population of at least 12 years of age resides in the 276 radio markets.10 This delineation of a local radio market, as defined by Arbitron, is widely used by buyers and sellers of radio advertising and generally reflects market data as determined by surveys of listeners.
All figures displayed in the associated charts represent “smooth” lines rather than the actual data. Smoothing is a statistical technique used to illustrate or reveal trends in the data. A line representing the actual data would be filled with jagged ups and downs, much like the representation of an earthquake on a seismograph. Such a representation would make it extremely difficult to discern a trend in the data. On the other hand, a smooth line uses averaging to blunt the jagged ups and downs of the actual data and to reveal any underlying trends. A point on a smooth line represents a weighted average of the actual data in an interval around that point.11 The difference in the lines represents general changes in the radio industry. Because the points on the lines are averages, the reader should not attempt to use these figures to make specific market to market comparisons.
( Arbitron's 276 markets represent about 78 percent of the U.S. population for those at least 12 years of age. Arbitron does not measure radio listening statistics for those under age 12.
( For market 100, for example, the smoothed line will show a weighted average of the actual data in markets 90 to 110. The data from market 100 gets the most weight, data from markets 99 and 101 get the next most weight, and so forth. The particular smoothing method employed is called “loess” and is described in by William S. Cleveland, Hobart Press, 1994. The specific implementation is from the “lowess” command in the statistical package “S-PLUS 2000”, with the smoothing parameter set to 0.5.