Overview
Geographic information systems combine spatial data (points, lines, polygons and rasters) with attributes, so you can map, measure and model patterns across space. Research uses range from mapping survey respondents and calculating travel distances to analyzing satellite-derived land cover and testing whether outcomes cluster geographically.
We work in QGIS (free and open source), ArcGIS Pro (Esri's current desktop GIS, which has replaced ArcMap) and in R or Python for reproducible spatial analysis. We begin with the data: coordinate reference systems, geocoding quality, boundary changes over time and the appropriate spatial unit. Then we move to analysis matched to your question.
Spatial data violate a core assumption of many statistical tests: nearby observations tend to be similar. We test for spatial autocorrelation and, where present, use spatial regression or geographically weighted regression so your conclusions account for it.
Who it’s for and the problems it solves
This is for you if you are…
- Students in geography, urban planning, environmental science, public health, economics and development studies
- Researchers who have location data but little GIS training
- Doctoral candidates needing publication-quality maps and defensible spatial statistics
- NGO and policy researchers producing maps for reports and evaluations
Common problems we help with
- Your layers do not line up or distances come out in degrees instead of meters
- You need to join survey data to administrative boundaries that changed over time
- Your map uses a color scheme that misleads or is unreadable in print
- A reviewer asked whether your regression residuals are spatially autocorrelated
- You are unsure whether to use hot spot analysis, spatial lag, spatial error or GWR
What’s included — and what isn’t
Included
- Data preparation: coordinate reference systems and projections, reprojection, geocoding, topology checks and attribute joins
- Spatial operations: buffers, overlays, spatial joins, zonal statistics, distance and network-based travel time analysis
- Thematic mapping: choropleth classification, proportional symbols, dot density, color-blind-safe palettes and map layout
- Spatial autocorrelation: global Moran's I and local indicators of spatial association (LISA)
- Hot spot analysis: Getis-Ord Gi* with appropriate spatial weights and multiple-testing awareness
- Spatial regression: spatial lag and spatial error models, with diagnostic tests
- Geographically weighted regression (GWR and multiscale GWR) for relationships that vary across space
- Raster analysis: land cover, elevation, remote-sensing indices such as NDVI
- Reproducible workflows in R (sf, terra, spdep, spatialreg, GWmodel) or Python (GeoPandas, PySAL)
Not included
- Geolocating or mapping individuals in ways that breach confidentiality or ethics approval
- Professional surveying, legal boundary determination or engineering-grade GIS work
- Supplying licensed ArcGIS software or proprietary datasets
What you receive
- Clean spatial datasets
- Projected, validated layers (GeoPackage or shapefile) with a data dictionary and source notes.
- Maps
- Publication-ready maps with legend, scale bar, north arrow where appropriate, projection note and accessible colors, in vector and image formats.
- Spatial statistics report
- Moran's I, LISA or hot spot results, and spatial regression or GWR outputs with diagnostics and interpretation notes.
- Project files and code
- QGIS or ArcGIS Pro project files, or R/Python scripts, so the work can be reproduced and updated.
- Walkthrough
- A session explaining methods and results so you can present them confidently.
How it works
Question and data review
We define the spatial question, study area, spatial unit and data sources.
Prepare
We clean, project and join data, checking accuracy and flagging boundary or scale issues.
Map and analyze
We produce exploratory maps, test for spatial patterns and fit models suited to your question.
Deliver and explain
We finalize maps and tables and walk you through the results.
Methods and software
- QGIS
- Free, open-source desktop GIS with processing tools and plugins; long-term release versions are recommended for stability in research projects.
- ArcGIS Pro
- Esri's desktop GIS, including the Spatial Statistics toolbox (Hot Spot Analysis, Cluster and Outlier Analysis, GWR). ArcMap has been retired, so older ArcMap tutorials may not match current menus.
- Spatial autocorrelation
- Global Moran's I and local Moran's I (LISA), with spatial weights based on contiguity, distance or k-nearest neighbors.
- Spatial regression
- Spatial lag and spatial error models chosen using Lagrange multiplier diagnostics, estimated in R spatialreg, GeoDa or PySAL.
- GWR and MGWR
- Local regression that lets coefficients vary across space, with bandwidth selection and care about multicollinearity and multiple testing.
- Projections
- Equal-area projections for density and area, conformal or local projections for distance, and consistent CRS across layers.
Projections: small setting, big consequences
Many errors start with coordinate reference systems. Latitude and longitude (for example, WGS 84) are fine for storing locations but poor for measuring distance and area. We reproject data into a suitable projected system for your study region before any buffers, distances or densities are calculated, and state the CRS in your methods.
Testing for spatial dependence
If neighboring areas have similar values, ordinary regression may produce overconfident results. We map residuals, compute Moran's I and use diagnostics to decide whether a spatial lag model, spatial error model or GWR is appropriate.
- Spatial lag: the outcome in one area is influenced by outcomes in neighboring areas
- Spatial error: unmeasured factors are spatially clustered
- GWR: the strength of relationships changes across the study area
- Hot spot analysis: where are statistically significant clusters of high or low values?
Scale and aggregation
Results can change when the same data are aggregated to different areas — the modifiable areal unit problem. We help you choose units that match your theory and data, test sensitivity where feasible and word conclusions to avoid ecological fallacies about individuals.
What the work can look like
Illustrative example — fictional data
Illustrative spatial analysis of clinic access
A fictional public health study examined travel time to the nearest primary care clinic across 420 districts. All values are invented.
Spatial autocorrelation
Travel time showed strong positive spatial autocorrelation (Moran's I = 0.52, p < 0.001 by permutation). LISA identified a cluster of high-travel-time districts in the northeast.
Model choice
OLS residuals were spatially autocorrelated. Lagrange multiplier tests favored a spatial error model, which was reported alongside OLS for comparison. A GWR model showed the effect of road density was stronger in rural districts.
Illustrative example with invented data; it does not describe a real client or location.
What you’ll need to provide
- Research questions and the study area
- Spatial data (coordinates, addresses, shapefiles, rasters) and attribute data with a data dictionary
- Source and date of boundary files, and any license terms
- Confidentiality requirements for location data
- Preferred software and output formats
You don’t need everything ready to ask for a quote — tell us what you have and we’ll say what’s missing.
What affects price and turnaround
Price depends on
- Number and complexity of data sources and joins
- Mapping only versus spatial statistical modeling
- Raster or remote-sensing processing requirements
- Number of final maps and formats required
- Software (QGIS, ArcGIS Pro, R or Python) and reproducibility needs
Turnaround depends on
- Data quality, especially address geocoding and boundary mismatches
- Size of raster datasets and processing time
- Rounds of map design revision
Every project is quoted individually. How we price our work.
Questions people ask before starting
Should I use QGIS or ArcGIS Pro?
QGIS is free and fully capable for most research needs. ArcGIS Pro may be preferred if your institution licenses it or you need specific Esri tools. We work in either, and in R or Python for reproducibility.
My tutorial uses ArcMap. Is that a problem?
ArcMap has been retired in favor of ArcGIS Pro, so menus and tools differ. We can translate an ArcMap workflow into ArcGIS Pro or QGIS.
Can you geocode addresses from my survey?
Yes, subject to your ethics approval and geocoding service terms. We report match rates and can aggregate or jitter locations for publication to protect confidentiality.
Do you do remote sensing?
We support common research tasks such as land-cover summaries, vegetation indices and zonal statistics. Specialized remote-sensing classification projects are scoped case by case.