This function decomposes the change in means between consecutive time periods into three components: composition effects (changing group shares), within-group change (groups shifting their outcomes), and an interaction effect (covariance of share and outcome changes).
Arguments
- data
A data frame containing group-level summary statistics.
- tname
The column name for the time variable.
- gname
The column name for the group variable.
- yname
The column name for the group mean of the outcome.
- sname
The column name for the group's population share at each time period, where shares should sum to 1 within each period.
- tol
The tolerance for the accounting identity check.