Selecting the Most Appropriate Statistical Test - Ordinal Data

Question 1

Which statistical test would you choose if the data was ordinal, you wished to compare frequencies or proportions, and there was 2 groups with that were independent of each other?

Answer 1

Mann-Whitney

Question 2

Which statistical test would you choose if the data was ordinal, you wished to compare frequencies or proportions, and there was 3 or more groups that had independent assessments of each other?

Answer 2

Kruskal-Wallis

Question 3

Which statistical test would you choose if the data was ordinal, you wished to compare frequencies or proportions, and there was 2 groups that had paired data?

Answer 3

Wilcoxon Signed Rank

Question 4

Which statistical test would you choose if the data was ordinal, you wished to compare frequencies or proportions, and there were 3 or more groups that had paired data?

Answer 4

Friedman Test

Question 5

When would the Kruskal-Wallis and Friedman tests be appropriate to use? Also, if they both showed a statistcal significance in the data, what statistical tests would be appropriate to use next?

Answer 5

• Student-Newman-Keul
  
  • compares all pairwise comparisons possible
  • ALL groups must be equal in size
• Dunnett
  
  • compares all pairwise comparisons against a single control
  • ALL groups must be equal in size
• Dunn
  
  • compares all pairwise comparisons possible
  • useful when groups are NOT of equal size

** All these tests are used in ordinal data for post-hoc tests for groups of 3 or greater **

• Would use the Kruskal-Wallis if groups of 3 or greater and the data was independent.
• Would use Friedman if groups of 3 or greater and the data was paired.

Question 6

Which test would you use if the data was ordinal, you did not wish to compare frequencies or proportions, but you wanted to assess for associations or predict group membership?

Answer 6

Multinomial Logistic Regression

Question 7

Which statistical test would you use if the data was ordinal, you did not wish to compare frequencies or proportions, you did not want to assess for associations or predict group membership, but you wished to check the data for correlation?

Answer 7

Spearman Correlation

Question 8

Which statistical test would you use if the data was ordinal, you did not wish to compare frequencies or proportions, you did not want to assess for associations or predict group membership, you didn’t want to check for correlation, but you wished to assess for proportion with event (survival)?

Answer 8

Cox Proportional Hazards