S2 lecture 4 - critical analysis Flashcards
What are experimenter effects?
When people’s experiences influence the study outcome. Can be true for research and participant, not generally conscious.
How do we deal with experimenter effects?
Use of RA’s, blinded study, and double-blinded study.
What is a confound?
A variable that varies systematically with the independent variable in your study.
What are confidence intervals?
A range, based on data, that gives an indication of what the population mean will be. We can say that 96% of the time, the true population mean will be in the confidence interval.
How do you find confidence intervals?
To find the lower bound of the CI use M - 1.96 x S.E. To find the upper bound of the CI use M + 1.96 x S.E.
What is standard error (S.E.)?
The standard deviation, divided by the square root of N.
How does standard error (S.E.) differ from standard deviation (SD)?
SD is a measure of variability in your sample, it describes your data. S.E. is part of inferential statistics where we estimate the true mean of the population.
Where does 95% confidence interval come from?
The area between -1.96 and 1.96 covers 95% of all values.
When testing means between groups, how far away do 2 means need to be different?
If based on our data the second mean is less than 5% likely to occur the mean is different.
True or False: p is the probability that the null hypothesis is true?
False, p is the probability that results like the one seen in the sample would occur under the null hypothesis.
What is problem with p-values?
As a p-value gets smaller, this means that finding the sample size, assuming the null hypothesis is true, becomes less likely. A p-value says nothing about the strength of an effect.
What are effect sizes?
How much results are changed by condition. Example: there is found to be a significant difference between exam grades of students who did not consume alcohol during the semester and the control group. To decide if you should give up alcohol you may not care that the difference is significant, you’ll want to how much of a difference it will make, this is effect size.
Explain effect size for assocations: correlations.
Using Pearson’s r, if x and y are measured variables, the correlation coefficient lets you test whether the two are associated. Correlations are on a standardised scale of -1 to 1, and give an indication of the strength of association.
Explain effect sizes for mean differences: Cohen’s d.
The mean difference as expressed in standard deviations. If cohen’s D = 1, the mean difference is equal to one standard deviation.
