Type I And Type II Errors Flashcards
What is a null hypotheses?
‘No difference between conditions’
What is an alternative/ experimental hypotheses?
There will be a difference between conditions
What hypotheses is rejected for a type 1 error?
The null hypotheses is rejected and alternative hypothesis is accepted when it should be the opposite.
Which hypothesis is actually true for a type 1 error?
The null hypothesis
What is a type 1 error?
A false positive which claims to have found a significant result when there isn’t one.
What hypothesis is accepted for a type 2 error?
The null hypothesis is accepted and the alternative hypothesis is rejected when it should be the other way round.
Which hypothesis should actually be accepted for a type 2 error?
The alternative hypothesis
What is a type 2 error?
A false negative which claims to have not found a significant result when there is one.
When does a type 1 error usually occur?
If the significance is too high/ lenient
When does a type 2 error usually occur?
If the significance is too strict/ low
What does 0.05 do to the two errors?
Balance them
How do you choose a stats test?
- Difference or correlation
- Experimental design
- Level of measurement
How do we know if the experiment is difference or correlation?
From the hypothesis
What experimental design is used for related design?
- Matched pairs
- Repeated measures
What experimental design is used for unrelated design?
- independent measures
What are the three levels of measurement which quantitative data can be split into?
- Nominal data
- Ordinal data
- Interval data
What is nominal data?
- Categorical
- Discrete = one item can appear in one category
What is ordinal data?
Data that is ordered in some way
What is interval data?
- Based on numerical scales e.g. units of equal, precisely defined size.
- Usually measurements you’d find in maths or other sciences.
What are parametric tests for?
Interval data only
What do parametric tests deal with?
Data with a normal distribution
What do parametric tests have?
Homogeneity of variance - set of scores in each condition should have similar dispersion or spread
Mann Whitney
- Difference
- Unrelated design
-Ordinal data
Wilcoxon
- Difference
- Related design
- Ordinal data
