Determining Confidence Interval

I have been wanting to write a paper about determining confidence or measurement in diversity of a sample but haven't had time to do any research. Give you some background, My work is to deploy civilian radar systems and before we put a radar system into operation we must conduct a flight test. During a flight test we hire/lease a aircraft that has the ability to record its position and some other information ... i.e. a "truth source". We then take this truth source and compare it to what the radar has calculated and we come up with performance figures and compare it to the standards and hopefully we pass. I have worked on papers about determining fidelity or confidence interval when we must detect the target 97% of the time in a 4 second window (update interval) and how many samples we must collect to have confidence in the measurement. However what I am interested in this time is the distribution of the flight. If you think of radar coverage as simple circle; then I could always fly just one half, the better half and skew my sample so my results are better. I am looking to understand if there is a way to measure that "skew" to prevent it from happening. Probability of detection (97%) is usually not an issue for this type of "skewing" however accuracy is and typically a radar requirement is an accuracy of 150m RMS. So I could fly my aircraft only in the areas that have good performance. In Radar, the radar people most think of, rotating radar, this is not a problem but the radar system I deploy are based on multilateration, so the geometry of the deployed stations effects the performance (DOP.)