Does the Public Prefer Hitting?

              It has long been my belief that the public reacts better to baseball in big-hitting eras like the 1930s and the 1994-2005 era than in low-run eras like the Dead Ball Era (1902-1919) or the 1960s (1963-1968).   Recently I undertook a study to attempt to prove that this was true. 

              I took in consideration all of baseball history from 1890 to the present, excepting the Player’s League (1890), the American Association (1890-1891) and the Federal League (1914-1915), excepting those because (1) they lack reliable attendance data, and (2) they are disconnected from a coherent history.  For the other leagues—the National League since 1890 and the American League since 1901—I worked toward two figures:  the runs scored per game in each season, and rate of increase in attendance in each season. 

              Without going into perhaps unnecessary detail, the runs scored per game was a moving average based on the 1-2-6-2-1 balance that I sometimes use, so that the "1940" figure is based 50% on the data from 1940 alone, and 50% on the data from the surrounding seasons.  I did this because each season’s attendance is (I believe) very much effected by what has happened in previous seasons.  People go to baseball games because they like baseball.  An exciting pennant race, a historic event, a good World Series or a popular innovation in one season effects the attendance the next season. What I am interested in is big-hitting eras and low-hitting eras, as much as big-hitting seasons and low-hitting seasons.

              So I figured the "runs per game" for each season, 1890 to 2017, and sorted those in two ways: high and low (two classes) and five classes (very high, moderately high, neutral, relatively low, and very low).   We can’t use the 1890 season because we can’t measure an increase/decrease in attendance in that season, since we have no "base" season behind it.  Reporting the classes:

              Runs Scored Levels are very high (top group) from 1891 to 1901, in 1925, from 1928 to 1932, from 1935 to 1938, in 1996, and from 1998 to 2000.

              Runs Scored Levels are moderately high (second group) from 1921 to 1924, in 1926-1927, in 1933-1934, in 1939-1940, in 1949-1950, in 1994-1995, 1997, from 2001 to 2009, and in 2017. 

              Run Scoring Levels are neutral (third group) in 1902, 1912, 1920, 1941, 1947-48, from 1951 to 1962, in 1977 and 1979, from 1985 to 1987, in 1993, in 2010-2011, and in 2016.  The 1950s are the "most normal" group of seasons in baseball history in this respect.

              Runs Scored Levels are relatively low in 1903, 1911, 1942, from 1944 to 1946, in 1970, from 1974 to 1976, in 1978, from 1980 to 1984, from 1988 to 1992, and from 2012 to 2015.

              Runs Scored Levels are very low from 1904 to 1910, from 1913 to 1919, in 1943, from 1963 to 1969, and from 1971 to 1973. 

              Then I looked at the percentage increase in per-game attendance in each season—2007 as opposed to 2006, 1997 as opposed to 1996, etc.   The best years for attendance growth were, in order, 1946, 1919, 1945, 1893, 1891, 1920, 1944, 1947, 1895 and 1903.   The worst years for attendance decline were, in order, 1898, 1914, 1932, 1931, 1950, 1951, 1917, 1933, 1961, and 1995. 

              My expectation was that attendance growth would, on average and overall, be higher in big hitting eras than in low-hitting eras.   The data does not show this to be the case.   Perhaps the thesis was wrong; perhaps I didn’t study it in exactly the right way, I don’t know.   But the study does not show that attendance increases more rapidly in big-hitting eras than in low-hitting eras.  It suggests that the opposite may be the case. 

              Looking first at the binary split (two groups). . . the overall attendance growth rate, from 1890 to the present, has been 2.6%.  In high-run seasons, the overall growth rate has been 1.6%.  In low-run seasons or low-run eras, it has been 3.6%. 

              That LOOKS like strong evidence against the theory that fans like offense, but the data is less convincing with a closer look.   This is the data from the five groups:

              If you focus on the MOST telling groups—very high runs and very low runs—you see a different pattern, a pattern consistent with my initial theory:

              Attendance has grown more rapidly in truly high-run eras like the 1920s and the steroid era than in pitching-dominated eras like the Dead Ball era and the 1960s.  The split in the binary data (1.016 vs. 1.036) is created entirely by the moderate eras.  Attendance has grown far more rapidly in eras when scoring was relatively low than in eras when it was above average, but not really high.  Weighting the data for its inherent intensity would largely neutralize the conclusion.

              So. . .what do we learn here?  This data certainly does support my initial thesis. It provides slight but unconvincing evidence of the opposite proposition.