Standard error, effect size and confidence intervals Flashcards
What is the standard error of the mean?
standard deviation divided by the sampling distribution of the mean (mean for a sample)
What are degrees of freedom?
the number of scores in a set that are free to vary
What is standard error?
the standard deviation of a number of sample of means
e.g. how much we can expect the sample’s mean to differ from the population’s mean
Why are standard errors useful?
helps us understand how accurately the sample represents the whole population
shows the reliability of a finding
What’s the difference between standard error and standard deviation?
SD: measures variability within a sample or population
SE: measures variability of the mean across different samples
How do you interpret a small standard error?
sample averages are usually close to the true population
How do you interpret a large standard error?
the sample averages are more spread out and less consistent than the true population average
What is a confidence interval?
a range within which the population value will most likely lie
usually calculated at a 95% confidence interval
How do you interpret a large confidence interval?
Wider intervals means more uncertainty
The range is bigger, so we are less sure about the true value
How do you interpret a small confidence interval?
We’re more certain about the true value because the range is small
What is the significance fallacy?
a statistically significant result automatically implies a practical or meaningful importance
What is psychological significance?
How meaningful or impactful a result is in terms of human behaviour/understanding or theory
What is effect size?
Refers to the magnitude of the impact of some variable on another
They don’t tell us whether the results occurred due to chance
Independent of sample size
What are the common measures of effect sizes?
Cohen’s d: comparing the differences between 2 groups
Pearson’s R: asesses the strength of the linear relationships between 2 variables. from -1 to +1
R squared: represents the proportion of the vairance in 1 variable that is used to explain another variable, from 0 to 1
When is an effect size meaningful?
Even small effect sizes can be important
The manipulation of the IV has to be quite minimal and the DV is hard to influence
