Correlations

After studying local newspapers and data from the city’s health department, it was evident that the AIDS rates in Philadelphia indicate a problem on both a city and national level. In order to address the AIDS problem, one first needs to determine where the problem is, both demographically and graphically. To uncover possible areas, correlations were drawn between relevant information from the detailed chart data of 2000 U.S. Census (specific charts are listed in the appendix) and the rates of AIDS for all of the Philadelphia ZIP codes. The resulting correlations suggest that AIDS has created a problem within two major demographics: the poor and African Americans. Although these target demographics seem to comprise of an overwhelming constituency, the isolation of ZIP codes can help narrow the focus of prevention.

The method for ascertaining correlations was as follows: in an Excel spreadsheet, AIDS rates of Philadelphia ZIP codes were plotted against several components of census data that seemed like they might provide a successful correlation. The success of a correlation was contingent upon the number of ZIP codes. The correlations listed as successful have p value of >.05, or <95% confidence interval. To achieve this value, my correlation coefficients must exceed the number .273 which corresponds to 50 df, degrees of freedom (a value of 1 displays a perfect correlation). In order to find significant correlations, two ZIP codes needed to be removed, 19112 and 19113, which were a Naval Base and an airport, respectively. These ZIP codes created disproportionately high rates because their populations were so small; therefore, the data from these ZIP codes made it impossible to correlate material. After removing the ZIP codes, not all of the correlations succeeded in reaching a value of .273, but there were several correlations that proved significant. Table A indicates the correlation values for all of the data collected:

The most important correlations suggest there is an AIDS problem among the poor population. Poverty data seemed important because of the media attention to increased poverty in Philadelphia, and also because many newspaper articles focused on the relation between AIDS and poverty. The second strongest correlation, with a value of .662, compares AIDS rates to levels of extreme poverty. The “poverty line” varies for each family because it compares two economic components. The U.S. Census Bureau states, “if a family’s total income is less than the family’s threshold, then that family, and every individual in it, is considered poor.”[2]  The data measured the ratio of family income to poverty level. The numbers of people with a ratio of less than .5, the most impoverished, had the strongest correlation with the AIDS data. Table 1 and Chart 1 indicate the findings. Chart 1 indicates the strong positive correlation between extreme poverty and AIDS rates. ZIP codes like 19133, North East Philadelphia, and 19121, North Fairmont, display both high poverty and high AIDS rates. The few outliers, 19107 and 19123, display relatively low poverty levels but high AIDS rates. This area is the “gayborhood” of which John Bell, a member of Philadelphia’s Project TEACH, refers to in an article in The Philadelphia Inquirer.[3]  In fact, the correlations on Chart 2 show that the gay neighborhood extends into ZIP codes 19147 and 19102 because these areas have high percentages of male-male households (aside from their proximity to the other gay ZIP codes on Map 2). Even within these ZIP codes, the relation between poverty and AIDS is evident. ZIP code 19107, with the highest poverty level, also has the highest AIDS rates. Therefore, the relationship of poverty to AIDS rates cannot be ignored. Map 1 indicates that the poorest areas are located just above the center city area (indicated by the “Art Museum” label). The affluence of ZIP codes increase as the distances from center city increase. The high affluence areas are important because they are inversely related to the AIDS rates on Map 1. The other correlations demonstrate the specific relationship of poverty and AIDS. In fact, my most successful correlation examines the median number of rooms in a household. The strong negative correlation (of -.665) in Chart 3 indicates that the fewer number of rooms help predict a greater AIDS rate. It is likely that a person who is poor is unable to afford a home with a great number of rooms. Therefore, this data helps strengthen the case for a relationship between poverty and AIDS. Similarly, the percent of vacant houses has a strong positive correlation (of .64), possibly because people in more impoverished areas do not have the economic wealth to occupy or own houses (a vacant house is one that is for rent, for sale, or unoccupied).[4] Map 4 shows that the ZIP codes closest to center city, 1921 (N. Fairmont) and 19133 (North Philadelphia) are once again implicated in data related to high poverty. I used median household income as another indicator of poverty. In Chart 5, the strong negative correlation (of -.45) indicates that lower income may predict higher AIDS rates. Once again, the problem is focused in areas north of center city (Map 5). One data set that related to poverty, average family size, did not have a significant correlation. Although it would seem that families with greater poverty would have greater numbers of children, this factor did not relate to AIDS rates (with a low correlation value of -.13)

                                     .

