When interpreting these charts, we looked for several key phenomena. First, we expected mean values to be both stable and relatively high over time, and thus we looked for changes in mean patient satisfaction.For example, a particular category or survey question may consistently elicit a mean response rate over 9 on a 10-point scale. However, if this category, although exceeding the benchmark, is consistently lower than the other categories, then staff should still focus process-improvement efforts toward this measure of patient satisfaction. Evidence of seasonality or location-specific differences in mean scores was also of concern. When analyzing standard deviations, we looked for unusually high measures, as this would indicate high process variability. Seasonality and location-specific differences in process variability were also a cause for concern.