34r

LABG

New Member
#1
A student wants to investigate whether people from northern USA are more severely affected by winter depression ("seasonal affective disorder") than people from southern USA. Using a questionnaire, the research fellow collected data from 20 people from northern USA and 20 people from southern USA.


I think this is a one tail test,
H0: u=0
H1: u > 0

Am I correct?? thank you in advance..
 

Karabiner

TS Contributor
#3
Do not perform one-tailed tests. In the present case, it would be impossible to determine if
perhaps the effect is the other way around.

With kind regards

Karabiner
 

LABG

New Member
#4
Do not perform one-tailed tests. In the present case, it would be impossible to determine if
perhaps the effect is the other way around.

With kind regards

Karabiner

Could you explain more why you think we should use two sided t-test?
I was thinking that since the student in the question above specifically is looking for if northern USA "have more severe winter depression" than Southern USA our Ho has to be u= 0 and H1: u > 0
 

Karabiner

TS Contributor
#5
A working hypothesis which contains a direction does not imply a one-sided hypothesis.
You very rarely see this. Why? Because it seems un-scientific to make oneself voluntarily
blind on one eye. It is not impossible that an effect has the opposite direction.

With kind regards

Karabiner
 
Last edited:

LABG

New Member
#6
A working hypothesis which contains a direction does not imply a one-sided hypothesis.
You very rarely see this. Why? Because it seems un-scientific to make oneselfu voluntarily
blind on one eye. It is not impossible that an effect has the opposite direction.

With kind regards

Karabiner
But if I was to use a two sided t-test then I would just find out if there is a difference between these two variables not if depresion in Northern USA is more severe than Southern USA???? what do you think about that.
 

Karabiner

TS Contributor
#7
WhaI am not sure what your problem is. If you reject H0 ("proportion in the South=proportion in the North"), then in the next step
you have a look at the descriptive statistics and see in which direction the effect was.

With kind regards

Karabiner
 

fed2

Active Member
#9
correct answer was one sided. Its obviously academic question, you get the points for 'one-sided', dollars to doughnuts. I like points, some people don't i guess.
 

hlsmith

Less is more. Stay pure. Stay poor.
#10
Both @fed2 and @Karabiner are correct. In practice, it would be two-sided, for teaching purposes, the instructor is likely looking for one-sided.

Dang, now I want a doughnut.
 

Karabiner

TS Contributor
#11
S/He did not bother to mention any context, so why should I bother about points?

But even if it is a homework question, we do not know the preferences of
the instructors. Plus, it is probably not wrong to give some advice on how this
usually dealt with in real life.

With kind regards

Karabiner
 

Karabiner

TS Contributor
#13
Agreeing with people for the sake of getting your points is most def what real life is about.
We still do not know for which answer s/he would get more points.
One-sided reseacrh questions are routinely dealt with by using
two-tailed tests. Maybe the instructors have the same in mind.
 

fed2

Active Member
#14
if the intended answer to this question was "two-sided impossible to determine if perhaps the effect is the other way around." i honestly don't want to know. Who knows what kind of demented teachers there are out there? maybe joeb is teaching...