What we do see, anecdotally and in formal studies, is that if you vary only the variable of age, it makes a substantial difference.
I personally discovered that during a job hunt where someone pointed out I should drop my age from my resume, extra controlled as it were because for a very very long time I looked 20 something (a bit beyond age 50).
No I didn't. I had group 1, 2.a, and 2.b all along. I think that group 1 may suffer discrimination.
To elaborate the 'may', I'd stipulate that group 1 probably has a harder time getting interviews than a younger person with the same resume. I don't think it is necessarily discrimination though. I think there's a legit reason someone who's 40 who has the same resume as someone who's 20 would be less attractive.
To be succinct, the guy who's 20 may be the next Jeff Dean or Steven Wozniak. The 20 year old has 'he may be the next Jeff Dean'-option value. The 40 year old ... if he's been programming for 20 years and he's not Jeff Dean ... he's probably not turning into Dean now. No 'he may be the next Jeff Dean'-option value.
A study that has somehow controlled for continued learning? I'm interested because I'd be genuinely surprised if that's the case.