Improved Monte-Carlo method vs. hit-and-miss method

#1
I am not understanding *Which is more accurate, the hit-and-miss method or the improved Monte-Carlo method?*

http://www-stat.stanford.edu/~susan/courses/s208/node14.html it is written that that the hit-and-miss has a higher variance but they showed
(variance of improved Monte-Carlo) - (variance of hit-and-miss method)>0
\(\sigma^2_M-\sigma^2_H>0\)

Doesn't it imply improved Monte-Carlo method has a higher variance?

Can you please explain it experimentally to me with an easy example ? Such as for a small integration.