c
Racial Concentration
and the Continuation of Poverty in Massachusetts
Melissa Gray
R. Schwartz
QR I101
12/20/02
According to the sociologist William Wilson, “the concentration of black poverty” is a result of the “alignment of economic and demographic forces”[1]. Wilson’s theory argues that the continuing economic conditions of a community are linked to the concentration of blacks in that population. In opposition, one could argue that the economic and demographic conditions of communities in a given area are evenly distributed and that these conditions have no relationship with the racial composition of their respective populations.
Two economic conditions that are commonly associated with poverty are the level of median income and the percentage of households that are renters in a given community, or census tract, the geographical unit of measurement in the Census 2000 data which will be used in the subsequent investigation of variable associations. Low median income contributes to the perpetuation of impoverished communities because residents lack the funds for the upkeep and improvement of their property, thus making the area unattractive to prospective residents with higher incomes that might boost the community average. Likewise, rented houses, which typically are not as well maintained as owner-occupied houses, also may turn away prospective residents with higher incomes. As with low incomes in general, the large presence of rental unites would keep a community from lowering its overall poverty rate. nq but there are exceptions, right? “the village” in nyc, for example. To make the analysis of data that follows easier to read, the variables for the black percentage of a population, the median income and the percentage of renting households have each been recoded into “Low”, “Medium” and “High”.
With
respect to the recoding of the levels of black populations, the histogram at
right (Table 1a) shows the distribution of census tracts and the percent black
population heavily skewed
the right. Therefore, it makes sense to break up the
clusters shown on the histogram into categories that break up the clusters of
tract awkdistributions
along the histogram so no one categories will contain too many tracts that no
comparison can be made. good reasoning
For this analysis, the low percent black category includes all census tracks
with 10% or fewer black populations, while the medium category includes tracts
with 10 to 20% (around the state average of 12.5%)[2]
and the high percent includes tracts with 20% or higher.
The
reasoning behind recoding the variable for median income was slightly
different. In order to ensure for
analysis that sufficient numbers of tracts with low, medium or high percent
black populations would be paired with low, medium and high levels of median
income, the categories were not made according to the state average. The bulk of the census tracts reported in
Massachusetts in 1999 reported average incomes between $40,000 and $70,000 (see
histogram at right, Table 2a). However,
as the purpose of this study is to analyze tracts whose residents live in lower
income communities, the “medium” for the median income variable was dropped to
include census tracts with average incomes between $30,000 and $40,000, while
“low” included everything below (lowest to $30,000) and the “high”, everything
above ($40,000 and higher). Good explanation. could add that the distribution being skewed to the left means
that the mean is pulled higher by the extremes; hence, as is typical practice
when reporting income, the median is used as the measure of center.
The rational for recoding the variable for the number of renter-occupied housing units in Census 2000 tracts ismuch the same. It was already stated that the tracts with high levels of poverty were the focus of this study. One common characteristic of living conditions for people living on lower incomes is that they are rented. As seen in the histogram at right (Table 3a), the majority of census tracts have between one and two thousand renters. Therefore, the clusters of census tracts reporting high numbers of renters are split up so they could be analyzed more closely. The “low” category includes census tracts with 1,000 or fewer renters, whereas the “medium” includes tracts with 1,000 to 2,000 renters and the “high”, 2,000 renters or above. the better variable would be the percent of rental units out of total number of occupied housing units, which is easy to construct in spss. this is the variable you refer to at the beginning of your paper.
The relationship between the three variables (median income, renter-occupied housing and percent black population) will be investigated using two crosstab [computerese for crosstabulation] contingency or two two-way tables. The first and second relationships addressed in this investigation are similar to Wilson’s theory mentioned earlier.
Percent Black and Median Family Income
Alternative Hypothesis:
On average, there is an inverse relationship between the percent black population and the level of median income in Massachusetts census tracts, as determined from the data provided by Census 2000. In other words, overall, as the percentage black population increases in Massachusetts Census 2000 tracts, the median income decreases. ws
Null Hypothesis:
There is no association between the percent black population in Massachusetts Census 2000 tracts and the level of median income for those tracts. All possible patterns that may be observed are purely the product of chance and are not the result of any external force.