Race was the second most important characteristic that was studied. The data was heavily focused on African Americans, because research of the media implicated this demographic in the study of AIDS in Philadelphia. The percent of Black population alone correlated significantly with AIDS rates with a value of .287 (Table 6). This correlation is startling because is based on race alone and does not take into account socioeconomic factors. However, the combination of poverty and race provide even stronger correlations. Chart 7 compares the percent of Blacks with income below the poverty level to achieve a significant correlation of .614. There are several interesting observations in this chart. First, it shows that ZIP code 19107 is not only a gay neighborhood, but it also has a significant poor, Black population. This data is dangerous because it indicates a triple threat in this location. Secondly, there is significant overlap of ZIP codes that have high percentage of poor blacks with ZIP codes that have a high percentage of overall poverty. For example, ZIP codes 1921 (N. Fairmont) and 19133 (North Philadelphia), once again correspond strongly with AIDS rates. This overlap suggests that the poverty problem in Philadelphia is namely a Black poverty problem. The poor Black constituency has a particular bearing on AIDS rates when compared to the correlation of poor whites. The correlation coefficient of Whites with income below the poverty level, .464, is indeed significant but not nearly as high as the correlation of Blacks with income below the poverty level, .614 (Chart 8). Another important correlation is between median household income of Blacks and AIDS. Black income values add important information because it indicates new ZIP codes, specifically 19146 (Schuylkill) and 19132 (NE Philadelphia), which correlate well with AIDS data (Chart 9). These zip codes do not appear to be as significant in the charts on poverty overall. Additionally, Schuylkill is located outside the center city area (Map 9). The distinctiveness of these ZIP codes may signify areas in which the problem of poverty is exclusively associated with race. I attempted to correlate the median age of Black females, because my research my research suggested that Black females were at the highest risk of new HIV cases. However, the correlation did not prove significant (possibly because the AIDS rates are based on cumulative data).

Although attention was given to the Black population, data from Hispanic populations was also collected in order to see if there is a problem within that community. The correlations for Hispanic populations are not nearly as strong as those of the Black population. In fact, the percent of the population Hispanic was not even a significant correlation (with a value of .07). However, the percentage of Hispanics living under the poverty level, Chart 10, indicates a significant correlation of .319. In fact, this correlation is visually not as convincing as the previous ones. Some ZIP codes have very high levels of Hispanic poverty, such as1921 (N. Fairmont) and 19133 (North Philadelphia), and others have very low levels of Hispanic poverty, such as 19106 and 19102, and yet these four ZIP codes have similarly high AIDS rates. Therefore, the most powerful determinant of AIDS in Philadelphia may not be race but rather poverty (or a combination of the two).

Although these sets of correlations do not prove that poverty, or race, causes AIDS in Philadelphia, they certainly indicate a relationship among these factors. Identifying what relationships are at the center of this AIDS problem are essential in knowing how to solving it. The correlations reveal that efforts in curbing the disease should focus on informing groups that are poor or Black. These two demographics are most highly correlated with AIDS rates, and the particular problem seems to be closest to center city. With this data, addressing the problem in Philadelphia is not only feasible but imperative.

                                                                                    
 

---

[1] All material from www.census.gov. The tables are indicated in the parenthesis.

[2] www.census.gov

[3] McCullough, Marie. A Philadelphia program illustrates further progress will be hard-won. The Philadelphia Inquirer. June 5, 2006. E01.

[4] www.census.gov. Table H5.