Selecting the Most Appropriate Statistical Test - Interval Data

Question 1

Which statistical test would you use if you wished to compare means of interval data, but the data was not normally distributed?

Answer 1

Would have to drop down to ordinal data flowchart and continue there.

Question 2

Which statistical test would you use if the data was interval, was normally distributed, and was 2 groups of independent data?

Answer 2

Student t-test (or ANOVA or MANOVA)

Question 3

Which statistical test would you choose if the data was interval, was nromally distributed, and had 2 groups of paired data?

Answer 3

Paired t-test

(or Repeated Measures ANOVA or Repeated Measures MONAVA)

Question 4

Which statistical test would you choose if the interval data was normally distributed, had 3 or more groups of independent data, and had > 1 DV?

Answer 4

MANOVA; if confounding is present use MANCOVA

Question 5

Which statistical test would you choose if the interval data was normally distributed, had 3 or more groups of independent data, but had 1 DV?

Answer 5

ANOVA; if confounding is present use ANCOVA

Question 6

Which statistical test would you choose if the interval data was normally distributed, had 3 or more groups of paired data, and had > 1 DV?

Answer 6

Repeated Measures MANOVA; if condounding is present use Repeated Measures MANCOVA

Question 7

Which statistical test would you choose if the interval data was normally distributed, had 3 or more groups of paired data, and has 1 DV?

Answer 7

Repeated Measures ANOVA; if confounding is present use Repeated Measures ANCOVA

Question 8

Which post-hoc tests would you use if the ANOVA or MANOVA show statistical significance for 3 or more groups of independent data?

Answer 8

• Bonferroni Test of Inequality
  
  • adjusts the p-value for # of comparisons being made
  • very conservative
• Tukey
  
  • compares all pairwise comparisons possible
  • ALL groups must be equal in size
  • slightly more conservative than Student-Newman-Keul
• Scheffe
  
  • compares all pairwise comparisons possible
  • ALL groups must be equal in size
  • less affected by violations in normality and homogeneity of variances
  • MOST Conservative
• 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 all groups are NOT of equal size

Question 9

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

Answer 9

Linear Regression

Question 10

Which statistical test would you choose if the data was interval, you did not wish to compare means, you did not want to assess for associations or predict group membership, but you wished to assess the data for correlation?

Answer 10

Pearson Correlation

Question 11

Which statistical test would you choose if the data was interval, you did not wish to compare means, you did not want to assess for associations or predict group membership, you did not wish to assess the data for correlation, but you wanted to assess the data for time to event or survival?

Answer 11

Kaplan-Meier product-limit estimate