This is a problem on interpretation of regression equations which
have quadratic terms in explanatory variables.
Context of data set:
A paper helicopter experiment
(http://www.paperhelicopterexperiment.com/)
was run to find some optimal dimensions.
Explanatory variables are body length and body width (both in cm) of a piece of
paper before the folding/cutting is done to produce the helicopter.
The response variable is the flight time (in seconds) to land on floor after
release from a height of 2.5 m.
You can compare multiple regressions from fitting (i) a plane,
(ii) a quadratic with the original variables and
(iii) a quadratic with centred variables (len.centered=len-6, wid.centered=wid-1.5).
For (ii) and (iii), compare the numerical results with what you would expect
based on transformed equations.
For your data set, the response variable is:
flight time, in seconds, of the paper helicopter.
flighttime=c(1.3, 1.43, 0.93, 2, 1.21, 1.73, 1.1, 1.41, 1.89, 1.98, 1.98, 1.25, 1, 1.85, 2.08, 1.94, 1.83, 1.69, 1.87, 1.82, 1.61, 1.69, 1.86, 0.93, 1.87)
The explanatory variables are:
(i) length of paper before folding
len=c(6.1, 6.1, 6.6, 6.6, 5.6, 6.1, 5.6, 6.1, 6.1, 6.6, 5.6, 5.6, 6.6, 6.6, 6.6, 6.6, 5.6, 5.6, 5.6, 6.1, 5.6, 5.6, 6.6, 6.6, 6.1)
(ii) width of paper before folding
wid=c(0.7, 0.7, 0.7, 2.1, 0.7, 1.4, 0.7, 0.7, 1.4, 1.4, 1.4, 0.7, 0.7, 1.4, 2.1, 1.4, 1.4, 2.1, 1.4, 2.1, 2.1, 2.1, 2.1, 0.7, 1.4)
For the th case,
let be the value of length and
be value of centered length;
let be the value of width and
be value of centered width; and
let be the value of flight time.
Consider two regression models that are quadratic in the original variables
and quadratic in the centered variables.
and
To answer the parts below, several separate regressions could
be done or you can answer based on theory.
Part a)
Compare the least squares coefficients for the two quadratic models.
Maybe some coefficients are invariant to the centering.
Which of the following are correct? Possibly more than one item is correct.
Part b)
What is the relationship of the least squares coefficient
with centered length and width variables,
and the least squares coefficient for the original length and width variables.
Which of the following are correct? There might be more than one correct answer.
Part c)
Which of following would suggest that quadratic model is a better fit
to the data?
There might be more than one correct answer.
Part d)
If is not significant, then choose appropriate answers below.
There might be more than one correct answer.
Hint:
You can earn partial credit on this problem.