Correlation 2 Flashcards
What is statistical significance?
The sample shows that it is fairly unlikely that the data comes from a bigger sample with no correlation.
Statistical significance shows
That a fair reflection of reality has been drawn
What does a extremely low p value mean?
Just that it’s significant but you have to look at the r value to see the strength.
What are the strength numbers of correlations?
.10 is small
.30 is moderate
.50 large
What is coefficient of determination?
r-squared which can tell you how much variance x and y share in common.
What is confidence intervals?
A way to see an estimate of the likely effect in the population
What is bivariate normal distribution?
Each variable is normally distributed.
Each variable is normally distributed at all levels of the other variable.
What is skewness?
It has to do with the lack of symmetry in a distribution.
Draw a positive skewed distribution
Large at the beginning and low at the end
Draw a negatively skewed distribution
Low at the beginning and large at the end.
What is kurtosis?
This has to do with the peakedness of a distribution.
What is leptokurtic?
A form of kurtosis. Very tall and slim
What is mesokurtic?
A form of kurtosis. A normal curve distribution
What is platykurtic?
A form of kurtosis. A low and wide distribution.
What are the best ways to assess normality in correlation?
If you have around 200 participants it’s best to look at a histogram. You should only be concerned if the deviations from normality are severe.
What if assumptions are not satisfied?
- change a level to .01 instead of .05.
- use stats that are not for interval data such as Spearman’s tho or Kendall’s T.
- Transform data
What can a non-linear relationship do to the results?
Underestimate the correlation between x and y. Departures from linearity are serious, then Pearson’s r is not appropriate.
Why are restricted range misleading?
Because they have a tendency to either reduce or inflate the correlation.
What are outliers?
A data point standing off by itself and they can strongly influence the value of r and thereby leading to a wrong conclusion about the extent to which the two variables are related.
How do we check for outliers?
Scatterplots
When do we have a perfect positive correlation?
When x and y distributions have exactly the same shape
When do we have a perfect negative correlation?
When x and y have exactly the same shape or when x and y distributions are oppositely skewed.
