Union success in representation elections: Why does unit size matter?

The author documents four facts regarding NLRB-supervised representation election activity over the period 1952-98: (1) election activity fell sharply and discontinuously beginning in the mid-1970s, after a two-decade rise; (2) unions' election win rate declined less sharply, though continuously, over the entire period; (3) a "size gap" characterized unions' win rates throughout the period, with a lower win rate in larger units; and (4) the size gap widened substantially after 1952. A simple optimizing model of the union decision to hold a representation election can explain (1)-(3). The author describes two possible explanations for (4), one based on differing behavior by employers in different size classes, and one purely probabilistic. Results of empirical tests using NLRB election data for 1952-98 suggest that the optimizing model of the union decision to hold an election combined with the purely probabilistic explanation can largely account for the observed patterns.