ASA_Andrew
- 1. Mining in Appalachia has been linked to higher birth defect rates (Ahern et al., 2011), water quality degradation (Lindberg et al., 2011; Bernhardt et al. 2012), increased cardiovascular disease (Esch & Hendryx 2011), higher cancer rates (Hendryx, Wolfe, Luo & Webb, 2011), and increased poverty levels (Partridge, Betz & Lobao, 2012). The evidence of detrimental coal mining impacts on the environment and local communities is robust and mounting. The evidence, though, is challenged in that the research is too focused and does not see the larger picture (Donovan, 2011) or that the results are caused by something else in the region like the mountains, forests, or other health and wealth factors (Ellis, 2013). There is then a substantial gap in the literature addressing those issues. This study seeks to fill that gap by using a regression model (opposed to a mining presence/non-presence factor) both on a small and large geographic scale, across a large range of variables. This study looks at the Appalachian region as defined by the Appalachian Regional Commission (ARC) to find correlations between mining and socioeconomic status. Socioeconomic Impacts of Coal Mining in Eastern Kentucky and Appalachia Andrew Gott, Dr. Buddhi Gyawali, Jeremy Sandifer, Ken Bates, Kentucky State University Introduction Methodology Abstract Results Coal production data were compared to socioeconomic variables with the hypothesis that increased mining activity will correlate with key socioeconomic indicators. The objective of this study was to identify those socioeconomic variables that correlate well (by means of significance and R2 > 0.02) with mining even after the inclusion of terrain factors and selected socioeconomic covariates of wealth, education, and employment. One study site containing the seven Eastern Kentucky counties of Floyd, Knott, Letcher, Magoffin, Martin, Perry, and Pike and one of the larger Appalachian region as defined by the Appalachian Regional Commission were used for analysis. The numbers of mines and mine production data were analyzed in a 52 km buffer around each Census Block Group (CBG) and were used in a bivariate regression analysis with the ACS variables. To understand the relationship between mining, terrain, and other socioeconomic data, a subsequent multivariate regression was performed. Results show that higher levels of mining activity in the ARC region minimally, but significantly correlated with source of income and employment levels. Future work should look not only at comparative levels of mining in the present day, but how changes in mining activity over time affect socioeconomic status in individual CBGs. The regression analysis shows that while source of income, employment, and education levels consistently showed to correlate best with mining compared to 1700 other investigated variables, these “top” variables still only minimally explained variance in the data along with small coefficients. They did, however, remain statistically significant contributors (p < 0.000001) with the inclusion of terrain and other socioeconomic covariates. Out of the 1700 variables, the ones that correlated best with mining were overwhelmingly focused on source of income and employment (Table 1.) These top variables are displayed in choropleth maps in Figure 2. Compering these to Figure 1 show that while there may be correlation in Central Appalachia, the northern and southern mining regions do not correlate as well. Future work should approach two different angles. One is to investigate factors other than socioeconomic status. Health and environmental quality should be investigated with the same methodology. The second approach involves investigating what makes Central Appalachia so unique. Central Appalachia tends to be among the highest or lowest percentages when looking at mining production, average slope, wealth, education, employment, those receiving benefits and more. As the results showed that mining is not the main explaining factor, not the mountains themselves (viewing Figure 1 shows that both mining and the mountains extend all through Appalachia), some unknown factors are creating this regional uniqueness. National Elevation Dataset (NED) 2013 1/3 arc second DEMs