Let's consider two groups, 150 and 15, with variability in each group, say
an SD of 2.
Let's look at the power of a "not significant" t test result at
significance level (alpha) = 0.05 (two-tailed)
Choose the smallest difference between the means that you think might be
interesting or important, and read the power of the trial to detect that
difference.
<NOBR
Power (%)
2.37
99
1.99
95
1.79
90
1.66
85
1.55
80
1.46
75
1.37
70
1.22
60
1.08
50
0.94
40
0.79
30
Regards,
Stan Alekman
In a message dated 11/14/2009 7:11:54 A.M. Eastern Standard Time,
[log in to unmask] writes:
Dear All
I have two groups named Middle Managers (MM) and Top Managers (TM) and I
want to test whether the mean differences between the two are significant
or
not.
My issue is that Group 1 (MM) has 150 respondents (10 respondents from one
organization each) while Group 2 (TM) has 15 respondents (One response
from
each organization- Because Top Managers are fewer in numbers in a given
organization)
I plan to use T-test for testing the significance but am worried about the
disparity between sample size of the two groups...The data has already
been
collected...Do I need to increase the sample size?
Would welcome any advice/suggestions in this regard...
--
Dr. Bilal Mustafa Khan
Assistant Professor
Department of Business Administration
Aligarh Muslim University (AMU)
India
0091-9897310838
|