What is the scoring methodology?
IP3 | Assess uses a Z-score approach to scoring individual indicators and data across domains—the “fuel gauge” visualizations used throughout the platform depict z-scores relative to the selected benchmark. This approach allows an "apples to apples" comparison of data from a variety of sources and with a variety of units and collection methods.
We use this approach because individual indicators are very heterogeneous in what they measure and in the units they are expressed (e.g., rates per thousand, percentages). Therefore, they must be standardized before adding them together to create an index. Z-scores provide a way of standardizing the indicators—converting them all to unit-free measures with mean=0 and standard deviation=1. The value of a Z-score tells you how many standard deviations you are away from the mean. A positive z-score indicates a raw score that is higher than the average mean, and a negative z-score reveals the raw score is below the mean average.
Z-scores are calculated by computing the geographically-weighted mean and standard deviation benchmarks at the national and state level using indicator scores for the original source geography and then computing Z-score values, where:
Z=(Local Value Mean) - (Benchmark Value Mean) / (Benchmark Value Standard Deviation)
Localities with z-scores <-3 or >3 are truncated to -3 and +3, respectively.
After all of the Z-scores are calculated, an index score is calculated for each domain representing the average of all composite indicator z-scores.
The fuel gauges show up bright red if an indicator or domain scores significantly worse than the benchmark, light red or light green if the data are not significantly different from the benchmark, and bright green if the data are significantly better than the benchmark.*
See County Health Rankings methodology for additional details on the z-score methodology.
*A note on colorblindness: We endeavored to select color shades and saturations that could be differentiated by those with colorblindness. However, we recognize that for many, red and green are difficult to discern. The boxes themselves are another way to interpret the results:
A box fully shaded to the left of the centerline represents a score that is significantly worse (i.e., greater than 1 standard deviation) than the benchmark score.
When the box to the left of the centerline is partially shaded, it means the z-score is worse than the benchmark but not a full standard deviation.
A box fully shaded to the right of the centerline represents a score that is significantly better (i.e., greater than 1 standard deviation) than the benchmark score.
When the box to the right of the centerline is partially shaded, it means the z-score is better than the benchmark but not a full standard deviation.