Nick Huntington-Klein

Screen Twice, Cut Once: Assessing the Predictive Validity of Applicant Selection Tools

Despite their widespread use, there is little academic evidence on whether applicant selection instruments can improve teacher hiring. We examine the relationship between two screening instruments used by Spokane Public Schools to select classroom teachers and three teacher outcomes: value added, absences, and attrition. We observe all applicants to the district (not only those who are hired), allowing us to estimate sample selection-corrected models using random tally errors and variation in the level of competition across job postings as instruments. Ratings on the screening instruments significantly predict value added in math and teacher attrition, but not absences—an increase of one standard deviation in screening scores is associated with an increase of about 0.06 standard deviations of student math achievement, and a decrease in teacher attrition of 3 percentage points. Hence the use of selection instruments appears to be a key means of improving the quality of the teacher workforce.

COLLEGE CHOICE AS A COLLECTIVE DECISION

Although the choice between colleges can be thought of as being made collectively by a family, models of educational choice almost universally portray the decision as made by the student alone. Using a novel experimental method for identifying collective decision functions, I find that students have more influence than parents over the decision, but not exclusive control. Students care more than parents about classroom experience and future earnings. Ignoring the dual‐agent nature of the decision can weaken predictions and lead to poorly targeted policy designs. (JEL I21, J24, D13)

Estimating local average treatment effects in aggregate data

ABSTRACT In some contexts, the effect of a treatment can be estimated with easily accessible aggregate rather than individual data, using difference-in-difference estimation. However, under imperfect assignment within groups, this produces intent-to-treat estimates, which may not be the treatment effect of interest. This article provides a method for estimating local average treatment effects using aggregate data. I also suggest a data source that allows the method to be applied when treatment rates are not recorded.