- Bayesian inference
- Causal inference
- Estimands
- Inference
- Linear regression
- Logistic regression
- Longitudinal and clustered data
- Measurement error / misclassification
- Meta-analysis
- Miscellaneous
- Missing data
- Randomized controlled trials
- Stata
- Survival analysis
Bayesian inference
- Bayesian inference: are parameters fixed or random?
- My first foray with Stan
- Automatic convergence checking in Bayesian inference with runjags
- ‘Monte-Carlo sensitivity analysis’ recommended against
Causal inference
- Potential outcomes, counterfactuals, causal effects, and randomization
- Conditional randomization, standardization, and inverse probability weighting
- Estimating causal effects from observational studies
- Using hazard ratios to estimate causal effects in RCTs
- Estimating effects when outcomes are truncated by death
- Why you shouldn’t use propensity score matching
- Causal interpretation of the hazard ratio from RCTs when proportional hazards holds
- PhD in estimands/causal inference in trials (UK/EU)
- Confounding vs. effect modification
- From DAGs to potential outcomes via Single World Intervention Graphs
- G-formula for causal inference via multiple imputation
Estimands
- Comments on FDA guidance ‘Adjusting for Covariates in Randomized Clinical Trials for Drugs and Biological Products’
- Hypothetical estimands – a unification of causal inference and missing data methods
- Causal (in)validity of the trimmed means estimand
- Is the ICH E9 estimand addendum compatible with model-based estimands?
- Estimating hypothetical estimands with causal inference and missing data estimators in a diabetes trial
Inference
- Standard deviation versus standard error
- The miracle of the bootstrap
- The difference between the sample mean and the population mean
- A/B testing and Pearson’s chi-squared test of independence
- The t-test and robustness to non-normality
- Wald vs likelihood ratio test
- A/B testing – confidence interval for the difference in proportions using R
- Wilcoxon-Mann-Whitney as an alternative to the t-test
- Adjusting for optimism/overfitting in measures of predictive ability using bootstrapping
- Banning p-values from journals
- On “The fallacy of placing confidence in confidence intervals”
- Frequentists should more often consider using Bayesian methods
- Independence of sample mean and sample variance
- P-values after multiple imputation using mitools in R
- Interpretation of frequentist confidence intervals and Bayesian credible intervals
- Does a Bernoulli/binomial model really assume everyone has the same probability p?
Linear regression
- Assumptions for linear regression
- Regression inference assuming predictors are fixed
- Linear regression with random regressors, part 2
- The robust sandwich variance estimator for linear regression (theory)
- R squared and adjusted R squared
- R squared and goodness of fit in linear regression
- The robust sandwich variance estimator for linear regression (using R)
- Interval regression with heteroskedastic errors
- R squared/correlation depends on variance of predictor
- The mean of residuals in linear regression is always zero
Logistic regression / Generalized linear models
- R squared in logistic regression
- The Hosmer-Lemeshow goodness of fit test for logistic regression
- Deviance goodness of fit test for Poisson regression
- Area under the ROC curve – assessing discrimination in logistic regression
- Checking functional form in logistic regression using loess plots
- Comparing predictive ability of two nested logistic regression models
- Interpreting odds and odds ratios
- Why shouldn’t I use linear regression if my outcome is binary?
- The mean of the residuals in logistic regression is always zero
Longitudinal and clustered data
- Robustness of linear mixed models
- Odds ratios, collapsibility, marginal vs. conditional, GEE vs GLMMs
- Mixed model repeated measures (MMRM) in Stata, SAS and R
- MMRM vs LME model
- Mixed models repeated measures (mmrm) package for R
Measurement error / misclassification
- Adjusting for covariate misclassification in logistic regression – predictive value weighting
- Using Stata’s sem to adjust for covariate measurement error
- What does correlation in a Bland-Altman plot mean?
- What might the true sensitivity be for lateral flow Covid-19 tests?
Meta-analysis
- Fixed versus random-effects meta-analysis – efficiency and confidence interval coverage
- Meta-analysis of last week’s Scottish independence polls
- Meta-analysis of Scottish independence polls (update 17th September 2014)
- Prediction intervals after random-effects meta-analysis
Miscellaneous
- Test for Alzheimer’s (allegedly) no better than a coin toss
- A statistician’s initial experiences of Git/GitHub
- Prof. Alan Agresti on modelling ordinal data, issues with Wald based inferences, and GEE for ordinal data
- Machine learning vs. traditional modelling techniques
- Running simulations in R using Amazon Web Services
- What’s the difference between statistics and machine learning?
