Matching methods are widely used in educational research to estimate causal effects when randomization is not feasible. These techniques pair treated units (such as schools receiving an intervention) with similar control units based on observable characteristics. However, current statistical inference procedures for these methods can produce…

Background: Effectiveness studies that ignore costs can lead to misguided policies. For instance, reducing class sizes gained traction due to positive outcomes on student achievement (Konstantopoulos & Li, 2012; Krueger, 1999), yet cost-effectiveness studies revealed the high-cost relative to its modest benefits (Brewer et al., 1999). More…

The difference in differences design is widely used to assess treatment effects in natural experiments or other situations where random assignment cannot, or is not, used (see, e.g., Angrist & Pischke, 2009). The researcher must make important decisions about which comparisons to make, the measurements to make, and perhaps the number of…

This paper proposes an extension of linear equating that may be useful in one of two fairly common assessment scenarios. One is where different students have taken different combinations of test forms. This might occur, for example, where students have some free choice over the exam papers they take within a particular qualification. In this…

Background/Context: Single-case experiment designs (SCEDs) are experimental designs in which a small number of cases are repeatedly measured over time, with manipulation of baseline and intervention phases. Because SCEDs often rely on direct behavioral observations, count data are common. To account for both the clustering and the non-normal…

Background: Double/debiased machine learning (DML) methods have been proposed to overcome the regularization bias from the naive approach of ML methods (Chernozhukov et al., 2018). DML methods use a partialling-out approach which removes the effect of confounders from both the treatment and outcome and then regresses the residualized outcome on…

We consider the cascading effects of researcher decisions throughout the process of quantifying racially polarized voting (RPV). We contrast three methods of estimating precinct racial composition, Bayesian Improved Surname Geocoding (BISG), fully Bayesian BISG, and Citizen Voting Age Population (CVAP), and two algorithms for performing ecological…

Background: One of the most vexing threats to education field trials is attrition--when some subjects drop out before it is complete. Since outcomes are not available for subjects who attrit, they are typically dropped from any analysis estimating average effects on the outcome. However, since attrition may itself have been affected by treatment…

This article presents methods for using extant data to improve the properties of estimators of the standardized mean difference (SMD) effect size. Because samples recruited into education research studies are often more homogeneous than the populations of policy interest, the variation in educational outcomes can be smaller in these samples than…

Research in education and behavioral sciences often involves the use of latent variable models that are related to indicators, as well as related to covariates or outcomes. Such models are subject to interpretational confounding, which occurs when fitting the model with covariates or outcomes alters the results for the measurement model. This has…

Although there is an extensive methodological literature on the measurement of intergenerational income mobility, there has been limited research on the conceptual interpretation of mobility measures and the methodological implications of those interpretations. In this article, I focus on the three measures of mobility most frequently used in the…

Item parameter estimation using an item response theory (IRT) model with fixed ability estimates is useful in equating with small samples on anchor items. The current study explores the impact of three ability estimation methods (weighted likelihood estimation [WLE], maximum a posteriori [MAP], and posterior ability distribution estimation [PST])…

Prior research has suggested that clustered regression discontinuity designs are a formidable alternative to cluster randomized designs because they provide targeted treatment assignment while maintaining a high-quality basis for inferences on local treatment effects. However, methods for the design and analysis of clustered regression…

This study compared Stepwise Logistic Regression (Stepwise-LR) and three machine learning (ML) methods--Classification and Regression Trees (CART), Random Forest (RF), and Generalized Boosted Modeling (GBM) for estimating propensity scores (PS) applied in causal inference. A simulation study was conducted considering factors of the sample size,…

Sensitivity analysis, a statistical method crucial for validating inferences across disciplines, quantifies the conditions that could alter conclusions (Razavi et al., 2021). One line of work is rooted in linear models and foregrounds the sensitivity of inferences to the strength of omitted variables (Cinelli & Hazlett, 2019; Frank, 2000). A…