Back to Basics
What is correlation?
What you use when you want to run a correlation but you have variable scaling issues, outlier sensitivity, or some violations of linearity.
What is a rank-based correlation?
When we run into the third variable problem, we can still examine a relationship using this.
What is a partial correlation?
The statistical procedure we use when we don't have an appropriate test to run.
What is bootstrapping?
When values on one variable are unevenly associated with values on another variable.
What is a correlation?
The most common rank-based correlation is Spearman's Rank Correlation Coefficient. But there's another less widely recognized option.
What is Kendall's Tau?
When we want to remove the influence of a third variable, from only one of the main two variables of interest, we use this.
Semi-partial correlation
If we were able to measure thousands of populations, we wouldn't need this branch of statistics.
What is inferential statistics?
What is a nominal variable?
There are two assumptions for the rank-based correlation hypothesis tests. One is that the data represent random, independent samples from a population (implied to be identically distributed). The other is more a condition of use. "The variables X and Y have at least ___________."
What is ordinal scaling?
The term used for "leftover" variation after you run a regression. The unpredictable portion.
What are the residuals?
It's not only hypothesis tests that are affected by non-normally distributed errors, but also ________.
What is a confidence interval?
The idea that the full strength of a relationship can be captured in statistics that assume linearity, like the Pearson r.
What is a misconception?
Something that makes it "feel" less impressive when looking at a result from a rank-based correlation.
What is a lower value?
After you remove the influence of Variable C on Variable A and Variable B, you could say the new relationship you are observing between Variable A and Variable B is __________ of Variable C.
What is independent?
The questionable value that comes from running a hypothesis test on badly skewed data.
What is a p-value?
Variation that bridges or links two variables, so that variations in one variable are connected to variations in the other.
What is covariance?
The four applications of rank-based correlations include...
1. Violations of normality
2. Outliers
3. Monotonic nonlinearity
4. ???
What is ordinal variable scaling?
Another name for semi-partial correlations, but that can cause a bit of a headache.
What is sampling with replacement?