WP 4 Statistics - The p value and Basic statistical tests

Question 1

to chose a test we must ask

Answer 1

how many samples (how many groups do we have)
how big is the sample
what type of data
is the data parametric

Question 2

what is parametric data

Answer 2

normally distributed
equal variances
no extreme scores

main parameters…. mean and sd

Question 3

equal sample size usually means

Answer 3

equal variance

Question 4

methods of comparing 2 samples (one person can either be in group A or B , not both)
SCALE
parametric
non-parametric

Answer 4

independent t test (compares mean of each group)

mann whitney U test (compare median of each group)

Question 5

methods of comparing 2 samples
ordinal
any sample size

Answer 5

mann whitney U test (compares median of each group)

Question 6

methods of comparing 2 samples
nominal
large sample , most expected frequency > 5
small sample

Answer 6

• chi-squared test or fishers exact (2x2) (compares proportions in each category)
• reduce number of catergories (compares ‘’)

Question 7

Methods for differences in one or paired sample :SCALE
parametric
non parametric

Answer 7

• pair t test (comparing means of differences in the sample)

- Wilcoxon (comparing ranks of differences)

Question 8

Methods for differences in one or paired sample : ordinal

any

Answer 8

sign test (comparing larger /smaller /same)

Question 9

Methods for differences in one or paired sample : nominal

any

Answer 9

McNemar’s test (comparing marginal probabilities)

Question 10

t-distribution

when do we use

Answer 10

n < 60

comparing the mean values of a variable in different groups.

Question 11

Paired tests

Answer 11

repeated measures on THE SAME PATIENT eg before and after

An example of a within group study

You are looking for the mean of the differences

more powerful than

Question 12

Independent sample t-test

Answer 12

Two separate/independent groups (one group control and other…)

An example of a between groups study

You are looking for the difference between the means.

Question 13

P-values

H0 means that

Answer 13

no difference between the means of the two samples (t-test)

but when run tests we will find a difference between the means

Question 14

p-values

what caused the difference

Answer 14

random chance

real effect

Question 15

p value defintion

Answer 15

probability that our results came about by random chance,

given that the null hypothesis is true.

Question 16

what each p value means

Answer 16

p > 0.05 (5%) → not significant (NS)

p < 0.05 → Reasonable evidence against H0

Question 17

Correlations

when is it used

Answer 17

to determine the strength of the relationship between two different variables
• E.g. ice cream sales and temperature

Question 18

regression asks

Answer 18

What is the relationship between variables?

Question 19

Regression tests for

Answer 19

causality (relationship between cause and effect)