METHODS - statistical testing

Question 1

why use a staistical test

Answer 1

whether a difference/association found in an investigation is statistically significant (occured by chance or real effect)

Question 2

3 criteria to choose a test

Answer 2

1) difference or correlation
2) experimental design related or non-related
3) level of measurement

Question 3

table of statistical tests

Answer 3

unrelated related correlation
nominal chi-squared sign test chi-squared
ordinal mann-whitney wilcoxon spearman’s rho
interval unrelated t-test related t-test Pearson’s r

Question 4

nominal data

Answer 4

categories

e.g people naming their favourite football team

Question 5

ordinal data

Answer 5

data is collected on a numerical, ordered scale but this scale is subjective (for this reason it lacks precision)
e.g people rating something 1-10

Question 6

interval data

Answer 6

based on numerical scales that include units of equal, precisely defined size (better than ordianl as data not converted to ranks)
includes counting observations and any ‘public’ unit of measurement

Question 7

requirements for a parametric test

Answer 7

interval data
normally distributed sample
homogeneity of variance (standard deviations are similar)

Question 8

if the results are found to be non-significant…

Answer 8

the null hypothesis is excepted (no difference or no correlation)

Question 9

significance level

Answer 9

the point at which the null hypothesis is accepted or rejected

Question 10

usual level of significance

Answer 10

0.05/5%

meaning the probability that the observed effect happened by chance is equal to or less than 5%

Question 11

calculated and critical

Answer 11

calculated value is compared with a critical value

Question 12

3 criteria for using table of critical values

Answer 12

1) direction of hypothesis
2) number of participants
3) level of significance

Question 13

type 1 error

Answer 13

the null hypothesis is rejected and the alternative hyopthesis is accepted when the null hypothesis is ‘true’
false positive as a significant difference found when one does not exist

Question 14

type 2 error

Answer 14

the null hypothesis is accepted but the alternative hypothesis is ‘true’
false negative

Question 15

what makes each error more likely

Answer 15

type 1 - significance level is too lenient 0.1/10%

type 2 - significance level is too stringent 0.01/1%

Question 16

degrees of freedom for parametric tests

Answer 16

unrelated t-test - (no. of ppts in A + no. in B) - 2
related t-test - no. of ppts -1
pearson’s r - no. of ppts -2