If a sports time series, such as attendance, is nonstationary, then the use of level data (e.g., demand estimation using panel data) leads to biased estimates, and the direction of the bias is unknown. In past works, authors have failed to reject nonstationary data, taken first differences, and proceeded with further analysis. That is a legitimate approach, although limiting (e.g., no elasticity estimates can be had from first differences). However, if the data are stationary, then all is well with the usual applications to level data (e.g., taking logs gives direct elasticity estimates). This article rejects that the Major League Baseball attendance time series is nonstationary with break points and suggests the break points deserve additional analysis to facilitate attendance demand investigations.