Practical 3: Questionnaires

Question 1

Structure of a Questionnaire:
Part 1: Demographic information (Contextual Data)
=> This info will help you put your results into a context (e.g. experienced female auditors tend to respond yes)
- give each suggested answer a numerical value to do analysis later
e.g. Sample: Auditors
Gender: Male / Female (Dummy variable!)
Experience: 1-5 yrs / 6-15 / 16-25 / over 25
Qualifications: ACA / ACCA / … / other
Seniority: Assistant manager/ manager/ senior manage/ partner

Answer 1

Structure of a Questionnaire:
Part 1: Demographic information (Contextual Data)
=> This info will help you put your results into a context (e.g. experienced female auditors tend to respond yes)
- give each suggested answer a numerical value to do analysis later
e.g. Sample: Auditors
Gender: Male / Female (Dummy variable!)
Experience: 1-5 yrs / 6-15 / 16-25 / over 25
Qualifications: ACA / ACCA / … / other
Seniority: Assistant manager/ manager/ senior manage/ partner

Question 2

Questionnaire structure - Part 2: Questions of ?(derived from ?):
- ? (? questions/ ? / ?):
> Questions are linked to the factors/ variables we are examining

e.g. Transparency is 1 factor so 1 of the questions can be:
To what extent does regulation affect transparency?
Answer choices:
Strongly agree/ agree/ neutral / disagree / strongly disagree

Answer 2

Questionnaire structure - Part 2: Questions of Discussion (derived from Literature):
- Factorisation (Research questions/ variables/ trends):
> Questions are linked to the factors/ variables we are examining
e.g. Transparency is 1 factor so 1 of the questions can be:
To what extent does regulation affect transparency?
Answer choices:
Strongly agree/ agree/ neutral / disagree / strongly disagree

Question 3

Questionnaire structure - Part 3:
?? questions:
- maximum ? questions!
e.g. In general, do you think that regulations can develop more in regards to Control / Transparency/ Communication (the factors)?

Answer 3

Questionnaire structure - Part 3:
Open ended questions:
- maximum 3 questions!
e.g. In general, do you think that regulations can develop more in regards to Control / Transparency/ Communication (the factors)?

Question 4

If the survey is in multiple languages, ???translation must be done:
e.g. I’ll translate the questionnaire into Vietnamese and then ask someone else to translate it back from Vietnamese to English.

Answer 4

If the survey is in multiple languages, back-to-back translation must be done:
e.g. I’ll translate the questionnaire into Vietnamese and then ask someone else to translate it back from Vietnamese to English.

Question 5

Sample:
1. Who are ? ? 
e.g. Auditors/ Regulators/ bankers
2. ? size:
- ? questionnaire: 
e.g. Auditors - 100 / R - 100 / B - 100
- Responses: minimum ?responses to be statistically significant!
e.g. 40 / 35 / 45
- Response rate:
e.g. 40%
> To check if the response rate is good enough, compare it to the ?.

Answer 5

Sample:
1. Who are responders? 
e.g. Auditors/ Regulators/ bankers
2. Sample size:
- Distributed questionnaire: 
e.g. Auditors - 100 / R - 100 / B - 100
- Responses: minimum 30 responses to be statistically significant!
e.g. 40 / 35 / 45
- Response rate:
e.g. 40%
> To check if the response rate is good enough, compare it to the literature.

Question 6

Test of normality:

• 1.96 <= Skewness <= 1.96
• 3 <= Kurtosis <= 3

Answer 6

Test of normality:

• 1.96 <= Skewness <= 1.96
• 3 <= Kurtosis <= 3

Question 7

If data is normal, use Parametric test (t-test)

=> check for Significant / Insignificant differences between different groups of respondents.

Answer 7

If data is normal, use Parametric test (t-test)

=> check for Significant / Insignificant differences between different groups of respondents.

Question 8

If data is not normal, use non-parametric test:
> Use Mann-Whitney test or aka Wilcoxon rank-sum test) to compare 2 groups
Do the t-test for each question separately
> Krascal Wallis test to compare more than 2 groups
—-
But it’s better to use Man-whitney test.

Answer 8

If data is not normal, use non-parametric test:
> Use Mann-Whitney test or aka Wilcoxon rank-sum test) to compare 2 groups
Do the t-test for each question separately
> Krascal Wallis test to compare more than 2 groups
—-
But it’s better to use Man-whitney test.

Question 9

In Stata, t-test result is given at:
‘Ha: diff !=0
Pr(|T| > |t|) = …’

If t-test result is:
< 0.01 (1%): *** highly significant;
< 0.05 (5%) : ** moderate significant;
< 0.1 (10%): * ;

Answer 9

In Stata, t-test result is given at:
‘Ha: diff !=0
Pr(|T| > |t|) = …’

If t-test result is:
< 0.01 (1%): *** highly significant;
< 0.05 (5%) : ** moderate significant;
< 0.1 (10%): * ;