Orange County NC Website
Browse       Search
14 <br />information needed to test for statistically significant <br />differences between occupant characteristics of recently <br />constructed apartments and single - family housing. The <br />minimum sample size was determined by reviewing previous <br />apartment surveys. The amount of variation in household <br />size and in number of children were used to estimate sample <br />size. The PI determined that a sample of 150. apartment <br />units and 150 houses would be sufficient to generate <br />statistically significant difference -of -means tests for <br />North Carolina urban areas. <br />The goal of having results that could be generalized to <br />North Carolina urban areas ideally required a listing of all <br />recent construction in all metropolitan counties. Of <br />course, this ideal sample frame does not exist. Instead, <br />expert sources were used to identify recent apartment <br />complexes and subdivision locations in the five North <br />Carolina metropolitan areas surveyed. AANC identified 19 <br />apartment complexes constructed since 1990 in these areas. <br />Local planners, economic developers and real estate brokers <br />were contacted for information on recent residential <br />subdivisions and on additional apartment complexes. This <br />process essentially randomized the selection of street <br />locations in each metropolitan area. <br />The sample was drawn from street address and phone number <br />listings derived from Select Phone Deluxe — software that <br />includes a set of 6 CD -ROMS and essentially compiles the <br />telephone listings available for every area of the U.S. The <br />software was satisfactory for the task although it was not <br />entirely current nor error free. It was superior to the <br />available R.L. Polk & Company's City Directories, which were <br />outdated and not available for all areas. AANC also provided <br />telephone number listings from member tenant lists for the <br />apartment complexes. <br />A sample of households was randomly drawn from these <br />listings. An equal number of apartment units and houses was <br />drawn. The number from each urban area roughly corresponded <br />to that area's relative size. The range was from 69 in <br />Wilmington to 118 in Charlotte. Thus, the sample could be <br />described as a cluster random sample. However, the sample <br />size is far too small to generate any reliable statistics <br />for each area. Instead, the PI compiled information from <br />the 1990 Census to generate the demographic profiles for <br />each metropolitan area in North Carolina and has forwarded <br />this "PUMS" data to AANC. <br />From the sample of 630 households, responses were received <br />from 465 households. The response,rate of 74% was very high <br />