Project MUSE®: Observational Studies - Latest Articles
https://muse.jhu.edu/journal/793
Project MUSE®: Latest articles in Observational Studies.daily12026-05-12T00:00:00-05:00text/htmlen-USVol. 1 (2015) through current issueLatest Articles: Observational StudiesTWOProject MUSE®Observational Studies2767-3324Latest articles in Observational Studies. Feed provided by Project MUSE®On Identification of Optimal Dynamic Treatment Regimes with Proxies of Hidden Confounders
https://muse.jhu.edu/article/985142
<p></p>
Estimating single time point optimal treatment regimes and multiple time point dynamic treatment regimes has received much attention in a diverse array of fields, including bio-statistics, computer science, and economics. When data are obtained from a randomized controlled trial, A/B test, or a sequential multiple assignment randomized trial (SMART), the assumptions required for identification and estimation of optimal rules/regimes can be met by design. Alternatively, when one wishes to estimate the optimal regime from observational data, unverifiable assumptions such as the no unmeasured confounding assumption (NUCA) are typically invoked in point exposure settings and its time-varying analog termed sequential
... <a href="https://muse.jhu.edu/article/985147">Read More</a>
Project MUSE®https://muse.jhu.edu/2026-05-12T00:00:00-05:00https://muse.jhu.edu/journal/793/image/coversmallOn Identification of Optimal Dynamic Treatment Regimes with Proxies of Hidden Confounders2026-03-14text/htmlen-USOn Identification of Optimal Dynamic Treatment Regimes with Proxies of Hidden Confounders2026-03-142026TWOProject MUSE®1804872026-05-12T00:00:00-05:002026-03-14Causal Mediation Analysis for Effect Heterogeneity
https://muse.jhu.edu/article/985143
<p></p>
Both mediation analysis (Pearl, 2001, 2009, 2012; VanderWeele, 2009, 2013, 2014, 2015; Valeri, 2012; Imai et al., 2009, 2010; VanderWeele and Robins, 2007; Valeri and VanderWeele, 2013, 2014; Lange et al., 2012; Hayes, 2022) and modification analysis (Wager and Athey, 2018; Hernán and Robins, 2020; Chmura Kraemer et al., 2008; Imai and Strauss, 2011; Imai and Ratkovic, 2013) shed light on the underlying mechanisms of cause-effect relationships; however, these analyses address fundamentally different questions. Mediation analysis assesses the pathways whereby an exposure affects an outcome through mediators; whereas modification analysis evaluates the extent to which an exposure has varying effects on an outcome
... <a href="https://muse.jhu.edu/article/985147">Read More</a>
Project MUSE®https://muse.jhu.edu/2026-05-12T00:00:00-05:00https://muse.jhu.edu/journal/793/image/coversmallCausal Mediation Analysis for Effect Heterogeneity2026-03-14text/htmlen-USCausal Mediation Analysis for Effect Heterogeneity2026-03-142026TWOProject MUSE®2236932026-05-12T00:00:00-05:002026-03-14Matching with Multiple Criteria and Its Application to Health Disparities Research
https://muse.jhu.edu/article/985144
<p></p>
Carefully-designed observational studies may provide a useful piece of evidence towards establishing or nullifying a scientifically meaningful association or causal conclusion. Observational studies often suffer from both overt and hidden bias due to systematic differences in pretreatment covariates between two comparison groups (Rosenbaum, 2002). The bias may lead to researchers concluding a spurious association or a cause-and-effect relationship after conducting a naïve outcome analysis comparing two groups. Statistical matching is a widely used nonparametric approach to reducing the overt bias by adjusting for observed differences in covariates between comparison groups. The overall guiding principle of
... <a href="https://muse.jhu.edu/article/985147">Read More</a>
Project MUSE®https://muse.jhu.edu/2026-05-12T00:00:00-05:00https://muse.jhu.edu/journal/793/image/coversmallMatching with Multiple Criteria and Its Application to Health Disparities Research2026-03-14text/htmlen-USMatching with Multiple Criteria and Its Application to Health Disparities Research2026-03-142026TWOProject MUSE®1548262026-05-12T00:00:00-05:002026-03-14Outcome-Assisted Multiple Imputation of Missing Treatments
https://muse.jhu.edu/article/985145
<p></p>
In observational studies, the treatment status of some study subjects may be unknown. For example, the treatment of interest may be an environmental exposure or health-related behavior that is measured only for a subset of the study subjects (Lee et al., 2021; Mitra, 2023; Feldman et al., 2024; Chen, 2025). Or, when the data for the observational study come from a survey, some individuals may not respond to the question that defines the treatment status (Molinari, 2010). In such cases, analysts can improve the accuracy of causal inferences by accounting for (i.e., not disregarding) the missingness in the treatment status, especially when the treatments are not missing completely at random.A flexible and convenient
... <a href="https://muse.jhu.edu/article/985147">Read More</a>
Project MUSE®https://muse.jhu.edu/2026-05-12T00:00:00-05:00https://muse.jhu.edu/journal/793/image/coversmallOutcome-Assisted Multiple Imputation of Missing Treatments2026-03-14text/htmlen-USOutcome-Assisted Multiple Imputation of Missing Treatments2026-03-142026TWOProject MUSE®2104352026-05-12T00:00:00-05:002026-03-14Synthetic Regressing Control
https://muse.jhu.edu/article/985146
<p></p>
The synthetic control (SC) method is a popular approach of evaluating the effects of policy changes. It allows estimation of the impact of a treatment on a single unit in panel data settings with a modest number of control units and with many pre-treatment periods (Abadie and Gardeazabal, 2003, and Abadie et al., 2010). The key idea under the SC method is to construct a weighted average of control units, known as a synthetic control, that matches the treated unit's pre-treatment outcomes. The estimated impact is then calculated as the difference in post-treatment outcomes between the treated unit and the synthetic control. See Abadie (2021) for recent reviews.The SC method utilizes constrained optimization to solve
... <a href="https://muse.jhu.edu/article/985147">Read More</a>
Project MUSE®https://muse.jhu.edu/2026-05-12T00:00:00-05:00https://muse.jhu.edu/journal/793/image/coversmallSynthetic Regressing Control2026-03-14text/htmlen-USSynthetic Regressing Control2026-03-142026TWOProject MUSE®4100742026-05-12T00:00:00-05:002026-03-14Monotonicity Test: An R Package for Efficient Nonparametric Monotonicity Testing
https://muse.jhu.edu/article/985147
<p></p>
Testing for monotonicity of a function has been a longstanding problem in statistics and is a popular subject in econometrics, particularly when the data of interest involve measurements over time e.g., daily stock prices. However, more generally, monotonicity in some form is a common assumption/condition underlying many statistical methods. Often, the assumption pertains to a true but unknown function that is estimated based on available data. For example, in surrogate marker evaluation research, monotonicity is fundamental for evaluating the risk of the surrogate paradox (Chen et al., 2007; VanderWeele, 2013). The surrogate paradox occurs when a treatment has a positive effect on a surrogate marker, the surrogate
... <a href="https://muse.jhu.edu/article/985147">Read More</a>
Project MUSE®https://muse.jhu.edu/2026-05-12T00:00:00-05:00https://muse.jhu.edu/journal/793/image/coversmallMonotonicity Test: An R Package for Efficient Nonparametric Monotonicity Testing2026-03-14text/htmlen-USMonotonicity Test: An R Package for Efficient Nonparametric Monotonicity Testing2026-03-142026TWOProject MUSE®1086782026-05-12T00:00:00-05:002026-03-14