Inferential Statistics/maths Flashcards
Test used with nominal data, independent measures design
Chi square
Test with ordinal data and independent measures design
Mann-Whitney U-test
Test with nominal data and repeated measures/matched participants
Binomial sign test
Test using ordinal data and repeated measures/matched participants design
Wilcoxon signed ranks
Test used ordinal data correlation study
Spearman’s Rho
Test used interval/ratio correlation study
Pearson’s product moment
Para and non parametric tests
Parametric tests are used when there is:
Interval/ratio level data (t tests and Pearson’s product moment)
A normal distribution of data
Similar variances between results from the different conditions (homogeneity of variance)
First step chi square
Add up row and column totals and overall totals
Second step chi square
Write in the observed frequency
Step 3 chi square
Work out expected frequency
Row total X column total divided by overall total
Fourth step chi square
Observed frequency minus expected frequency
Fifth step chi square
Square the observed frequency minus expected
Sixth step chi square
After you square the observed-expected
You divide this answer by the expected frequency
Seventh step chi square
X squared = add up last column
Last column was where you divide by the expected frequency
Chi square: X squared from the table is…
Calculated value
Chi square: how to calculate degrees of freedom
(Number of rows-1) X (number of columns-1)
Always 1
Chi square: how do you find critical value on the table
Look at level of significance needed at the top (columns)
Look at degrees of freedom at the side (rows)
For chi square and Spearman’s Rho, what’s needed to be significant? (Confirm alternative hypothesis)
Calculated value bigger than critical value
Binomial sign text what’s needed to show significant difference? (Confirm alternative hypothesis)
Critical must be bigger than calculated value
Measures of dispersion
Range, variance, standard deviation
Variance method
- Calculate mean
- Original score - mean (gives ‘d’ difference)
- Square it (d2)
- Add all new d squared values and divide by number of participants minus 1
(Like mean but minus one from number of participants before you divide)
Standard deviation method
Square root of variance
Difference between bar graph and histogram
Gaps in bar charts (non continuous data)
Histogram used with interval/ratio data
Histogram Y axis frequency density Frequency density= frequency/ class width
Frequency given by area of the bar, not height in histogram
Bar charts bars can be unrepresentative especially in unequal categories
Binomial sign test method
- Put plus or minus indication direction of difference
- Add up least frequent sign to get calculated value
- Use level of significance+one/two tailed hypothesis (column) and number of participants (rows) to find critical value
- To show level of significance (confirm hypothesis) critical value must be bigger than calculated value
Experiment hypothesis v correlation hypothesis language
Experiment ‘difference’
Correlation ‘relationship’
Strength of variance
Takes into account all data collected
Histogram eval
S- representative as shows correct proportions
Bar chart eval
W- unrepresentative bars don’t always show correct proportions
How to work out frequency density for histogram
Frequency divided by class width (This is how tall bar will be)
What needs to be included for statement of significance
- Operationalised hypothesis
- level of significance (eg. P smaller or equal to 0.05)
- explain why (is calc value higher or lower than critical)
Find ratios of given amount
Eg. 2:7 of £45
Divide value by total rations Multiply by ration you want to find out Eg. 45/9 = 5 5 X 7 = 35 5 X 2 = 10
Find fraction of given amount
Divide by denominator
Multiply by numerator
Pie chart
Data for each group divided by total data
X 360
= each sector
Conversion between %, fractions & decimals
% into decimal = divide by 100
% into fraction = over 100 then simplest form