- 3 year post-doc in Bath – Clinical trial estimands – from definition to estimation
- PhD on causal inference for competing risks data
Missing data
- When is complete case analysis unbiased?
- Multiple imputation with interactions and non-linear terms
- Multiple imputation using random forest
- Missing covariates in structural equation models
- When is complete case/records logistic regression unbiased?
- Including the outcome in imputation models of covariates
- Substantive model compatible imputation of covariates – smcfcs in R
- Multiple imputation followed by deletion of imputed outcomes
- Missing covariates in competing risks analysis
- Weighting after multiple imputation for MNAR sensitivity analysis not recommended
- smcfcs in R – updated version 1.1.1 with critical bug fix
- On the missing at random assumption in longitudinal trials
- Multiple imputation for missing covariates in Poisson regression
- Combining bootstrapping with multiple imputation
- Imputing missing covariates in nested case-control and case cohort studies
- Maximum likelihood multiple imputation
- smcfcs – non-linear relationships between covariates
- Multiple imputation when estimating relative risks
- Missing not at random sensitivity analysis with FCS multiple imputation
- Combining bootstrapping and multiple imputation under uncongeniality
- Critical bug fix for smcfcs in Stata
- Understanding missing at random dropout using DAGs
- Reference based imputation for continuous missing data in R with bootstrap inference
- Bootstrapping multiple imputation using multiple cores/processors in R
- Conference talk video – Bootstrap Inference for Multiple Imputation Under Uncongeniality and Misspecification
- Imputation of covariates for Fine & Gray cumulative incidence modelling with competing risks
- P-values after multiple imputation using mitools in R
- Convergence plots for smcfcs in R
- Auxiliary variables and congeniality in multiple imputation
- Is MAR dropout classified as MNAR according to Mohan and Pearl?
- Reference based multiple imputation – what’s the right variance and how to estimate it?
- Multiple imputation separately by groups in R and Stata
- smcfcs imputation in R – now with parallel functionality
- Summary statistics after imputation with mice
- Conditional mean reference-based multiple imputation
- How many imputations with mice? Assessing Monte-Carlo error after multiple imputation in R
- Multiple imputation with splines in R using smcfcs
- Perfect prediction handling in smcfcs for R
- Multiple imputation for missing baseline covariates in discrete time survival analysis
- Multiple imputation and its application – 2nd edition published
- Variance estimation for reference-based multiple imputation – the debate continues
- Multiple imputation for missing covariates in the Fine & Gray model for competing risks
- Multiple imputation with flexible parametric survival models
Randomized controlled trials
- Adjusting for baseline covariates in randomized controlled trials
- Leveraging baseline covariates for improved efficiency in randomized controlled trials
- Clustering in randomized controlled trials
- Robustness to misspecification when adjusting for baseline in RCTs
- Is the two sample t-test/ANOVA really biased in RCTs?
- Improving efficiency in RCTs using propensity scores
- Multiarm trials – should we allow for multiplicity?
- Matching analysis to design: stratified randomization in trials
- Estimands and assumptions in clinical trials
- Confidence intervals for the hazard ratio in RCTs which agree with log rank test
- Covariate adjustment and prediction of mean response in randomised trials
- Live stream seminar 26th October 2017: Covariate adjustment and prediction of mean response in randomised trials
- Randomisation as the basis for inference in trials
- Comment on ‘Conditional estimation and inference to address observed covariate imbalance in randomized clinical trials’
- ANCOVA in RCTs – model based standard errors are valid even under misspecification
- Robustness of ANCOVA in randomised trials with unequal randomisation
- The hazards of period specific and weighted hazard ratios
- Is stratified randomisation in trials (at least large ones) pointless?
- ‘An introduction to covariate adjustment in trials’ – PSI covariate adjustment event
- On improving the efficiency of trials via linear adjustment for a prognostic score
Stata
- Stata-Mata’s st_view function – use with care!
- Why I think Stata’s old xi: prefix is still useful
- Estimating risk ratios from observational data in Stata
Survival analysis
- Interpreting changes in hazard and hazard ratios
- Multiple imputation for informative censoring R package
- Handling competing risks in randomized trials
- Testing equality of two survival distributions: log-rank/Cox versus RMST
- A simulation introduction to censoring in survival analysis
- What can we infer from proportional hazards?
- Non-proportional hazards – an introduction to their possible causes and interpretation
- Does the log rank test assume proportional hazards?