Normal or not?

#1
hi, my first post on this site. im in need of desperate help. currently going through my dissertation process which includes stats which im hopeless at. i think i have completed everything but just wanted to check something. some of my tests involves comparing and observing relationships between 2 variables. therefore, i know i need to use spearmans or pearsons test. however, my normality test gave to different significance for each variable (.067 and .522). As one assumes normality and the other doesn't what shall I do? Assume normality anyway?

If someone can reply asap that will be great. Thanks x
 

JohnM

TS Contributor
#2
Neither of those significance values are below .05, so you can say with 95% confidence that your data does not depart from a normal distribution.
 
#3
Apologies I typed in the wrong figures. If the normality tests have shown one variable to be 0.18 and the other .522 should I assume normality or not?

Also having conducted a t test should I decide on significant using the 2 tailed significant p value? Sorry to sound really clueless.
Thanks
 
#4
Sig.
.003
.127

Thats another normality test I have that has confused me as one shows normal distribution and the other not normal. Which should I assume?
 

JohnM

TS Contributor
#5
xjennax said:
Apologies I typed in the wrong figures. If the normality tests have shown one variable to be 0.18 and the other .522 should I assume normality or not?
The .18 and .52 are above .05, so you can assume normality.

xjennax said:
Also having conducted a t test should I decide on significant using the 2 tailed significant p value? Sorry to sound really clueless.
Thanks
That depends on your hypothesis.

Is it directional, like this (requires the one-tail prob):

Ho: mu1 = mu2
Ha: mu1 > mu2

or is it non-directional, like this (requires the two-tail prob):

Ho: mu1 = mu2
Ha: mu1 not= mu2
 

JohnM

TS Contributor
#6
xjennax said:
Sig.
.003
.127

Thats another normality test I have that has confused me as one shows normal distribution and the other not normal. Which should I assume?
I wouldn't be concerned unless every normality test for that variable shows a significance of less than .05.