CI Width Differences With cband = TRUE/FALSE #244
Description
My understanding is that confidence intervals with cband = TRUE should always be wider than those with cband = FALSE, since the simultaneous bands use a larger critical value. However, in my analysis with an unbalanced panel, I noticed that for some outcomes, the confidence intervals become wider when I set cband = FALSE compared to the default. This does not happen for all outcomes, only for some, which is confusing. Should we be concerned when the CIs become wider after turning off the simultaneous band?