Dutchess County: Statistical AIDS Evidence
As a follow up to my first paper, I decided to do a correlation analysis of the AIDS rates per 100,000 of the zip codes in Dutchess County with demographic factors that seemed to have relations with AIDS rates. After analyzing the correlation data, I was able to see the actual relations between demographic factors and with the AIDS rates per 100,000 of zip codes within Dutchess County. Demographic factors that showed significant correlation with AIDS rates within the zip codes in Dutchess County are the percentage of people with public assistance, per capita income, percentage of people born outside the United States, percentage of Whites, percentage of Blacks, percentage of Hispanics, percentage of urban area, percentage of rural area, income below poverty level and income above poverty level. In order for the correlation coefficients to be significant at the p < .05 level, the correlation coefficients had to be at or greater than 0.413. However, in order for the correlation coefficients to be significant at the p < .01 level, the correlation coefficients had to be at or greater than 0.526.
In the very beginning of my correlations, I noticed that I kept getting correlation coefficients less than 0.413, showing that whatever demographic factor I was attempting to correlate proved to be insignificant. In response, I decided to remove zip code 12582 (Stormville) from the data. I figured Stormville needed to be removed because its value was exceptionally high for a single population. After removing Stormville, the correlation changed and the p-value became significant.
As I looked back on my first paper, I believed race played a distinct relationship with the AIDS rates of the zip codes in Dutchess County. According to my data in my first paper, Blacks had the highest rate among all of the other races. Due to this particular data, I expected a significant correlation. I came to find that my assumption was correct. On a correlation significance of p < .01, the correlation coefficient for the percent of Blacks correlated with the AIDS rates of the zip codes in Dutchess County was 0.637. This correlation coefficient was actually the most significant ones of the nine other significant demographic correlations tested. The significantly positive correlation coefficient shows that, the higher the percentage of Blacks residing in a given zip code, the higher the AIDS rate is expected to be within that zip code. The high AIDS rates for the black community may greatly be due to the fact that, some of these people were probably poor and have no medical insurances. Perhaps, since Intravenous Drug Users (IDU) was the highest percentage of the mode of transmission for cumulative AIDS rates in Dutchess County since 1980, Blacks within Dutchess County may be using drugs more often than other races. The outlier at point (725.71, 26.96) is zip code, 12601, which is Poughkeepsie; its value is simply exceptionally high for a single population (see table 1, see graph 1, see map 1). There is also another outlier at point (798.58, 2.55) is zip code 12533, which is East Fishkill; its particular value is simply exceptionally low for a single population (see table 1, see graph 1, see map 1).
With regards to the White community living in the zip codes of Dutchess County, the correlation coefficient showed the highest correlation coefficient of -0.640 (see table 1). On a correlation significance of p < .01, the negative correlation meant that the higher the percentage of Whites residing in a given zip code, the lower the AIDS rate is expected to be within that zip code. In contrast to the reason for Blacks, Whites may have been generally richer and have medical insurances; that is why the AIDS rates for places with more Whites in Dutchess County is generally lower (see map 2). Nonetheless, the significant distribution of the points, zip codes showed no outliers for the Whites (see graph 2).
I also tested the correlation between AIDS rate and the percentage of people born outside the United States. I tested this because I wondered if the correlation would be significant since foreigners of different cultures have different ways of living and living behaviors than natives who are born in the United States. The correlation coefficient for this particular correlation is 0.451. On a correlation significance of p < .05, the positive correlation signifies that the higher the percentage of people born outside the United States residing in a given zip code, the higher the AIDS rate is expected to be within that zip code. There are two evident outliers at points (687.64, 4.17) which is zip code 12508, Beacon and (494.23, 4.20) which is zip code 12592, Wassaic (see table 1, see graph 3, see map 3). Both of the value of the two zip codes is exceptionally high for a single population.
Another demographic factor I have tested is of the percentage of people with public assistance within Dutchess County. The correlation coefficient was 0.512, on a correlation significance of p < .05. Thus, the positive correlation implies that within a given zip code in Dutchess County, if the percentage of the people with public assistance is high in a given zip code in Dutchess County, the AIDS rate would most likely be high there as well. The outlier at point (494.23, 9.49) is zip code 12592, which is the town of Wassaic (see table 1, see graph 4, see map 4). This is due to the exceptionally high value for a single population. I was not surprised with the positive correlation result because I thought that people with public assistance did not have some essentials of living, like medical insurance; as for this I thought that AIDS rates had a direct relationship with the percentage of people with public assistance.
Per capita income in 1999 correlated with AIDS rate in Dutchess County showed a correlation coefficient of -.413, based on a correlation significance of p < .05 (see table 1). The negative correlation signifies that the lower the per capita income in a given zip code in Dutchess County the higher the AIDS rate is expected within that particular zip code (see map 5). The graph did not show an evident outlier (see graph 5). There were no zip codes with exceptionally high values for a single population. Much like per capita income in 1999 correlated with the AIDS rates in Dutchess County the percentage of people with income above poverty level had a negative correlation. The correlation coefficient was -0.428, based on a correlation significance of p < .05 (see table 1). In addition to the similarity of per capita income, since the percent of people with income above poverty level had a negative correlation, it simply implies that in a given zip code in Dutchess County where there is a higher percentage of people with income above poverty level, the higher the AIDS rate is expected within that particular zip code (see map 6). There are no outliers for this graph also because there were no zip codes with exceptionally high or low values for a single population (see graph 6). In contrast to the negative correlation of the percentage of people with income above poverty level with AIDS rate in Dutchess County, the percentage of people with income below poverty level had a positive correlation, implying that if the percentage of the people with income below poverty level is high in a given zip code in Dutchess County, the AIDS rate would most likely be high there as well. Moreover, this demographic factor had an outlier. This outlier could be seen at point (725.71, 18.709) which is zip code 12601, Poughkeepsie (see table 1, see graph 7, see map 7). I was not at all surprised by the outcomes of per capita income, the percentage of the people with income above poverty level and the percentage of the people with income below poverty level because I thought that people with less money are the ones that would less likely get tested for HIV and also be the ones that are unaware of the disease.
Lastly, the final two correlations I tested were percent rural and percent urban. Looking at the graphs, there are evidently many outliers (see graph 8, 9). Some were exceptionally high and some were exceptionally low for a single population. Surprisingly, both correlations were significant, based on a correlation significance of p < .05. Rural had the correlation significance of -0.423 (see table 1). The negative correlation signifying that as a zip code in Dutchess County is more rural, it would be expected to see the that AIDS rate in that particular zip code is low (see map 8). On the other hand, urban had the correlation significance of 0.423 (see table 1). The negative correlation signifying that as a zip code in Dutchess County is more urban, it would be expected to see the that AIDS rate in that particular zip code is high (see map 9).
Works Cited
1) US census bureau
http://factfinder.census.gov/servlet/DatasetMainPageServlet?_program=DEC&_submenuId=&_lang=en&_ts=