1
$\begingroup$

Any help is much appreciated!! Im just querying some p values generated in SPSS from some data i have recently collected. The three groups were type 1 diabetes, type 2 diabetes and both together (type 1 and 2). Independent t tests were used to analyse depression (normal/ abnormal), fatigue (none/ severe) in relation to sedentary behavior. E.g is higher depression associated with increased sitting time. After collecting data for each group, i have come up with this data:

DEPRESSION: Type 1= 0.039 , Type 2= 0.05 , Type 1 and 2= <0.01

FATIGUE Type 1= 0.122 , Type 2= 0.012 , Type 1 and 2= <0.01

My question is how can the both group (type 1 and 2) be highly statistically significant if the other groups aren't? Does this look correct? Thank you!!

$\endgroup$
2
  • 2
    $\begingroup$ The p-values for depression and fatigue for type 1 and type 2 diabetes were significant or nearly significant. When you combine the two types you get a larger sample size and hence the p-value could easily go further down. $\endgroup$ Apr 7, 2018 at 23:14
  • $\begingroup$ "Independent t tests were used to analyse depression (normal/ abnormal), fatigue (none/ severe) in relation to sedentary behavior." IT is vague. What are your hypotheses ? and objectives ? please be specific . $\endgroup$
    – user10619
    Apr 11, 2018 at 16:47

1 Answer 1

1
$\begingroup$

an example with the same (more clear) effect:

        a=0   a=1    mean
b=0     11    39     25 
b=1     41    9      25
mean    26    24     25

Note that the mean values are very similar, $\mu_{a=0} \sim \mu_{a=1}$ and $\mu_{b=0} \sim \mu_{b=1}$, all around 25.

But the corner-cells are very different from the mean cell value 25. So if you compare 'a=1 and b=1' with 'a != 1 or b !=1' the you will find a big difference.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.