stats tests Flashcards
stats test where observed value has to be less than the critical value to be significant
- sign test
- Wilcoxon test
- Mann Whitney test
stats test where observed value has to be higher than the critical value to be significant
- chi-squared test
- spearman’s rho
acronym for stats test table
NO Is Charlotte Mad Really/Maybe She’s Worried Shhh C.
N = nominal data
O = ordinal data
I = independent measures
C = chi squared
M = Mann Whitney
R/M = repeated measure/matched pairs
S = sign test
W = Wilcoxon test
S = spearman’s rho
C = correlation
Nominal data
Data which is a tally/total of the number of times something occurred
ordinal data
Data which can be put in rank order, from smallest to largest (but the spaces between the ranks may not be the same size)
independent measures design
this involves using different people in each condition
repeated measured design
this involves using the same people in each condition
matched groups design
this involves using different people in each condition but an attempt is made to make the participants as similar as possible on certain key characteristics
sign test (nominal data + repeated measures/matched pairs design)
- data in table to show scores for each ppt in each condition
- difference between the scores for each ppt
- ignore ‘0’ differences
- sign of difference in 3rd column (+ve or -ve)
- add up number of each +ve and -ve
- smaller is the observed value
- compare observed to critical value in critical value table
- observed must be less than critical to be significant
chi-squared (nominal + independent measures design)
- calculate degree of freedom ((no of rows - 1)x(no of column-1))
- add totals of columns
- add totals of rows
- E = row total x column total / overall rows totals
- observed values for each cell = O
- E-O
- square O-E
- divide O-E squared by the expected frequency
- add scores together = observed value
- use degrees of freedom to to look up critical value in table
Wilcoxon (repeated measures design + ordinal data)
- write data in table comparing ppts score for each condition
- calculate difference between scores for each ppts
- ignore ‘0’ differences
- rank differences
- sum of +ve differences
- sum of -ve differences
- smaller is the observed value
- N= number of differences (so not 0)
Mann Whitney U (independent measures + ordinal data)
- call smaller group 1 and larger group 2
- rank ALL scores as one group
- find sum of ranks in Group 1 and sum of ranks in group 2
- apply equation to group 1 + 2
- observed value smaller than critical
spearman’s rho (correlation + ordinal data)
- rank data for each co-variable individually
- add the ranks to table
- calculate diff between ranks for each ppt
- ignores cares with same ranks
- square the differences
- sum of squared differences
- insert results into equation
- observed value larger than critical
(remember tp add + to - based on he results)
