you criticize the use of a Cooks-distance-type mea-
sure to identify influential subjects?
b. Suppose you identified five influential subjects in
the dataset. How should you decide whether or not
to drop any of these subjects from the data?
c. How can you criticize using the model stated in
question 4 to identify influential subjects?- In attempting to determine a best model, starting with
the model stated in question 4, suppose you wanted to
use a Bonferroni-type adjustment to correct for multi-
ple testing of interaction terms. What problem will you
have in trying to determine the significance level for
individual tests in order to achieve a family-wise error
rate (FWER) not higher than 0.05?
Test THE SCENARIO.A firm mattress is often believed to be
beneficial for low back pain, although evidence supporting
this recommendation is lacking. A randomized, double-
blind, controlled, multicenter trial was carried out to deter-
mine the effect of mattress firmness on the clinical course
of patients with chronic, nonspecific low back pain.
A series of 313 adults with chronic, nonspecific low back
pain, without referred pain, who complained of backache
while lying in bed and upon rising were randomly assigned
to four types of mattresses:
3 ¼firm, 2¼medium firm, 1¼medium soft, 0¼soft.
Clinical assessments were carried out at baseline and at
90 days. The endpoint (i.e., the health outcome variable)
was improvement in pain while in bed and upon rising
from bed between 0 and 90 days. The data set included
the following variables:
D¼improvement score for pain in bed from baseline to 90
days
(0¼no improvement or improvement in bed only,
1 ¼improvement upon rising only or improvement
both in bed and upon rising)
F¼firmness of mattress (3¼firm, 2¼medium firm,
1 ¼medium soft, 0¼soft)
BASE¼type of base of bed (1¼firm, 0¼not firm)
POST¼posture while sleeping at baseline
(1¼supine or fetal, 0¼other)
PF¼subjective perception of firmness of mattress (3¼
firm, 2¼medium firm, 1¼medium soft, 0¼soft)
OCC¼type of occupation (1¼sedentary,
0 ¼not sedentary)
AGE¼age in years of subject
GEN¼gender (0¼male, 1¼female)Test 293