This repository contains code for the paper "Sampling for computational efficiency when conducting analyses in big data," In Press at Epidemiology.

CODE DESCRIPTION

Step 1. Generate the simulated data using churn_1_data.R

Step 2. Run the analyses

• Estimate rate using crude approaches: churn_2_crude-rate.R
• Estimate risk using crude approaches: churn_2_crude-risk.R
• Estimate risk and rate using weights (not including Z): churn_2_wt.R
• Estimate risk and rate using weights (including Z): churn_2_wt-z.R
• Estimate risk and rate using multiple imputation (not including Z): churn_2_mi.R
• Estimate risk and rate using multiple imputation (including Z): churn_2_mi-z.R
• Estimate crude risk allowing for interval censoring: churn_2_ic.R

Step 3. Combine results across approaches and scenarios using churn_3_stack.R

Step 4. Summarize results in tables and figures

• Build Table 1: churn_4_table.R
• Build figures for crude results: churn_4_fig-unadj.R
• Build figures for adjusted results: churn_4_fig-adj.R