statistical tests

Question 1

Unpaired T test (Mann-Whitney U-test)

Answer 1

2 independent samples from same population
(plasma glucose level between intervention and placebo group)

Question 2

Paired T-Test (Wilcoxon Matched Pars Test)

Answer 2

2 sets of observations on a single sample
(plasma glucose level in intervention and placebo group before and after intervention)

Question 3

One way analysis of variance (F test) (Kruskall-Wallis analysis of variance by ranks)

Answer 3

generalisation of paired T-test where three or more observation sets are made on a single sample
(plasma glucose levels between depression, BPAD and schizophrenia)

Question 4

Two way analysis of variance (analysis of variance by ranks)

Answer 4

same as a once way analysis of variance but tests the influence of two different covariates

(plasma glucose levels between BPAD, depression and schizophrenia and whether they differ in male and female subjects)

Question 5

Chi squared test (fisher’s exact test)

Answer 5

Test null hypothesis that the distribution of a discontinuous variable is the same in two or more independent samples

Distribution free test to investigate significant difference between two groups

(cannabis use differs between intervention and placebo)

Question 6

product moment correlation coefficient (pearson’s r) (Spearman’s rank correlation coefficient

Answer 6

assess strength of straight line association between two continuous variables (glucose related to triglycerides in diabetic patients)

Question 7

Regression by least squares

Answer 7

numerical relation between two quantitative variables (how PEF rate varies with height)

Question 8

multiple regression by least squares

Answer 8

numerical relation between dependent variable and several predictor variables (age, weight, sodium intake determines blood pressure)

Question 9

long rank

Answer 9

non-parametric test to compare the survival times and any significant differences between two or more groups in a Kaplain Meyer (survival) Curve