Potential bug in control of alpha level in covariate.predictions function

[jcarlis3]

Hi Jeff,

Based on the help in covariate.predictions, the alpha argument should control the predicted CIs, with alpha = 0.025 being a 95% CI. When I change the alpha value, the width of the resulting CI never changes. Is this a bug, or am I missing something? I noticed there are 4 lines in covariate.predictions.R where 1.96*SE is hard-coded in, so perhaps any alpha specification is being overruled?

Thanks,
Jason

[jcarlis3]

Just realized David already reported this issue on phidot. Sorry for the cross-posting. Any resolution?
http://www.phidot.org/forum/viewtopic.php?f=21&t=3526