On statistical methods for labor market evaluation under interference between units [Elektronisk resurs]

• Evaluation studies aim to provide answers to important questions like: How does this program or policy intervention affect the outcome variables of interest? In order to answer such questions, using the traditional statistical evaluation (or causal inference) methods, some conditions must be satised. One requirement is that the outcomes of individuals are not affected by the treatment given to other individuals, i.e., that the no-interference assumption is satisfied. This assumption might, in many situations, not be plausible. However, recent progress in the research field has provided us with statistical methods for causal inference even under interference. In this paper, we review some of themost important contributions made. We also discuss how we think these methods can or cannot be used within the field of policy evaluation and if there are some measures to be taken when planning an evaluation study in order to be able to use a particular method. In addition, we give examples on how interference has been dealt within some evaluation applications including, but not limited to, labor market evaluations, in the recent past. 

Ämnesord

Natural Sciences  (hsv)

Mathematics  (hsv)

Probability Theory and Statistics  (hsv)

Naturvetenskap  (hsv)

Matematik  (hsv)

Sannolikhetsteori och statistik  (hsv)

Statistics  (umu)

statistik  (umu)

Genre

government publication  (marcgt)

Indexterm och SAB-rubrik

causal effect

causal inference

contagion effect

direct and indirect effects

evaluation studies

neighborhood effect

peer effect

peer influence effect

policy intervention

spillover effect

SUTVA

treatment effect