Media firm Magna research today published its estimates of the audiences for the World Series from 1991. They broke the data down by audience per game in each series (of which, of course, there are a possible 7). There was no Series in 1994 due to a strike.
These numbers are estimates, and I have no idea of their potential error; that is, their appropriate plus/minus. Also, the ratings do not show other ways besides traditional TV that people use to follow the Series. I listen to the radio, for example, and sometimes surf to Major League Baseball’s website to listen or watch. These new media sources could be important as we look for trends because, for example, the MLB website feed did not exist until about 2000.
The newspapers published the data in tabular form, which are always hard to read for time series. So I plotted the data in this figure:
The number of people who are watching is going down. On average, earlier games have lower ratings than do later games. The opening games (black line) are typically the lowest rated, while games 7 (yellow line) have the highest ratings. Crudely extrapolating the observed trend gives us a guess that next year’s Game 1 will see about 16 million viewers.
But this data, even ignoring the measurement error and the possibility of non-television outlets, can still be misleading because the US population has not remained constant during this period. The next plot shows the same data, but by the percent of US population (I used the Census data for estimates).
Back in 1991, about 10-15% of the population watched the Series, but by 2009 that number was down to about 5-7%, a fairly dramatic plunge. Looking at the data this way shows that the slide has more or less halted itself since about 2006. Crudely extrapolating this information says that next year’s Series will be watched by about 6% of the population (again, all is relative to the original data’s potential measurement error).
But this data can still be misleading, especially if we want to estimate the extent of the slide. Plotting just the ratings through time can lead to overconfidence. To show what I mean, I replotted the data for percent US population for Games 1 and 7, and then replotted the same data but relative to the population in 1991.
Take Game 1. By 2009 the percent watching was about 6.2%, but the unadjusted percent relative to the 1991 population was 7.5%. Now, this is not a huge difference, but it could be to advertisers, especially if they want to estimate their reach. If we used just the change in ratings and did not normalize by population, we’d be overestimating the percent of viewers.
None of this analysis answers “Why?” I have no clear idea. Maybe starting the Series the same time kids are going trick-or-treating has something to do with it. If the trend towards longer seasons continues, we’ll soon watch Game 7 as we open our presents on the Federally Recognized Holiday of December 25th That Shall Not Be Named.
Maybe baseball shouldn’t overlap with football, which is now the most popular sport in the US. Viewers have a choice between college ball, pro basketball, hockey, Premier League soccer, among others. Summer is long forgotten by the time the Series wraps.
The steroid scandal couldn’t have helped, nor could the infiltration of soccer in high schools. Ideas?