Latest Contributions by CharlieBryan

Thanks for the resources! I have a BS in GIS and I have performed this type of regression in the classroom and as a research assistant. However, I did not have much experience with  the missing data issue until now. yea-aah statistics with real world data! ... View more

I tried using euclidean distance (cell size 250 ft) from city centers because I could not get all the the park features for the entire study area ( which I had to expand in order to run OLS because we only have 22 census tracts and to run any regression you need at least n=30). Then executed the spatial join tool to join and  create a new variable average distance to incorporated city centers for each census tract. Tobacco retail is generally located closer to city centers ( like all retail ). I really thought I needed this variable ( or some other spatial reference variable like average distance to highway) in order to have a get a good OLS regression model that would identify significant socio-economic ( data from census) disparities caused by density variation of tobacco retail.Turns out tobacco density has a significant negative correlation with median income and positive  correlation with higher density of people with a disability (adj R-squared of .80).  However, the all models it tried over and under predicted in all the high tobacco density census which is what we are trying to explain. Then I found and tried a completely different approach (I would just focus on the 22 census tracts in our county). First I calculated the  tobacco retail per 1,000 residents in the county. Then identified the tracts that had a rate higher the county rate. These tracts are considered to have high density compared to the tracts that had a lower rate than the county rate.I calculated the average of the following socio-economic indicators  for  high rate tracts  and  low rate tracts: percent below poverty, median income, percent in labor force, gini index, percent uninsured, percent white, percent household with one + under 18 years, average house size, percent male, percent female, percent 15-19 year old, percent 20-24 year old, median age, percent disabled, percent veterans. I then calculated the difference between the two averages for each soci-economic indicator. Differences greater than 4% were highlighted as potential disparities. I would then perform a two sample test ( if possible) to determine if the differences in the high and low rate tracts are statistically significant. This method is more effective in my opinion. Comments and advice is welcome. ... View more

Yes I waited but was not redirected to the link you provided. Thank you ... View more

Wow thanks for the detailed response! The map attached is a rough draft. I included the schools just because I forgot to turn off the layer when I exported the map. Survey locations were not picked, we contacted all retailers and 16 chose not to participate in the survey. The only evidence that can indicated that these locations would not be different from closest surveyed locations or other surveyed locations is by understanding the survey data better. For example if our data exploration reviles that store type is the main factor that affects advertising, we can classify the not surveyed by store type and see if they are represented in the surveyed locations. ... View more

Thanks ... View more