Correlation
Correlation ranges between these two numbers
-1 to +1
In lecture, there was an image showing a high correlation between chocolate consumption in a country and number of Nobel winners from a country. What was the confounding variable?
High GDP, high quality of life
What is the range of scale for R-Squared? What R-squared value would suggest a perfectly predicting model?
0 to 1. 1 suggests perfect.
what does the "hat" over the response variable mean?
Population correlation is a _____________ .
parameter
What happens to the sample correlation as the number of observations increases, if no bias in the sample?
it gets closer to the value of the population correlation, ie. becomes more stable
There are always an equal number of positive and negative error terms in a regression. T or F?
False
True or False? A lower R-squared value (vs. higher) means greater predictive power
False. A higher R-squared value means greater predictive power.
What is the hypothesized value of correlation in testing for significance?
0 (zero)
What calculation would you perform on "r" to arrive at R-squared?
You would square "r" to get R-squared
In the critical value approach method for hypothesis testing, will a test statistic with a higher or a lower magnitude give you a better chance of rejecting the Null hypothesis?
higher. The higher the magnitude (absolute value) of the test statistic the further out in the tail it goes, increasing your chance of reject the Ho and accepting the Ha.
If we summed up all the error terms against a true regression line, they would approximate this number.
0 (zero)
how do you pronounce the population parameter for correlation?
pronounce "roh" or "row"
If there is a 0.5% chance we incorrectly rejected the null hypothesis, what is the p-value?
0.005 (0.5%)
In hypothesis test for a simple regression, the formula for degrees of freedom is "n-2". Why is 2 subtracted?
2 variables (response and explanatory)
interpret the slope coefficient with the below information:
b1 = 2.5
x is debt (in $000's)
y in income (in $000's)
What is the name of the variable that links two highly correlated variables?
Confounding variable
An Eller undergrad is interested in testing whether an association exists between these two variables and surveys 32 randomly selected people in downtown Phoenix, asking them their student loan debt and salary. She finds the sample correlation to be 0.54. Assuming a significance level of 5%, what is the t-critical value? df = n-2
2.042
What is the difference between correlation and regression
You use regression to predict. Correlation only gives us strength of relationship between two variables.
interpret R-squared given the below information.
R2=0.18
x is number of credit hours taken
y is hours per week watching Netflix