Supporting Tables for
To assess how local union strength and size affect per-pupil spending, we initially estimated an ordinary least squares (OLS) regression model with measures of both strength and size. This is the first model in table A6.1 and includes all states. All states can be included because public opinion is not included in this model but will be introduced later in the chapter. “Strong union in district” is coded “1” when the local union has collective-bargaining rights and represents a majority of the teachers. Union size is the number of union members (logged because it is highly skewed).
For many of the analyses in this chapter, we employ state-level independent variables (e.g., state government spending for education). These would be perfectly predicted by (collinear with) the set of state dummy variables employed in all our regression models in chapters 4 and 5. In order, then, to account for clustering within states, these analyses use Huber-White standard errors (also known as robust standard errors). Including opinion does not change the results, and we do show this later in the chapter.
The results show that school districts with strong local unions spend, on average, $597 (or about 12 percent) more for each child but that large union memberships seem to have no effect. However, we have argued that unions may need both size and collective bargaining strength to be effective. We test this in the second model in table A6.1, which includes both union power measures and the interaction of the two dimensions. By the logic of statistical interactions, we then have a measure of the membership size in districts lacking a strong union and a measure of membership size in districts with strong unions in place. Here, the results tell a very different and interesting story. When we estimate the impact of membership separately for districts with and without strong unions, we see that membership has a large impact on spending only when the union is strong. Going from the mean of 135 unionized teachers to 1,000 unionized teachers (slightly less than a standard deviation increase) increases instructional spending by about $545 per pupil. The ap-