What do I use
Urban planning, estate management and architecture
Land use, valuation, housing studies, post-occupancy evaluation, accessibility and transport.
Your question, and what answers it
One number for the whole study area, tested by permutation.
In ZStat: open your study, go to Analysis, and choose morans i.
The local version, which produces the map planners actually use.
In ZStat: open your study, go to Analysis, and choose getis ord.
For point patterns where there is no value attached, only the positions.
In ZStat: open your study, go to Analysis, and choose nearest neighbour.
Hedonic pricing: which attributes and by how much. Check Moran's I on the residuals - if they cluster, the model is missing something spatial and its standard errors are too small.
In ZStat: open your study, go to Analysis, and choose ols regression.
Build the instrument, check it holds together, then compare groups or model satisfaction.
In ZStat: open your study, go to Analysis, and choose reliability.
Time series for a single series, panel regression when you have several areas over several years.
In ZStat: open your study, go to Analysis, and choose arima.
Rather than running one test per measure and collecting a false positive.
In ZStat: open your study, go to Analysis, and choose manova.
What goes wrong in urban planning, estate management and architecture
- A hedonic model with no spatial check. Two flats on one street are not independent observations and the standard errors will be too small.
- A study area taken from the bounding box of your own points, which is smaller than the real one and makes any pattern look more regular.
- Swapping latitude and longitude. In Nigeria both sit in the same numeric range, so nothing detects it and every distance is wrong.