One component of the Drug Court program is frequent random drug testing for a variety of drugs – THC, Meth, Cocaine (crack or powder), Opiates (painkillers), Barbiturates, Benzodiazapines (Xanax), and Alcohol.
The program generally consists of an available census of 50 people, which are in our program for a period of two years. When we started six years ago our original sample size design was 10% of the census per day – generally Monday through Friday – yielding an average of one test per week per participant. At that level we were getting a 99.4 % negative (no drug use) test result rate.
Last March, the State has instituted a standard that calls for all participants who have been in the program for less than one year to be tested twice per week. This pretty much doubled the number of tests per day that had to be observed and processed ($12.00/ea). We tested at this new level throughout last summer and didn’t notice a change in our negative result rate – in essence, increased testing didn’t increase the number of positive results.
What we’re looking for is some help in designing a cost-efficient yet reasonably reliable and defensible method of determining sample sizing. I took a stat class in 1992 at (then) Dalton Junior College and, unfortunately most of my knowledge in this area has dribbled out of my brain. I do remember enough of it to know that we are asking for a simple solution to a complex challenge with many variables.