Regression worksheet Last updated Sept. 15, 2007. Send comments/corrections to mparker@austincc.

edu page 1 of 2

Worksheet for Non-Computer Work on Correlation and Regression (September 15, 2007)

Data: Each student has a Quiz 1 score (x variable) and Quiz 5 score (y variable) in a class. Each quiz has a
maximum of eight points. We'd like to predict the Quiz 5 score from the Quiz 1 score.

x y
0 3
1 3
1 4
2 5
5 6

1. In the blank space above, plot these data on a scatterplot with y on the vertical axis.
2. Guess what r is. (Is it positive or negative? Is it close to zero or close to positive or negative one?)

3. For each variable (x and y) find the mean and standard deviation. The standard deviations are provided here
in case you don’t have the technology you usually use to find standard deviations easily available right now.

x y

sx  1.924 sy 
1.304

4. On homework and quizzes, use MINITAB to compute the correlation coefficient. On tests, you won't
have to compute the correlation coefficient. For these data r = 0.917.
Is this consistent with your guess?

5. On your scatterplot, guess where you would draw a line that comes close to describing the data. Do a very
light sketch of it. You will compare your guessed line with the one you actually compute below.

6. On homework, quizzes and tests, be able to compute the equation of the regression line by hand, using
a scientific calculator, using these formulas. On homework and quizzes, use this often enough to learn to do
it for the test. You may use the computer to find the equation of the regression line on most homework and quiz
problems.

Compute the slope and intercept of the regression line.

sy
br 
sx

a  y  bx 

7. Write the equation of the regression line: ( . You write this, but put in your computed values for
a and b.)
Regression worksheet Last updated Sept. 15, 2007. Send comments/corrections to mparker@austincc.edu page 2 of 2

8. Use the equation of the regression line to predict y twice: when and when .

x3 ŷ 

x5 ŷ 

9. Using a different colored pencil, put these two points you just computed on your graph and then draw a line
through them. This is the graph of the regression line that you just computed. Does it look pretty close to
the line you guessed?

10. Compute the residual for . (The residual is the difference between the observed y value and the
predicted y value.)

11. Use the line you drew to predict y when . Write your prediction here.

12. After reading the example in the text about interpreting the regression coefficients, write your
interpretations about quiz grades here:

Intercept: (If , then the predicted value of y is a.)

Slope: (If x increases by 1, then we predict that y will increase by b.)

--------------------------------------------------------------------------------------------------
NOT REQUIRED: Computing the correlation coefficient by hand.
Look at the formula to see what you must do to find the correlation coefficient. Notice the formulas in the
top row of this table. Check to see how the first row of values was found. Then fill in the next blank row.
The rest of the rows are already filled in to save you time. Then find the sum of the last column.

xx y y  x  x  y  y
x y   
xx sx y y sy  sx   s y 

0 -1.8 -0.93555 3 -1.2 -0.92025 0.86094

1 3

1 -0.8 -0.41580 4 -0.2 -0.15337 0.06377

2 0.2 0.104395 5 0.8 0.61350 0.06405
5 3.2 1.66320 6 1.8 1.38037 2.29583
 x  x  y  y 
     
 sx   s y  3.66695

Compute Answer: r = 0.917

Do you see why we don’t require you to compute the correlation coefficient by hand? How much work would it
take if the dataset had twenty observations?!!