were downloaded from USGS in 1 degree tiles. Using ArcGIS Pro, all tiles were merged using the Mosaic to New Raster Tool and clipped to the study site (Figure 1a). The elevation range, and average elevation were calculated for each Census Block Group (CBG) using the Zonal Statistics to Table tool. The Slope tool was then used, and the Zonal Statistics to Table tool repeated, finding the range and average slope for each CBG (Figure 1c). Next, coal mining production data for the years 2000-2014 were downloaded from the Mine Safety and Health Administration (MSHA), and XY data displayed. The Summarize Within tool was then used, aggregating the number of mines found in each CBG neighborhood (52 km around each CBG centroid) for all 15 years (Figure 1). ArcGIS Pro was then used to run the Exploratory Regression tool to find the model of best fit for American Community Survey socioeconomic variables using: 𝑦 = 𝛽0 + 𝛽1 𝑥1 + 𝛽2 𝑥2 + ⋯ + 𝛽 𝑛 𝑥 𝑛 + 𝜖 with y being the value of the ACS variable, 𝛽0 being the intercept (the predicted value of the ACS variable when all covariates are zero), 𝛽1 being the coefficient, and 𝜖 being the residual error. Figure 2 a.) B19051e2: Percent with Earnings Income b.) B19052e2: Percent with Wage or Salary Income c.) B19053e2: Percent with Self-Employment Income d.) B19055e2: Percent with Social Security Income e.) B19056e2: Percent with Supplemental Social Security Income f.) B23022e25: Did not work in the past year g.) B23024e29: Above Poverty line and employed h.) B23025e4: Employed i.) B23025e7: Not in labor force j.) B27010e38: Percent of those on Medicare alone k.) B11005e5: Fraction non-married households with a member under age 18 and overlaid mine locations l.) B11005e5: Fraction of those receiving Social Security Income and overlaid mine locations. Literature Cited Ahern, M., Hendryx, M., Conley, J., Fedorko, E., Ducatman, A., & Zullig, K. (2011). The association between mountaintop mining and birth defects among live births in central Appalachia, 1996–2003. Environmental Research, 838-846. doi:10.1016/j.envres.2011.05.019 Bernhardt, E., Lutz, B., King, R., Fay, J., Carter, C., Helton, A., . . . Amos, J. (2012). How Many Mountains Can We Mine? Assessing the Regional Degradation of Central Appalachian Rivers by Surface Coal Mining. Environmental Science & Technology Environ. Sci. Technol., 8115-8122. Donovan, T. (2011, June 27). Mountaintop Removal Mining Birth Defects: New Study Suggests Controversial Coal Operations Linked To Adverse Health Effects. Retrieved August 19, 2015. Ellis, R. (2013, March 16). Report: Health risks high for mountaintop removal areas. Retrieved August 19, 2015 Esch, L., & Hendryx, M. (2011). Chronic Cardiovascular Disease Mortality in Mountaintop Mining Areas of Central Appalachian States. The Journal of Rural Health, 350-357. doi:10.1111/j.1748-0361.2011.00361.x Lindberg, T., Bernhardt, E., Bier, R., Helton, A., Merola, R., Vengosh, A., & Giulio, R. (2011). Cumulative impacts of mountaintop mining on an Appalachian watershed. Proceedings of the National Academy of Sciences, 20929-20934. Partridge, M., Betz, M., & Lobao, L. (2013). Natural Resource Curse and Poverty in Appalachian America. American Journal of Agricultural Economics, 449-456. a. b. c. d. e. f. g. h. i. j. Variable Beta Description ARC Region B19051e2 -0.00041 Earnings Income B19052e2 -0.00036 Wage or Salary Income B19053e2 -0.00017 Self-employment Income B19055e2 0.000348 Social Security Income B19056e2 0.000179 Supplemental Social Security Income B23022e25 0.000422 Did not work in the past year B23024e29 -0.00036 Above Poverty Line and Employed B23025e4 -0.00034 Employed B23025e7 0.000413 Not in Labor Force B27010e38 0.00013 Medicare alone Seven County Study Site B11005e5 -0.03379 Other Household with children under 18 B19055e2 0.05496 Social Security Income Table 1. All ARC Region and Seven County Study Site ACS variables with bivariate OLS regression results of R2 > 0.02 and p- value < 0.001, and for which mining remained a significant variable in the multivariate model. Figure 1. Mine Count per CBG neighborhood defined as 52 km from the CBG centroid. The count represents the cumulative total for 15 years. Data from the Mine Safety and Health Administration. k. l.