Thanks @simonbrown, @olibclarke ! I spoke with our microscopist and he says that he aims the defoci between 1 and 2um with 0.25um increments. Is it surprising that the average defocus misses this intended range by as much as is observed in this dataset especially on the lower defocus end?
The above image also gets to your comment/prediction @simonbrown about those images close to focus being affected disproportionately by the water ring: it seems to my eye that the distribution across CTF estimates at lowest defocus is most equal among the calculated defocus values: it spans the entire range of CTF estimates.
@olibclarke I performed CTF estimation again but this time increasing max resolution from the default 4A to 4.5A and the problem worsened in that more of the images were estimated at CTF values worse than 5A (i.e. excluding data worse than 5A would lead to retention of 43% of the data with 4.5A cut-off whereas with 4A cut-off, excluding data worse than 5A would lead to retention of 55% of the data) . I will test out the defocus range search in another CTF estimation job; where could I see the 1D search plot?