I feel that the difference between the residents and the extenders stems from the fact that the extenders were trained to do a specific task, presumably by the same attendings who would final sign the studies. The RRAs likely were trained to cater to a specific style, in contrast to residents who learn from multiple different attendings, divisions, and texts. Remember, this study specifically looked at time to final sign and not overall accuracy (which was the same among residents and RRAs).
Differences among residency programs would likely be negligible. If you are training someone to specifically emulate your own style and verbage, you are obviously going to sign off on their report faster versus someone who may say the same thing differently. It seems specious to criticize the residents when the outcome measured does not reflect their accuracy.
For what it's worth, I cannot find the original JACR article. Perhaps they pulled it?