applied statistics:
note:
There will be no alternative hypothesis of the equal sign
Even accepting H0 hypothesis can not prove H0 hypothesis, assuming that this role is to reject H0 hypothesis thereby accepting H1 hypothesis, so it should be the first to write H1 hypothesis.
When the true value and the estimated value very similar, beta can be large, map:
FIG apparent from the above derivation n becomes large data more focused to a value, the smaller becomes β
The whole process is hypothesis testing: statement H0H1 hypothesis, select the check method of sampling to determine the level of significance, then compare and determine whether there are statistically significant. Does not necessarily have a significant statistically meaningful, since n is small, a large beta], i.e., a second type of error has 90% of the time is meaningless.
P advantage is that the value of the error can be obtained directly and significantly limit the extent of the value of the first type, may be directly with 0.05 or 0.01 Comparative
The overall proportion of testing large samples need to meet two conditions:
