o. Consider two candidate models below where Y is a response variable, and r is

Question

o. Consider two candidate models below where Y is a response variable, and r is a known explana- tory variable. Model Mo Y po Model M Y Bo Bar where is a random variable with mean zero and variance a Suppose that n independent observations are available: (ri, yi) i n) from model Mi, where the values of the explanatory variable (ri i 1, were controlled by the experimenter so that Fri 0. Note that model Mi is the true model that generated the data. Suppose that our goal is to estimate E(YIr aro), where ro is a single specified value of the explanatory variable

Explanation / Answer

the answer is YES

it is possible when the difference in scale deviance of the two models is greater than twicw the difference in number of parameters of the models i.e.

let So be scale deviance of model Mo with q parameters

and let S1 be scale deviance of model M1 with p parameters

if So-S1 > 2(p-q), then Model Mo is better than Model M1

Related Questions

Navigate

• Browse (All)
• Browse O
• Subjects

---

---

Integrity-first tutoring: explanations and feedback only — we do not complete graded work. Learn more.