also used, which will be further discussed.

From what is understood about ethnic concentration and ethnic certain hypothesis could be created. These hypotheses would need to test the relationship between ethnic concentration and signs of social inequality. Since this could never be proved, a null hypothesis was created to show that the opposite does not happen.

On average, Massachusetts Census tracts with high levels of Hispanics, report lower median incomes than tracts with low levels of Hispanics.

The relationship between tracts with high levels of Hispanics and low levels of median family incomes is not significantly different from chance.

On average, Massachusetts Census tracts with high levels of Hispanics, have higher levels of renter occupied housing than tracts with low levels of Hispanics.

The relationship between tracts with high levels of Hispanics and their high levels of renter occupied housing is not significantly different from chance.

The relationship between median family income and ethnic concentration was the first relationship to be tested. If ethnic concentration concentrates poverty then lower median family incomes would be expected in tracts with higher racial concentrations. Furthermore, an area with low median incomes would not attract business and therefore jobs, keeping the median income lower. This would feed in on the tract and perpetuate its low median family incomes. The relationship between tracts with high ethnic concentration and high numbers of renter occupied housing was also studied. As stated above a tract in high poverty will show signs of social instability. Renter occupied housing was chosen here as a sign of social instability. High levels of renter occupied housing can reflect social instability for several reasons. First, these high levels show that not many people, within a given tract, can afford to buy a house. It also could be sign of mobility, if many people are renting it could be showing that people are often moving in and around that tract. These people could be moving or could be renting instead of buying because they do not have a steady income. If people don’t in fact have steady jobs, this could be a sign that they are facing ethnic discrimination in the form of access only to low income, and therefore undesirable jobs. They could also have difficulty obtaining loans to buy a house. This is a fairly large assumption, but high rates of renting are defiantly a sign of lower incomes and an inability to buy housing.

The first relationship that was explored was that between median family income and ethnic concentration. Here, the ethnicity is the dependent variable because it theoretically changes the median family income of that census tract. As one can see in Crosstab 1, of the tracts with high levels of Hispanics, not a single tract reports a high median income. Of tracts with medium levels of Hispanics, 7% have high median family incomes. However, of tracts with low levels of Hispanics 43.5% report high median family incomes. As was said before the cut-offs for median family incomes were slightly lowered to compensate for the few tracts with high levels of Hispanics. However, even this does not create a single high Hispanic, high median family income tract. What is even more dramatic is of the tracts with low numbers of Hispanics, only 4.1% report low median family incomes. Of tracts with medium amounts of Hispanics 30.1% experience low median family incomes. Significantly, 89% of tracts with high levels of Hispanics report low median incomes. This is a staggering difference. It means that few tracts with few Hispanics have low median family incomes and yet the vast majority of tracts with high levels of Hispanics live in areas with low median family incomes. This obviously creates an inverse relationship, were the more Hispanics there are in a tract, the lower the median family income, and vice versa. This can be seen when looking at the overall percentages. Of all Massachusetts, only 19% of those census tracts have low median family incomes. However, of tracts with high levels of Hispanics, 89% have low median family incomes. There are obviously signs of income inequality in those tracts with concentrated levels of Hispanics.

This inequality becomes even clearer when looking at the chi-square statistic on Chi-Square 1. The level of statistical significance selected is 0.5 and there are 4 degrees of freedom. The chi-square statistic here is 704.190, an extremely high chi-square value. The chi-square value shows the difference between the observed and the expected values for a crosstab. The critical value for this is 9.488. There is obviously a large difference between the two. The null hypothesis has to be rejected because the chi-square value is so much larger than the critical value. Basically, the percent chance of the residential and income patterns in Crosstab 1 happening by chance, is less than 1 in 10,000. This means that the residential patterns most likely did not happen by chance. Instead, as much literature would suggest, this is a result of ethnic concentration. Due to the high numbers earning low median family incomes and the low probability of this happening by chance, it can be said that there could be a direct relationship between high ethnic concentration and low incomes.

The second relationship that was theorized is very similar to the first. This explores the relationship between the concentration of Hispanics and renter occupied housing. In Crosstab 2 the findings are more limited and not as dramatic. One of the limitations of studying renter occupied housing is that it is a number and not a percentage. That is it is hard to tell what percentage the renter occupied housing is of its tract. However, there is still a fair amount of difference. Of areas which exhibit low levels of Hispanics, only 2.3% have what is considered high levels of renter occupied housing. Of tracts with high levels of Hispanics, 9% have high levels of renter occupied housing. This is actually a fairly significant, as will be seen in the chi-square statistic. Again, of tracts with low levels of Hispanics, 84.4% have low levels of renter occupied housing. This is the vast majority of those tracts with low percentages of Hispanics. Contrasting to this, of tracts with high numbers of Hispanics, 46.9% report low levels of renter occupied housing. Here the difference between the two can be seen. While the majority of mostly non-Hispanic tracts have only low levels of renters, concentrated Hispanic tracts have mostly medium (44.1%) to high levels of renters. This discrepancy can be better understood once the chi-square statistic is taken into account, as can be seen in Chi-square 2. Again the level of statistical significance was set at 0.5, and there are 4 degrees of freedom on the chart. Here the chi-square statistic is 190.010, a dramatically high value. The critical value here again is 9.488. Again the chi-square value is a great deal larger so the null hypothesis is rejected. Here too, the chance of the crosstab happening as the way it did by chance is less than 1 in 10,000. This means that the statistics found are significant, they show something about that population and its residential trends. These numbers show that this residential pattern did not result from pure chance. There is a reason for this trend. This could possibly be a result of ethnic concentration leading to negative social consequences.

