cookiemonster
05-02-2011, 03:09 AM
I would like to ask all of you about some queries.
Apologies for my poor English.
I tried to search for previous posts and also Google but cannot find a clearer explanation.:shakehead
Information : Survey 3 companies regarding level of Job satisfaction and social anxiety
Company A : 96 respondents
Company B : 120 respondents
Company C : 135 respondents
Divided into demographic information (Percent) of each company
Gender (Company A, B, C)
Male 32, 25, 70
Female 68, 75, 30
Marital Status
Single 47, 35, 22
Not-single 53, 65, 78
Work
Blue-Collar 35, 22, 48
White-Collar 65, 78, 52
+ The result of Levene’s Test and normality find that there are not violated assumptions.
+ All companies are similar in fundamental information and three companies are dependent.
+ The sampling is a purposive way and response rate is 80%
H1-H3 : There are differences on job satisfaction in terms of their gender, marital status and work
H4 : There are differences of their social anxiety among employees working in 3 companies.
:yup:
Some questions I would like to ask you guys :
Q1. How large the difference between the groups can be judged as unequal sample sizes ? The larger one should more than half or ….. ?
Q2. The size I have to recognize about unequal sample sizes is the variable of each hypothesis (H1 : Male/Female) rather than the total respondent (96/120/135)? Or both has to take into account.
Q3. Should I employ Mann Whitney U instead of t-test? And the Kruskal-Wallis is more appropriate than One-way Anova, isn’t it? As I have found some texts explain that unequal sample size is not a big deal for the large sample. I am confused now :mad: Should I go on with Para or Non-parametric?
Please help me improve myself as I am not good at statistics and
always avoid studying Stats. I am now designing to kick off this weakness.
Hope you all give me any suggestions. Thank you in advance.
Apologies for my poor English.
I tried to search for previous posts and also Google but cannot find a clearer explanation.:shakehead
Information : Survey 3 companies regarding level of Job satisfaction and social anxiety
Company A : 96 respondents
Company B : 120 respondents
Company C : 135 respondents
Divided into demographic information (Percent) of each company
Gender (Company A, B, C)
Male 32, 25, 70
Female 68, 75, 30
Marital Status
Single 47, 35, 22
Not-single 53, 65, 78
Work
Blue-Collar 35, 22, 48
White-Collar 65, 78, 52
+ The result of Levene’s Test and normality find that there are not violated assumptions.
+ All companies are similar in fundamental information and three companies are dependent.
+ The sampling is a purposive way and response rate is 80%
H1-H3 : There are differences on job satisfaction in terms of their gender, marital status and work
H4 : There are differences of their social anxiety among employees working in 3 companies.
:yup:
Some questions I would like to ask you guys :
Q1. How large the difference between the groups can be judged as unequal sample sizes ? The larger one should more than half or ….. ?
Q2. The size I have to recognize about unequal sample sizes is the variable of each hypothesis (H1 : Male/Female) rather than the total respondent (96/120/135)? Or both has to take into account.
Q3. Should I employ Mann Whitney U instead of t-test? And the Kruskal-Wallis is more appropriate than One-way Anova, isn’t it? As I have found some texts explain that unequal sample size is not a big deal for the large sample. I am confused now :mad: Should I go on with Para or Non-parametric?
Please help me improve myself as I am not good at statistics and
always avoid studying Stats. I am now designing to kick off this weakness.
Hope you all give me any suggestions. Thank you in advance.