Causation vs correlation
The cause of some change
Independent Variable (X)
What you do when you find a p-value less than .05
Reject the null hypothesis
The variable that goes on the vertical axis of a scatterplot
Y variable (the DV)
The expected value of Y when all X's are 0
The intercept
Drawing conclusions about individuals based on collective data
Ecological fallacy
The effect or result
Dependent Variable (Y)
How you know that there is no statistically significant effect of X on Y
The p-value is larger then .05 (so you fail to reject the null hypothesis)
The difference between the observed outcome and the predicted outcome (y - y(hat))
Residual
What you do when there are other variables that influence your DV
Include control variables
Something that leads result to deviate systematically from the truth- like only collecting data that supports your hypothesis
Selection bias
A change in the IV makes a change in the DV more likely
Probabilistic causation
A way to show your data so that you can compare variables when you have both binary/ordinal/categorical IV and DV
Crosstab
The line of best fit on a scatterplot does this
Minimizes the RSS
Including a binary variable in your model does this
When the operationalization of your variable does not measure what you say you are measuring
Measurement error
A graph you could make with a continuous IV and a binary DV
Mosaic plot
A statistical test that compares the mean of one category of a binary variable to the other to determine how likely the difference between the sample means is due to chance assuming there is no relationship
2 sample t-test
The % of the variation that your model explains
the R-squared value
What you should do if the effect of one IV is dependent on another IV
Include an interaction term
The DV causes the IV
Endogeneity (reverse causality)
Statistical support for the idea that the IV and DV vary together
Evidence of correlation
A statistical test that determines if categorical variables are independent of each other
Chi-squared test
A one-unit change leads to a beta change in y
Interpretation of regression estimates
We observe a relationship between X and Y, but there really is no relationship between these variables and something else is causing both of them (the result of failing to include a confounding variable)
Spurious correlation
Observations for one unit influence the observations of another unit (ie unemployment rate in 2016 influences unemployment rate in 2017)
autocorrelation