Lay people are pretty good at estimating the relative contribution of genes and environment to human traits (correlation with actual estimates is .77). Women are better at it than men, especially if they have had multiple children:
Lay estimates of genetic influence correlate strongly (r = .77) with published heritability estimates
"Isn’t it obvious to anyone w more than one child...that people are born with certain talents and temperaments?" 2nd sentence, Blank Slate. Now confirmed by Emily Willoughby, former student James Lee et al in study of laypeople's concept of heritability.
People tend to underestimate the heritability of obesity, and to overestimate the heritability of sexual orientation and breast cancer. Overall, though, people's estimates of the heritability of traits are fairly accurate.
A group of 1000+ MTurk workers were rather good at estimating how heritable various traits are (according to current best published estimates)
People (on MTurk) are pretty good at estimating the heritability of many traits, but overestimate heritability of sexual orientation
"lay estimates of genetic influence … Despite apparent ideological associations with these beliefs, the correspondence between mean lay estimates and published heritability estimates for the surveyed traits is large (r = .77)"
Study finds that lay estimates of the heritability of various traits are fairly accurate: the correlation between lay estimates and published estimates is 0.77.
Recent paper suggested that lay estimates of the heritability of various traits wasn't that off. But you may be thinking of different issues.
Watching multiple siblings grow up provides the best intuition about heritability: “…educated mothers with multiple children emerge as particularly accurate in their estimates of the genetic contribution to these traits” via