finish the saying… spearmans and chi
must be high, all the rest must be lower
what test is for independent measures and nominal data
chi square
what test for independent measures and ordinal data
mann u whit
test for repeated measures and nominal data
sign
test for repeated measures and ordinal data
wilcox
test for relationship ordinal data
spearman ro
how to work out binominal sign test
ignore ppts who have the same result
give either a + or - to each ppt depending on whether there was a positive or negative change between them
identify least frequent occuring sign and count no. of times this appeared
this is calculated value
how to work out spearmans rho?
Rank the data for each variable separately- first rank the data for Variable A
Give the lowest value a rank of 1
Find the difference between rank A and rank B for each participant. (it does not matter if some of the differences are negative because the next step cancels these out).
Square the difference between ranks for each participant
Calculate the Rs value (the correlation coefficient)
Compare the calculated r value to the critical value table
If the calculated value is less than the critical value, the result is not significant.
how to work out wilcoxon
ignore scores of ppt who scored the same
calculate the difference between each score
rank the differences
add together the ranks fpr least frequent sign and call this T
look at critical value table
how to calculate mann whittney
Rank the scores from both groups rank order
Add up the total of the ranks for each condition separately
Use the formulae to calculate the two U values
The lowest of the two U values is the calculated value of U
Consult the table of critical values. Compare U value to the critical value table to see if the results are significant at the level of 0.05.
in mann u whittney what happens If the calculated value is equal to or less than the critical value
the results are significant and you can accept the alternative hypothesis.
in mann u whittney what happens If the calculated value is equal to or less than the critical value
the results are significant and you can accept the alternative hypothesis.
chi square how do you work it out?
rank the cells a-d
Total row of cells and column cells
total of all cells
CELL A/B/C/D= FIND TOTAL USING E=RC/T
Make a table and compare the raw data and expected frequencies
(O-E) then (O-E) squared then (O-E) sqaured / E
Total up the last column which gives calculated value
Use critical value table
how do you work out the variance and standard deviation
you first of all need to calculate the mean score for each condition in the experiment.
For each participant, you then subtract the mean score from their score. This is ‘d’ (the difference). Don’t worry if the number is positive or negative!
Then, you square each ‘d’ score. (d2)
Add up all the d2
Divide this total by n-1
Divide this value by n-1
strengths and weaknesses of variance
Takes into account all values in the data set
Less likely to be affected by outliers (by calculating differences)
More time consuming to calculate
Not in the same units as the original measure (e.g now in cm²)
standard deviation strengths and weaknesses
Same units as original measure
Easy to calculate if variance already done!
Takes into account extreme outliers
when do u use parametric tests
when you have interval data
when theres normal distribution
all groups have similar variances
when data is normally distributed what test do we use
t test
How is the critical value used to determine if the findings are statistically significant?
It is compared to the appropriate/correct calculated/observed value.
Explain what p > 0.05 would mean if it appeared as part of the significance statement when reporting the findings from this study.
Greater than 5% probability that the results are due to chance.
Observed value was greater than the critical value, therefore
results are not significant.
The results are not significant/there is no significant difference.
what are the two hypothesis for mann whitney
null: two samples, ranks do not differ sigificantly
alternitive: ranks do differ significantly
two critera for using parametric tests to analyse data
no extreme scores
interval data
scores normally distributed
