Question: Networked computers tend to slow down when they are overloaded. The response time is how long it takes from when you press the Enter key until the computer comes back with your answer. Naturally, when the computer is busier (either with users or with other work), you would expect it to take longer. This response time (in seconds) was measured at various times together with the number of users on the system and the load (the percent of the time that the machine is busy with high-priority tasks). The data are shown in Table 12.5.4.
a. Explore the data by commenting on the relationships in the three scatterplots you can produce by considering variables two at a time. In particular, do these relationships seem reasonable?
b. Compute the correlation matrix and compare it to the relationships you saw in the scatterplots.
c. Find the regression equation to predict response time from users and load. (You will probably need to use a computer for this and subsequent parts of this problem.)
d. To within approximately how many seconds can response time be predicted by users and load for this data set?
e. Is the F test significant? What does this tell you?
f. Are the regression coefficients significant? Write a sentence for each variable, interpreting its adjusted effect on response time.
g. Note that the two regression coefficients are very different from each other. Compute the standardized regression coefficients to compare them and write a sentence about the relative importance of users and load in terms of effect on response time.
