Inferential Statistics Flashcards
What do statistical tests show us
The stats tests calculate whether a result is significant using different levels of significance. For instance it would suggest whether the probability of the result is due to chance or whether we can have a confidence in the significance of the results obtained
Different levels of significance
O.1% - highly significant - used in important research when the risk of chance must be minimised
1% - very significant - may be used when challenging previous research findings
5% - significant - conventional and accepted minimal level of significance used in psychology
10% - not significant but may be an effect - may be worth doing follow up studies
The 5% level of significance
Means there is a 95% confidence that the result is significant and there is only a 5% probability that the results are due to chance
It strikes a balance between making a type 1 and type 2 error so is the best to use
Type I error
False positive
You have rejected the null hypothesis when you should have accepted it. More likely to make the error with a 10% level of significance
Type II error
False negative
You have accepted the null hypothesis when you should have rejected it
More likely to be made with a 1% level of significance
Nominal Data
Worst kind of data
Categories/ labels
No individual scores they are gathered into a category
Eg yes or no, win or lose
Ordinal data
Order and ranked position
It’s on a subjective scale and the gaps between scores are not equal
Eg 1st 2nd 3rd, sad neither happy
Interval data
Best data
Fixed gaps between data, fixed intervals
Real measurements that can be plotted on a ruler
Gaps between scores are the same
Objective scores
Time (secs), height (cm)
Concerting types of data
You can only convert good data into worse data. You cannot make bad data better
Eg interval to ordinal to nominal works but nominal to ordinal to interval doesn’t
Choosing a stats test table
Design. Nominal. Ordinal. Interval
Difference. Related. Sign test. Wilcoxon Related T
Difference. Unrelated. Chi2. Mann W. Unrelated T
Relationship. Related Chi2. Spearman. Pearson
Calculating the sign test
Work out the difference between each participant scores and note if it is a negative or positive difference
Exclude any ppts with a difference of 0 from the results
Count the number of positive and negative signs
The one with the smallest number is known as the observed/calculated value (S)
How to work out critical values
Figure out the type of hypothesis
Work out number of ppts or degrees of freedom (N value)
What is the level of significance
Find N value and test and level of significance and match up to find CV
Compare observed value to critical value
Write a conclusion
Conclusion you would write accepting the null (example)
There is no significant difference because the calculated value is greater than the critical value. Therefore the null hypothesis can be accepted
Conclusion you would write rejecting the null hypothesis (example)
There is a significant difference. The calculated value is greater than the calculated value so the null hypothesis can be rejected