Visual inspection of data for obvious changes in pattern is easy and should be done in all cases, especially when you seek quick, approximate information that will be used rapidly in formulating...

Visual inspection of data for obvious changes in pattern is easy and should be done in all cases, especially when you seek quick, approximate information that will be used rapidly in formulating ongoing practice hypotheses. We’ve presented some basic patterns of data simplified as mean and trend lines to draw out these implications more clearly. We’ve distinguished three attributes of graphed data in particular—the level of the data, the trend or directionality of data within one phase (slope), and the trend or directionality of data across phases (drift). When baseline phases are compared to intervention phases using these three attributes, we can indicate general patterns of improvement, strong improvement, or very strong improvement (depending on the number of attributes affected), or we can indicate the same three patterns of deterioration, depending on the definition of desired or undesired events. It’s difficult to interpret data when they’re variable or autocorrelated. Therefore, visual analysis contains inherent risks. We believe it is advisable to supplement visual analyses with some of the simple statistical analyses.