Statistics Flashcards
What are the three levels of measurement?
Nominal: data organised into categories. Order is meaningless.
Ordinal data: data can be ranked but the intervals between each is subjective. Ie degree grade
Interval or ratio data: measured along a numerical scale that has equal distances between adjacent values
What is variance?
- The spread of the data around the mean.
- Average squared difference from the mean.
What is standard deviation?
- Square root of the variance.
- In original units.
- Easier to interpret than variance.
What are confidence intervals and why can they be useful?
We can use confidence testing to inform whether to reject the null. If no overlap between ranges, it is likely we have statistically significant differences.
Tell me about type 1 error
Where you incorrectly reject the null (you think there is an effect when there is not)
More likely if p value to high ie 10%. But can still happen at 5%.
For example, a drug does not actually work, but due to random chance a study shows it leads to a significant reduction in depression scores at 5% level. Researcher rejects the null when in reality it was due to chance. Results just happened to fall in the 5% region.
Tell me about Type 2 error
Where you do not reject the null when there is a good probability the results are due to chace
How to calculate confidence interval for 95% of people
It is the mean - the standard dev number
And the mean + standard deviation number. This gives you the range.
So if at 95% level on standard deviation this ranges to -1.96 and + 1.96, and the mean was 7. It is 5.80 and 8.20.
What are some ways to improve NHST (null hypothesis significance testing)?
Bayesian analysis
* Contrasts the probability of the null hypothesis being true against the
probability of the alternative hypothesis being true.
Pre/registration
* Prevents some of the abuses of NHST.
Despite these issues, NHST and p values remain important in
psychological research.
What is greek symbol for the mean
mu or u
