Propensity score techniques and the assessment of measured covariate balance to test causal associations in psychological research

By V. Harder, Elizabeth A. Stuart, J. Anthony. Psychological methods, 2010

787 citations recorded by the source corpus.

There is considerable interest in using propensity score (PS) statistical techniques to address questions of causal inference in psychological research. Many PS techniques exist, yet few guidelines are available to aid applied researchers in their understanding, use and evaluation. This study gives an overview of available techniques for PS estimation and PS application. It also provides a way to help compare PS techniques, using the resulting measured covariate balance as the criterion for selecting between techniques. The empirical example for this study involves the potential causal relationship linking early-onset cannabis problems and subsequent negative mental health outcomes, using data from a prospective cohort study. PS techniques are described and evaluated based on their ability to balance the distributions of measured potentially confounding covariates for individuals with and without early-onset cannabis problems. This paper identifies the PS techniques that yield good statistical balance of the chosen measured covariates within the context of this particular research question and cohort.

Subfields: Statistics and Probability

View Paper: https://www.semanticscholar.org/paper/acf0b73bcbbbcecf407f6d7a83cf6aee89be8aad

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Statistics and Probability

Citations in the corpus

Matching methods for causal inference: A review and a look forward (2010, background): “For example, if all adolescent heavy drug users are also heavy drinkers, it will be impossible to separate out the effect of heavy drug use from the effect of heavy drinking.”
Relationship Between Entrepreneurship Education, Entrepreneurial Mindset, and Career Readiness in Secondary Students (2020, methodology): “…alternatives, we used the suggested decision criterion of standardized bias, or the standardized difference of means of the propensity score, of less than 0.25 for all variables, ensuring that the most important variables had standardized bias less than 0.10 (Harder et al., 2010; Ho et al., 2007).”
Co-Curricular Connections: The Role of Undergraduate Research Experiences in Promoting Engineering Students’ Communication, Teamwork, and Leadership Skills (2016, methodology): “This imbalance of the measured covariates motivates the use of PS techniques to help bring into balance the distributions of the measured covariates (Harder et al., 2010).”
On the Unnecessary Ubiquity of Hierarchical Linear Modeling (2017, methodology): “Propensity weights were then stabilized using the method outlined in Harder, Stuart, and Anthony (2010).”
Examining Group Walks in Nature and Multiple Aspects of Well-Being: A Large-Scale Study (2014, neutral): “First, while the PSM method ensured there were no significant group differences on measured covariates, it remains possible that differences existed on unmeasured confounding variables (Harder et al., 2010).”
A Tutorial on Propensity Score Estimation for Multiple Treatments Using Generalized Boosted Models (2013, methodology): “It has automated variable selection [5] and, in simulations, has proven to provide more stable weights than parametric models [17, 18].”