This morning’s Rocky Mountain News repeats a frequently misused statistic, the so-called “union premium.” In its story on shifting union membership numbers in Colorado, the News reports the Bureau of Labor Statistics findings that “[f]ull-time workers covered by union contracts earned median pay of $833 a week vs. $642 for those not represented.”

What the News doesn’t report, however, is the Bureau’s own disclaimer attached to those statistics: “The difference reflects a variety of influences in addition to coverage by a collective bargaining agreement, including variations in the distributions of union members and nonunion employees by occupation, industry, firm size, or geographic region.”

Put into plainer English, the Bureau warns that union membership is more common among jobs that pay more for reasons that may be entirely unrelated to unionization itself. Big-city jobs often pay more than their rural counterparts, big businesses often pay more than small businesses, skilled employees usually get paid more than unskilled employees; unionization may be more common among certain types of jobs (big city, big business, high skills), but that doesn’t mean that unions necessarily cause employees to live in cities, work for big firms, or have useful skills. Add in the fact that union organizers have an incentive to seek to represent employees with higher incomes in the first place (since they can fund a better lifestyle for the union staff).

Naturally, the AFL-CIO flacks the “union premium” without any statistical context. (Bret adds: Nor do they disclose the full cost of membership!)