I'm working with confidential survey data that contain each respondent's Census tract (but nothing below that). There are ~6 respondents per Census tract, and about 2,000 distinct Census tracts in the data.
I ran a logistic regression model using respondent-level variables to predict whether the respondent has ever been diagnosed with asthma. I wanted to map the residuals, to examine any potential spatial clusters or patterns, but don't know how to proceed when my unit of analysis = individuals who share the same location (tract) as others in the data.
Can I convert each respondent's location from the Census tract to a unique spatial point within the Census tract? I.e., can I assign a unique spatial point to each respondent within the same Census tract (at random)? Or is there another way to go about this?
Would love any feedback! (Note: I'm currently working with a Spatial Polygons Dataframe in R.)
I was able to do this using @mrhellmann's advice. I used the steps outlined on this site to accomplish this: https://rstudio-pubs-static.s3.amazonaws.com/200263_e77d00d6b6b24aa8890c8c4f074bcdff.html
The only thing I added at the end was to merge my points data points with my actual data frame df:
points.df <- SpatialPointsDataFrame(points, data = df)
I then plotted the residuals using the following code, where sp is my Spatial Polygons data frame:
plot(sp, main= "Residuals"); points(points.df, col= points.df$poly_id, pch= 1, cex= points.df$residuals)