Data Analysis I (Week 5)

Question 1

What is Cronbach’s alpha and its formula? What’s the value for items to be considered to have a high internal consistency?

Answer 1

Measures inter-item reliability of multi-item scales

Formula: alpha = (N* r ̅ ) / (1 + (N-1)*r ̅ )

r: average inter-item correlation
N: number of items

r >0.6 = high internal consistency

Question 2

What happens when Cronbach’s alpha is <0.6?

Answer 2

There are some variables which are not related to your construct

Get a table: Item-Total Statistics (see: Cronbach’s alpha if item deleted)

Leave out an item, check again, if necessary leave out another item etc.

Q: Did you measure 1 construct or multiple? Possible to create 2 variables instead of 1.

Question 3

What is the purpose of hypothesis testing?

Answer 3

To make a judgment about the difference b/w two sample statistics or b/w the sample statistic and a hypothesised pop. parameter

Question 4

What is Type I error?

Answer 4

Reject Ho when Ho is true

Question 5

What is Type II error?

Answer 5

Accept (i.e. Don’t reject) Ho when Ho is false

Question 6

What is confidence interval?

Answer 6

̅X ± sampling error = Interval estimate of μ

Question 7

What are the 2 ways to draw conclusions for hypothesis testing?

Answer 7

1) Confidence interval
- If value is outside confidence interval = Reject Ho
- E.g. Does not contain 0

2) p-value
- If p-value < (1 - confidence level) = Reject Ho

Question 8

What is the purpose of the one-sample t-test?

Answer 8

Does the pop. mean differ significantly from μ?

Two-tailed:
Ho: μ = 4
H1: μ ≠ 4

One-tailed: (Direction)
Ho: μ = 4
H1: μ > 4

Question 9

What is the paired sample t-test? What is the conclusion for paired sample t-test?

Answer 9

Both samples are from the SAME people

Are these two pop. parameters significantly diff. from each other?

Ho: μ1 = μ2 (There is no relationship)
H1: μ1 ≠ μ2 (There is a relationship)

Conclusion: On average, there is/is no difference between μ1 and μ2 at the population level.

Question 10

What is the independent sample t-test?

Answer 10

Both samples are from DIFFERENT people (i.e. indpt)

Are these two pop. parameters significantly diff. from each other?

Ho: μ1 = μ2
H1: μ1 ≠ μ2

Conclusion: Reject Ho = Group 1 and Group 2 are independent samples and e.g. males are > females.

Question 11

What is correlation analysis?

Answer 11

Used to explore relationship b/w variables

BIVARIATE relationships (only 2 variables at a time)

Typically hope to see:

• Sig. r/s b/w x & y variables
• Not v. strong r/s b/w x variables