✓ Inferential Statistics - significance testing (AO1) Flashcards
What are the 4 steps to significance testing?
Probability
Significance
Calculated Value
Critical Value
What is probability?
“the likelihood a certain outcome will happen”
How is probability used in psychology?
- look at the likelihood our IV caused the DV (or that there is a link between 2 variables)
- at the start of the experiment you write experimental AND null hypothesis
- if there is a high probability of our results being due to chance we accept the null hypothesis & reject the experimental hypothesis
- if there is a low probability “ vice versa
What is signficance?
in psych we represent probability with a “significance level”
How is significance used in psychology?
- the normal level used is 5% (probability of the result being due to chance/other extraneous variables is less than 5%)
- written as p <_ 0.05
- this means the researcher is 95% sure the results are due to the IV causing the DV/or the 2 variables are linked
- sometimes we use the p<_0.01 (1% chance & 99% certain)
- OR p <_0.10 (10% & 90%)
What is calculated value?
after completing the statistical test you will get a “calculated value”
(e.g. for the sign test it could be s=3)
- this number needs to be compared to the critical value
What is critical value?
the number in the statistical table that needs to be compared to the calculated value from the statistical test
- to find the correct critical value you need the 3 pieces of information
What are the 3 pieces of information you need to find the correct critical value?
number of ppts:
- this is referred to as “N”
(for a chi-squared test “Df” is given instead)
hypothesis table:
- directional = one-tailed
- non-directional = two-tailed
level of significance:
- usually p=0.05
- sometimes p=0.10/0.01
What are potential errors?
we can never be 100% sure of an outcome because there is always potential for error to occur
What is a type 1 error?
false positive:
- when we mistakenly accept the experimental hypothesis (thinking our results are signif.) & reject the null hypothesis
- most likely to happen with p<_0.10 which leaves 10% room for chance/error (overly lenient)
What is a type 2 error?
false negative:
- when we mistakenly accept the null hypothesis (thinking our results are non-signif.) & reject the experimental hypothesis
- most likely to happen with p<_0.01 which only leaves room for 1% chance/error (overly harsh)