I find it very useful to watch whole short program events from the first couple JGPs of each season, where most of the skaters are unknown to me and they're skating according to a random draw, and try to score the program components as well as I can based on the video.
I'm not being influenced much by reputation (I admit I do expect the Japanese and Russian skaters to do well and I usually have heard of the Americans and have some idea what to expect).
And then afterward I check the protocols to see how often I was in range with the panel and where I deviated. Especially for Skating Skills, if I was out of line I assume that something got lost between live and video.
I haven't done a statistical analysis of the JGP scores, but my impression is that the average of component scores tends to cluster around high 4s: lots of scores in the 4s, many in the 5s, and usually some outliers in the 6s and maybe 3s. At some events we see outstanding junior skaters earning 7s and even 8s, and occasionally there are skaters from small countries on the circuit who really aren't at a junior skill level earning scores in the 2s . . . or more rarely 1s and even 0s.
I think it's a good exercise for trying to see what constitutes average, above- or below-average, or especially good or weak junior skating. And to see where they are mostly not quite up to the standard of the elite seniors. Occasionally an exceptional skater will look like they could absolutely hold their own with senior GP skaters PCS-wise (or only in some areas), and those would be the exceptions that might inspire lots of 7s and maybe some 8s. But most junior skaters aren't at that level -- and if they're already over 16, that's exactly why they're competing in junior and not senior.
