Please explain why do we use a 2 sided test?
Since the point of T-value test is to determine if the modifications to the website increase average session time compared to the control website.
So why perform a 2-sided test instead of a right-sided test where we can determine if the variation website average session time > control average session time?
Isn’t the goal is to determine if the increase in average session time in the newly designed website is statistically sufficiently strong enough to say it is better than control version?
Rather, our goal is NOT just to determine if the change in average session time is statistically significant enough to say there is a change in average session time?
Isn’t the 2 sided test will just tell you if the change is statistically sufficiently strong enough to declare the new website caused a change (neither improvement or degradation in average session time, just a change)