Stata Lecture 3

Question 1

What 3 things must you consider when deciding how to analyse data?

Answer 1

Is the data normally distributed or skewed?
Are you comparing two groups or more?
Are the groups independent or are they paired?

Question 2

What is the difference between independent groups vs paired groups?

Answer 2

Independent groups come from separate populations (e.g. how men vote vs how women vote)
Paired groups are drawn from the same population (e.g. womens views on politics vs womens views on art)

Question 3

What are 3 possible reasons for incurring a high p value?

Answer 3

There truly is no difference between the two groups
The sample was atypical and not representative of the population
The sample was not large enough for the data to be deemed significant

Question 4

3 Disadvantages of non-parametric statistical tests?

Answer 4

Do not provide a confidence interval
Based on rankings, not scores
Less sensitive than parametric test - best to use parametric test if assumptions hold for it

Question 5

How do you check if data is normally distributed?

Answer 5

Box and Whisker plot for the data in each individual group

Question 6

What are the assumptions of all parametric statistical tests?

Answer 6

Normally distributed data

Standard deviation is similar across all groups

Question 7

3 reasons when could you potentially use a parametric test for skewed data?

Answer 7

When there are 50 observations in each group
The skewness is not very large
The standard deviations are similar across groups

Question 8

How to calculate variance?

Answer 8

Standard deviation squared

Question 9

How do you determine if standard deviation is similar across groups?

Answer 9

The variance (standard deviation squared) should be no more than 4 times larger the variance of any other group

Question 10

When would you use a non-parametric test?

Answer 10

Data is skewed
Small sample size
Standard deviation differs across groups

Question 11

Which test would you use for:
parametric
2 groups
independent

Answer 11

Two sample/unpaired t-test

Question 12

Which test would you use for:
parametric
2 groups
paired

Answer 12

Paired t-test

Question 13

Which test would you use for:
parametric
3 or more groups
independent

Answer 13

ANOVA (analysis of variance test)

Question 14

Which test would you use for:
parametric
3 or more groups
paired

Answer 14

Repeated measures analysis of variance

Question 15

Which test would you use for:
non-parametric
3 or more groups
independent

Answer 15

Kruskal-Wallis test

Question 16

Which test would you use for:
non-parametric
2 groups
independent

Answer 16

Mann Whitney test

Question 17

Which test would you use for:
non-parametric
2 groups
paired

Answer 17

Wilcoxon signed rank

Question 18

Which test would you use for:
non-parametric
3 or more groups
paired

Answer 18

Friedman test