Missing at Random (MAR) means that the probability of data being missing is purely random after conditioning on observed data (e.g., in multiple imputation). That is to say, it depends only on the observed data, not the missing values themselves.
Dr. Guy Prochilo 🏳️🌈
- 281 Followers
- 56 Following
- 43 Posts
Psychological Scientist | Lecturer of Research Methods | Fourth Year Psychology (Honours) Program Coordinator at ISN Psychology | http://guyprochilo.com | https://www.linkedin.com/in/guyprochilo/
Just wrapped up today's lecture on missing data! Covered item- and construct-level missingness, detection, mechanisms, analysis (using Welch's t-tests with missingness indicators), solutions, and a theory intro to multiple imputation (since #jamovi lacks modules).
Don't use a significance test of heterogeneity (eg Cochran's Q) to choose between fixed or random effects meta-analysis. If studies estimate a single common effect size, use fixed effects. If studies differ in populations, measures, or protocols, opt for random effects.
A fixed effects meta-analysis assumes a single common effect size across all studies, which is often unrealistic given differences in interventions, populations, and measurements. In psychology especially, a random effects meta-analysis is usually the more plausible
#phdchat
#phdchat
A one-stage IPD meta-analysis aggregates data across all studies and can be analyzed using a linear mixed model with the study ID as a random effect. This aligns with a random effects meta-analysis where we assume the true effect size differs across studies #phdchat
IPD Meta-Analyses can be analyzed using a one-stage or two-stage method. A one-stage method analyzes raw data while accounting for within-study clustering, while a two-stage method calculates effect sizes for each study and combines them like a traditional meta-analysis #phdchat
The idea of Double Hermeneutic in psychology is fascinating—when people learn about findings, they might change their actions, altering the effect. Meanwhile, cells and rocks remain unaware of the latest research trends, untroubled by the burden of self-reflection. #phdchat
The Zeigarnik effect suggests we remember unfinished tasks best, creating a mental burden. Case in point: that email I didn’t reply to? It’s been living rent-free in my head for 8 months. If a colleague doesn’t reply, take comfort—they might be thinking of you daily.
Predictive Mean Matching is a multiple imputation method that fills missing data using values already in your dataset with similar predicted values. This ensures imputations are realistic and consistent with the original data, making it ideal for most missing data cases.
In my studies I encounter item- and construct-level missing data. Handle item-level missingness (e.g., missing response in a 10-item survey) by imputing with the mean of available items. But use multiple imputation for construct-level (e.g., entire variable) missingness.
#phdchat
#phdchat