In light of these conclusions, the general idea that ethnic concentration is associated with low incomes and signs of instability, holds some amount of truth. When testing the relationship between concentrated Hispanic tracts and low median family incomes, the trend held. The null hypothesis was able to be rejected because there was a significant difference between what was observed and what was expected. This means that there are definite signs of social inequality in those tracts with high concentrations of Hispanics. The second relationship between social instability and ethnic concentration also gave significant results. Here, there was a positive relationship between the concentration of Hispanics and the amount of renter occupied housing. This is a sign of social instability of a census tract. Again the null hypothesis was rejected. The probability of either of these happening by chance is so small that there is most likely some form of relationship between them. To pursue findings on this further one could look at other signs of social instability or the poverty levels of ethnically concentrated tracts. One could also look at similar scenarios in other states besides Massachusetts and in other census years (although, the definition for Hispanics was unique to the 2000 census). Furthermore, the research only implies that this is an ongoing form of inequality. That is, from others studies on this issue, we can see that ethnic concentration creates inequality. Here there are clear signs of both, which implies that this inequality has resulted from the ethnic concentration. We can also understand that this concentration will feed in on itself isolating the group and allowing few chances for these residents to better their situation. That is to say, we can understand how these statistics are like a snapshot of the ongoing process. But it would be interesting to study this more in depth and view the pattern of concentration and inequality over more time. This would give even more insight into the ongoing relationship between ethnic concentration and ethnic inequality.

Works Cited

Massey, Douglas S. and Nancy A. Denton. American Apartheid: Segregation and the

Making of the Underclass. Harvard University Press, Cambridge University,

1993.

Smith, Preston H. Hidden History of Racial Segregation in Springfield and Holyoke.

United States Census Bureau, United States Department of Commerce.

http://www.census.gov/ Dec 14, 2002.

Table 1

Breakdown of Hispanic Residence in Massachusetts

	Population	Percentage
Total Hispanic in Mass	428,729	100%
Identifying as Puerto Rican	199,207	46.5%
Hispanics in Boston	85,089	19.8%
Hispanics in Springfield	41,343	9.6%
Hispanics in Holyoke	16, 485	3.8%
Total Boston, Springfield, Holyoke	142,917	33.3%

1. The populations used in this table were obtained from the Census Website and were calculated by Elissa Barratt

Level of Hispanics in Massachusetts Census Tracts 2000

Graph 2.1

Graph 3.0

Crosstab 1

Case Processing Summary 1

Tracts

	Cases
Valid	Percent	Missing	Percent	Total	Percent
Census
1356	100.0%	0	.0%	1356	100.0%

Relationship in Massachusetts Census Tracts between percent Hispanic and Median Family Income (Results from 2000 Census)

			Percent Hispanic
		Low 0-5%	Medium 5-20%	High +20%	Total
Low 0-$40,000	Count	37	91	129	257
Median Family Income
	% within level	4.1%	30.1%	89.0%	19.0%
Medium $40,000 - $70,000	Count	477	190	16	683
	% within level	52.5%	62.9%	11.0%	50.4%
High +$70,000	Count	395	21	0	416
	% within level	43.5%	7.0%	0%	30.7%
Total
	Count	909	302	145	1356
	% within level	100.0%	100.0%	100.0%	100.0%

Chi-Square Test 1

	Value	df	Asymp. Sig. (2-sided)
Pearson Chi-Square
704.190	4	.000
Likelihood Ratio
657.210	4	.000
Linear-by-Linear Association
524.502	1	.000
N of Valid Cases
1356

a 0 cells (.0%) have expected count less than 5. The minimum expected count is 27.48.

Crosstab 2

Case Processing Summary 2

Tracts

	Cases
Valid	Percent	Missing	Percent	Total	Percent
Census
1356	100.0%	0	.0%	1356	100.0%

Relationship in Massachusetts Census Tracts between percent Hispanic and total renter occupied housing (Results from 2000 Census)

Renter

			Percent Hispanic
		Low 0-5%	Medium 5-20%	High +20%	Total
Total
Low 0 – 900	Count	767	151	68	986
Occupied Housing
	% within level	84.4%	50.0%	46.9%	72.7%
Medium 900- 1750	Count	121	129	64	314
	% within level	13.3%	42.7%	44.1%	23.2%
High +1750	Count	21	22	13	56
	% within level	2.3%	7.3%	9.0%	4.1%
Total
	Count	909	302	145	1356
	% within level	100.0%	100.0%	100.0%	100.0%

Chi-Square Tests 2

	Value	df	Asymp. Sig. (2-sided)
Pearson Chi-Square
190.010	4	.000
Likelihood Ratio
182.761	4	.000
Linear-by-Linear Association
149.980	1	.000
N of Valid Cases
1356

a 0 cells (.0%) have expected count less than 5. The minimum expected count is 5.99.