Inferertial Statistics Flashcards
Why use inferential statistics
Draw conclusions from findings from sample of a population
-see if results are statistically significant or due to random chances
If significant- Alternative hypothesis accepted as something is happening (directional or non-directional)
If random chance - null hypothesis as nothing is happening or opposite thing is happening
How to use inferential statistics
Different formula gives an observed/calculated value - compare this with critical value from a table at 0.05 probability
-to make sure random chance is less or equal to 5% and 95% sure result not due to chance
Can’t just use descriptive statistics (mean, median) - they only show trend not draw conclusion
Spearman’s rank order correlation coefficient
Find a correlation between 2 co-variables (data that are related). Can only be used with ordinal + ratio/interval data.
Observed value between -1 and +1 (given in exam)
Find critical value =
-look if study is directional or non-directional
-how many participants (N)
Find critical value from N
Compare observe with critical on table (based on directional or non-directional)
-observed equal or higher than critical = null reject, alternative hypothesis accepted (ignore + and - symbol when comparing)
-if lower or shows a opposite correlation = null accept, alt rejected
Chi squared test
For nominal data only (lowest)
-calculate frequency or occurrence of data in value
-test for association of variables or difference
-only for repeated measures
Observe value = chi squared
Crit value =
-directional or non-directional
-degree of freedom (number of row 1-1) x (number of row 2-1) so (2-1)x(2-1) =1
Observe greater or equal to critical = reject null, accept alt
Vice versa if lower
Mann-Whitney U test
Test of difference
-used for independent measures only
-ordinal + interval/ratio data only
Observer value = U
Crit value =
-directional or non-directional
-how many participants in each group N1 + N2
2 data so look down and across for critical value
-observed less or equal to critical value= null reject, alt accept
-vice versa if higher
Sign test
Test of difference
-used for repeated measures compare results from same participant
-nominal data only (category)
Observer value = S
-the lowest frequency/number from a set of data is picked as S, so if data is 7 smart, 2 dumb, 5 normal, 2 is picked
Crit = participant N, one or two tailed hypothesis
Observe lower or equal to critical value= null reject alt accept
-vice versa if higher
Wilcoxon T test
Test of difference
-for ordinal + ratio/interval data
-used for matched pair design
Observe value = T
Critical =
-one or two tailed
-participant number N
Observe less or equal than crit = null reject alt accept
-vice versa if higher
R up rule
If there are R in test (SpeaRman, Chi squRe)
-observe value must be gReateR or equal to crit
If not R (Man Whitney, Wilcoxon, sign)
-observe must be lower or equal to crit
Summary picture
https://resource.download.wjec.co.uk/vtc/2020-21/el20-21_6-23a-kos/pdf/_eng/4-14-1-inferential-statistics.pdf
Sign, chi, sign, spearman critical value table might look like
https://homework.study.com/cimages/multimages/16/capture1419019533866701582.jpg
Chi square critical value table
https://www.scribbr.com/wp-content/uploads/2022/05/chi-square-distribution-table-critical-value.png
Mann Whitney U critical value table
https://encrypted-tbn0.gstatic.com/images?q=tbn:ANd9GcQDMt4t5zQif1GbzJt96ytC9KM-yg6f3aI6ng&s
