Lower
Accept lower p-values (meaning lower in magnitude; values tending toward zero).--And don't forget that by reducing the probability of getting a type I error, you increase the probability of getting a type II error (inverse relationship).
Yes. It can be higher or lower
It's a lower-case Greek sigma followed by a superscript 2, in other words, "sigma-squared".
Before conducting a significance test, the statistician will choose an alpha level. Depending upon the severity of having type I or type II error, the statistician will make the alpha level higher or lower. Generally in courts, the alpha level is .05. The other common alpha levels for significance tests are .10 and .01.
The lower the better
I would have thought this blindingly obvious but no matter, a lower percentage error is better because it means your approximation to a solution is closer to the real answer than an approximation with a higher error.
If you are receiving interest on an assett, a higher interest is better. If you are paying interest on a debit, a lower interest is better.
Higher. 4.0 is the best.
Higher
Yes, the Higher the Better.
Lower
its better
Higher
A better drop weight is one that has a lower drop. -3 is the most powerful bat.
Standby times reflect overall battery capacity, the higher the better.
A lower accident frequency rate is better as "frequency" means "how often something occurs".