Link: Statistical test
Alternative hypotheses
The opposite of Null hypothesis.
Null hypothesis: There’s no difference in recovery times between drug C and D.
Alternative hypothesis: There is a difference in recovery times between C and D.
Left: represents null hypothesis, the distance around the single mean d(n) Right: represents alternative hypothesis, the distance around the two separate means d(a)
If d(a) < d(n), it suggests using two means makes more sense -> we reject the null hypothesis
An intuitive way to think about why it makes more sense.
The distribution that are closer to normal distribution usually makes more sense. In this example, the distribution of group C and of D in alternative hypothesis are more closer to normal distribution (compare the data points vs the mean)
However, we only reject or fail to reject the null hypothesis/primary hypothesis. There are too many alternatives out there and it’s hard to say which one can be accpeted.