Table 4a
* All data from Census 2000 website
Following the recoding of the variables, “Low” median incomes are any that are below $30,000 and, according to the above table, only %8.2 percent of all Massachusetts census tracts fall into this category, compared with %81 percent of tracts that are included in the highest category, with households earning an average of over $40,000 (see above table, Table 4a). However, this is not the case for Census 2000 tracts with high levels of black populations. good comparison Overall, as the percent black population decreases, the median family income increases. Within the 96 census tracts, while the number of census tracts with high black populations are fairly evenly distributed (32, 33, 34% for low, medium, high family incomes, respectively) in all levels of family income.uc However, for census tracts with low levels of black populations, only 5% have median family incomes below $30,000, compared to 87% which have incomes above $40,000. okay: the best way to make comparisons here is not down the columns but across the rows. Rule of thumb, make comparisons along the axis that is perpendicular to the direction of the calculated percentages, e.g., percentage down the columns, compare across the rows. Whereas 34 percent of the 96 tracts with high percentages of black populations had high incomes, 87 percent of the tracts with low black populations had high incomes.
Table 5a
Consider
the table at right (Table 5a). According
to the chi-square tests, the likelihood that the calculated value of 223.259
for the observed chi-square value, compared with the expected chi-square value
of 9.488, with 4 degrees of freedom at the .05 level happened by chance is less
than one out of one thousand. The high unlikelihood that this pattern happened
by chance justifies the rejection of the null hypothesis. Therefore, it is probable that there is
an association between the percent black population in Massachusetts Census
2000 tracts and the level of median income for those tracts. However, the research hypothesis that states
that in general, there is an inverse relationship between the percent black
population and the level of median income in Massachusetts census tracts. In other words, as the percentage black
population increases, the median income decreases. With the elimination of the null hypothesis, the alternative
hypothesis still stands. provides a better
explanation for the pattern.
Percent Black and Renter-Occupied Housing
Alternative Hypothesis:
On average, there is a direct relationship between the percent black population and the percentage of renters in Massachusetts census tracts, as determined from the data provided by Census 2000. In other words, overall, as the percentage black population increases in Massachusetts Census 2000 tracts, the percentage of renters also increases. ws
The second null hypothesis is as follows:
There is no association between the percent black population in Massachusetts Census 2000 tracts and the percentage of household that rent their home for those tracts. All possible patterns that may be observed are purely the product of chance and are not the result of any outside influence.
Table 6a
* All data from Census 2000 website
According to the table above, 1052, or 78% of all Massachusetts census tracts have “Low” levels of renter-occupied housing. As with the levels of median family income, this trend is not the same for all tracts. As the percentage black population increases, the percentage of census tracts with low levels of renter-occupied housing decreases. inac In census tracts with low black populations, 80% have less than 1,000 reports of renter-occupied housing. This statistic is nearer the normal than for the census tracts with medium or high percent black populations. In census tracts with medium levels of black populations, 60% have low levels of renter-occupied housing and of high percent black population tracts, almost 62% fall in the lowest category of renter-occupied housing. Both these values are below the state average.
Table 7a
If one considers the above table (Table 7a), it is clear that there is a large difference in the expected chi-square calculation, 9.488, and the observed, 40.707. Though the difference in the observed and expected chi-square statistics is less than the preceding case, with 4 degrees of freedom and at the .05 level, the likelihood that the patterns observed in Table 6a occurred by chance is less than one in one thousand. As a result, the null hypothesis is abandoned and it can be concluded that there is a probable association between the percent black population in Massachusetts Census 2000 tracts and the percentage of household that rent their home for those tracts. Again, as before, with the elimination of the null hypothesis, the research hypothesis still stands. there is a problem that should be addressed, namely that the more than 20 percent of the cells have expected values of less than 5 so the chi-square statistic is subject to error and uncertainty. the alternative would be to reassess the construction of the variable renter occupied housing: note that only 27 tracts fall into the high renter category.
