Setting up spatial data for spatial regressions

[Tarjei W. Havneraas]

Hi everyone

I've been working on map plots in R lately, and want to advance to spatial regressions in Stata using the -spmap- package.

My aim is to do cross-sectional analysis. However, I am having some difficulties with formating my data correctly to -spset-.

I have a data set with 19 counties, longitude and latitude and two variables of interest. The variable "group" is used in R to group counites, e.g.

 plot <- ggplot(data = Map_diag_and_med, aes(x=long, y = lat, group = group)) +
  geom_polygon(aes(fill = med))

Although I only have 19 counties and observations from one year on my two variables of interest, the total data set has 82 112 observations in total, due to all the longitude and latitude observations.

* Example generated by -dataex-. To install: ssc install dataex
clear
input float county str6 group double(longitude latitude) float diag
1 "01.1"  10.68531156276028  59.48859956745566 4.55
1 "01.1" 10.692866471099695 59.490463187251976 4.55
1 "01.1" 10.696115331962785 59.493375407845775 4.55
1 "01.1" 10.708315889128597 59.490451533272335 4.55
1 "01.1" 10.713956978021121   59.4778906633188 4.55
1 "01.1" 10.718585124375561  59.47630827739814 4.55
1 "01.1" 10.720418520606842   59.4736656505165 4.55
1 "01.1"  10.72591715474232 59.471882092417836 4.55
1 "01.1"  10.73608854099581  59.47583202527449 4.55
1 "01.1" 10.741668744010008  59.47906380163044 4.55
end
label values county county_n
label def county_n 1 "Østfold", modify
label var group "group" 
label var longitude "long" 
label var latitude "lat"

I've tried to look up documentation on -spset-, but I have the impression that I need polygons for my counties for -spset- to work, and I wonder if I can work around this with my data set.

I'll add that I've discussed combining data set on a separate post.