How are index scores calculated?

A domain index score is calculated by finding the average of the percentiles for indicators (including both community statistics and resident sentiment data) within that specific domain. While some of these indicators appear in the Track Module dashboards, many other indicators exist behind the scenes and factor into these calculations.

Here’s a breakdown of how index scores are calculated:

1. For each indicator within a domain, we first determine the community’s percentile (or rank) compared to all other communities. For example, a city may be in the 65th percentile for property crime rate (meaning that this city’s rate is “better” than 65 percent of all communities), the 47th percentile for motor vehicle crashes, the 88th percentile for the overall feeling of safety (resident sentiment data from The NCS), and so on. 
2. After we have identified a percentile for each indicator, we then consolidate all the resident sentiment indicators (when available) into one data point. This is done by finding the average of all the resident sentiment percentiles (excluding community statistics percentiles for now) identified in the previous step.
3. This sole resident sentiment percentile is then combined with all the percentiles from the other community statistics indicators. The resident sentiment indicator is weighted to be 10% of the overall index score (with community statistics as the remaining 90%). This method ensures that resident sentiment data is appropriately balanced with community statistics data, regardless of the number of indicators initially included.

A few additional notes:

• If resident sentiment data is not available, the domain index scores will be calculated solely using the community statistics data. In this case, we’ll determine the average percentiles of each of the community statistics indicators, which will make up 100% of the weight for the final index score. 
• All available indicators are included in the index score calculations EXCEPT those that we have identified as being "non-directional", meaning that a higher number is not necessarily better or worse. For example, violent crime rate has directionality (where a higher number is worse), so we can use that to calculate a percentile (reflecting the percentage of communities that received a lower score than yours). But some indicators, such as median age or percentage of minorities, don’t have clear directionality: we can’t say a certain value is better or worse than another, so these indicators are not included in the index score calculations.

Here’s an example of that calculation in action:

Let’s say that the index score for Demo City’s safety domain was 62. To get this score, we first determined the percentiles for each of the 7 safety-related community statistic indicators (violent crime rate, property crime rate, motor vehicle crashes, etc.). We also identified the percentiles for each of the 13 safety-related resident sentiment indicators from the community’s most recent iteration of The NCS (e.g., overall feeling of safety, etc.), and then calculated the combined resident sentiment indicator by finding the average percentile score of those 13 indicators. Finally, we calculated the final index score by finding the weighted average of these 8 values (7 community statistics indicators + 1 combined resident sentiment indicator), where the combined resident sentiment indicator received 10% of the weight, and the other 7 indicators received 90% of the weight.