According to the calculation of the chi-squares, both null hypotheses are rejected. In both cases, it is plausible that the percent black population was related to other economic conditions, namely the level of median family income and the number of renter-occupied houses in a Massachusetts 2000 Census tract. Also, with the rejection of the null hypotheses, each research hypotheses remained standing. As theorized by William Wilson, it is likely that there is a link between the percent black in a community and other economic circumstances such as those mentioned above. Nevertheless, there are more conditions characteristic of impoverished communities that should be looked into. As seen the case involving the number of renter-occupied households and the level of black population, the pattern was not consistent. There was not a steady decrease in the number of tracts with low levels of renters as the black population became higher. good: here you correct a mistaken description above, but the first statement should be corrected as well. It is possible that the same inconsistencies might be found when comparing the black population and other factors associated with poorer communities. Though this study has made the some aspects of racial segregation more clear, there are still some unexplained fluctuations, as evidenced by those in the patterns of renter-occupied households and black populations. consider boston and cambridge with very high levels of renter units in non-black areas. Also, to further support the claim that low median income and high numbers of renter-occupied housing units contribute to the perpetuation of impoverished conditions of communities, similar data from earlier time periods should be analyzed to see if the same trends exist. If such conditions do exist it would supporting the theory that “racial segregation- and its characteristic institutional form, the black ghetto- are the key structural factors responsible for the perpetuation of black poverty in the Unites States.”[3] In other words, communities with high levels of black population are kept in poverty over time because such factors as high levels of renter-occupied housing units and low median incomes, which continue to persist.
Work Cited
Denton, Nancy, and Douglass Massey. American Apartheid: Segregation and the Making of the Underclass. Cambridge, Massachusetts: Harvard University Press, 1993.
“Quick Facts.” US Census Bureau. 2000 Census. Dec. 12, 2002. <http://quickfacts.census.gov/qfd/states/25000.html>
QR Fall 2002 Evaluation of Paper 3
Addresses and tests 2 hypotheses related to one of the following:
1.Residential concentration is higher among African-Americans than Hispanics in MA. Non-white Hispanics are the most concentrated group of all.
2.Minority members tend to move out of poorer and more segregated tracts if they can i.e., if they can afford to. This idea is supported by a large literature in immigrant communities in the United States where the pattern has been describes as “up and out”.
3.More recent social arrivals, e.g., immigrants are more likely to live in more racially/ethnically concentrated communities than community members of long-standing.
4. Social instability is an intervening factor between racial/ethnic concentration and negative social outcomes.
|
Introduction. · sets research context
|
Yes, with Wilson as quoted in Massey and Denton: Massey and Denton go beyond Wilson to talk about the links between concentrations of racial minorities, poverty, and instability as worsening. |
|
· foreshadows argument or thesis |
not really |
|
Description of data and variables.
|
The reasoning behind the recoding is thoughtful and clearly stated, as well as being attentive to the distributions of the variables in question. There is some inconsistency between your original statement about the percentage of rental units and then your use of the number of rental units; the former is a more appropriate variable to get at the rate of rental units. The cut points for rental occupied units becomes the source of a later problem when so few (27) tracts fall into the “high” area. This contributes to the problems with the chi-square statistic that I describe in the text. |
|
Formal statement of the Research and Null Hypothesis. #1 |
Well stated |
Analysis #1· discuss patterns in the table |
This is solid and makes the point. Making comparisons across the rows would be an improvement. The use of the marginal percentages as a baseline is excellent. |
|
· interpret the chi-square test. |
good |
Hypothesis #2
|
Well stated |
|
Analysis #2 · discuss patterns in the table |
your reading here is flawed by a misstatement about the pattern |
|
· interpret the chi-square test. |
okay but the problem with the chi-square statistic should be addressed and results qualified accordingly. We went over this in the third lab. |
|
Summary and Conclusion. · assess your hypotheses in light of the data analysis · conclusions or implications? · further possibilities for analysis? |
You make a more accurate characterization of the problematic renter occupied unit issue and your qualifications are in order. Your link with Wilson is restate, which is good, though the relevance of Massey and Denton remains an opportunity not as well taken up as it could be. |
Presentation /Writing· claims supported by evidence and reasoning |
yes |
· Organization coherent; logical |
yes |
· Language specific |
yes |
· Focused paragraphs |
yes |
· Proofread for errors |
some remain |
· Tables and graphs well labeled |
well presented |
Key: GP: good point; GE: good evidence; WS: well stated; GQ: good qualification
CWE: claim with little no pertinent evidence; NQ: needs qualification; E?: evidence for this seems lacking; Elab: could be improved by elaborating; INAC: inaccurate; Highlight: something problematic
UC: unclear statement, difficult to follow/understand; V: vague, needs to be more specific or concrete; AWK: awkward style; P: punctuation error; G: grammar error; Underscore: typo or misspelling
B+