(a) yes, because the results come from an observational study
(b) no, because the results come from an observational study
(c) * yes, because the results come from a statistical experiment
(d) no, because the results come from a statistical experiment
It’s an experiment because a treatment was imposed on the subjects. Therefore it can produce evidence
of cause and eﬀect, and because statistically signiﬁcant results were obtained, it does.
4. Two variables xand yare believed to have a straight-line relationship. We would like to predict y
from x. Minitab tells us this about xand y:
Descriptive Statistics: x, y
Variable N N* Mean SE Mean StDev Minimum Q1 Median Q3 Maximum
x 7 0 7.00 1.63 4.32 1.00 3.00 7.00 11.00 13.00
y 7 0 13.43 1.67 4.43 8.00 10.00 13.00 17.00 21.00
Correlations: x, y
Pearson correlation of x and y = 0.941
P-Value = 0.002
Use this information for this question and the two following.
What is the intercept of the regression line for predicting yfrom x?
(d) * 6.7
Slope is (0.941)(4.43/4.32) = 0.96, intercept is 13.43 −0.96(7) = 6.7.
5. Using the information given in Question 4, what is the predicted value of ywhen x= 10? The slope
of the regression line is 1.0.
(c) * 16.7
6.7 + (1)(10) = 16.7. Or: x= 10 is higher than average for x, and the correlation is positive, so y
should be higher than average for ytoo. The only alternative that is is 16.7.
6. In Question 4, some information is given about two variables xand y. From the information given,
does it make sense to ﬁnd the regression line?
(a) No, because xand yhave outliers
(b) * Yes
(c) No, because the correlation is not a good measure of the relationship between xand y
(d) No, because the relationship is not a straight line
You can do a quick 1.5×IQR check to verify that there are no outliers. The correlation is a good
measure of the relationship if it is a straight line, and there is nothing in the output to suggest that a
curve would be better.