I found this breakdown of JPL's CNEOS data to be every helpful. The close approach time chart shows % chance of impact vs day. From this I take it that we just didn't have enough info at the time of publishing.