More about determining levels of risk
Levels of Risk
Please note that the levels of risk represented here are relative to the counties and independent cites in the Commonwealth of Virginia only as determined by quartile rank.
These levels of risk are not meant to represent an objective state of risk to children. Instead it it meant to measure the quartile rank of each locatilty relative to other counties/cities in Virginia only.
How are quartiles determined?
Quartiles divide the number of OBSERVATIONS into 4 equal groups such that 25% of the localities are in the 1st Quartile, 25% in the 2nd etc. Quartiles do not divide the range of observations into 4 parts, but sort the 134 localities by the indicator from lowest to highest and then determine which OBSERVATION (locality) represents the cutoffs for the quartiles and use that value. For an EVEN number of values (e.g., 134 localities) the position containing the cutoff for each quartiles in a sorted list is determined as follows:
- The median is (134+1)*0.5 = 67.5. So the median is the average of the values found in localities in the 67th and 68th positions (sorted lowest to highest by that value). Every locality < or equal to this value is in the 50% percentile.
- The 1st quartile cutoff is (134+1)*0.25 = 33.75. So the Q1 cutoff is the average of the values found in localities in the 33rd and 34th positions.
- The 3rd quartile cutoff is (134+1)*0.75 = 101.25. So the Q3 cutoff is the average values found in localities in the 101st and 102nd positions.
Since 134 is not evenly divisible by 4, there will not be equal numbers of localities in each quartile. Q1 will have 33 localities, Q2 will have 34, Q3 will have 34, and Q4 will have 33.
In this way, the top ¼ of localities are represented by the “high” group regardless of how big or small the range of values measured are.