11.07.2015 Views

Mapping Cultural Participation in Chicago - Cultural Policy Center

Mapping Cultural Participation in Chicago - Cultural Policy Center

Mapping Cultural Participation in Chicago - Cultural Policy Center

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Appendix IIIModel<strong>in</strong>g and PredictionTo calculate our measure of unexpected participation, we specified and estimated theparameters of a statistical model that related the participation density <strong>in</strong> artsorganizations for each census tract to several socioeconomic characteristics for thattract.• To account for differences <strong>in</strong> <strong>in</strong>come among areas we used the natural log of themedian household <strong>in</strong>come <strong>in</strong> the tract. We also experimented with us<strong>in</strong>g median<strong>in</strong>come and its square.• To account for educational atta<strong>in</strong>ment we used the percentage, and its square, ofresidents over 25 years of age who had a bachelor’s degree. We also controlled forthe percentage of residents that had a master’s level or higher degree.• To account for the age structure of the census tract we used the percentage ofresidents between the ages of 35 and 49 and between the ages of 50 and 65. We also<strong>in</strong>teracted these two age variables with the percentage of residents with a bachelor’sdegree.• To account for differences <strong>in</strong> household structure among tracts we considered theaverage household size, percentage of households headed by a female with children,percentage of households with children, and percentage of households with onemember. Once we accounted for these variables, the percentage of householdsheaded by a woman with children had no explanatory power <strong>in</strong> the model and so itwas dropped from our analysis.We measured participation <strong>in</strong> our analysis as the natural log of the participation density<strong>in</strong> large organizations. Each of the socioeconomic characteristics was related l<strong>in</strong>early tothis variable. We estimated our statistical model us<strong>in</strong>g the method of least squares. Weconsidered several variations on the model where we <strong>in</strong>cluded or excluded some of thevariables described above. We also alternatively excluded some census tracts from theanalysis depend<strong>in</strong>g on their small size. We weighted the observations by the number ofoccupied households <strong>in</strong> the census tract. F<strong>in</strong>ally, we adjusted downward the participationdensities <strong>in</strong> the downtown Loop community area to account for the relatively largenumber of bus<strong>in</strong>ess addresses from this part of the city <strong>in</strong> our database. The purpose ofestimat<strong>in</strong>g these variations <strong>in</strong> our statistical model was to see whether our f<strong>in</strong>d<strong>in</strong>gs onethnic composition of a community were sensitive to our specification and assumptionsabout the data.The unit of analysis was the census tract. There are 865 such tracts <strong>in</strong> <strong>Chicago</strong>. Becausewe excluded tracts with small numbers of people, we used only 810.55

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!