In: Math
Compounds in the saliva, hair and urine of pets may potentially trigger asthma in some patients. To test this hypothesis, some researchers conducted a study. As test subjects, they used 8 families, 4 of which had a pet in their home and 4 did not. They measured respiratory flow of each family member, as this is known to decrease in patients with asthma. Recent research has shown body weight to correlate with asthma in children, and sex, age, and height are also known to affect respiratory flow measures. Therefore, the researchers measured and included these covariates in their model to control for any confounding effects that they may have had before testing their treatment of interest (whether the home had pets). The analysis conducted by the researchers is presented below. They concluded that pets in the family do not influence respiratory flow.
> attach(pets.data)
> pets.data
resp.flow body.weight treatment sex age height family
1 266 41 pet F 13 130 Smith
2 400 95 pet M 37 190 Smith
3 369 81 pet F 36 180 Smith
4 365 78 pet M 42 185 Wilson
5 351 75 pet M 35 192 Wilson
6 334 62 pet F 41 170 Taylor
...
13 391 78 no pet M 41 180 Singh
14 340 62 no pet F 27 177 Singh
15 337 55 no pet F 34 185 Singh
16 389 87 no pet M 27 172 Campbell
17 376 78 no pet F 33 167 Campbell
18 338 57 no pet F 47 155 Li
19 337 62 no pet M 50 172 Li
20 359 69 no pet M 18 173 Li
> model<-lm(resp.flow~treatment+sex+age+height+body.weight)
> anova(model)
Analysis of Variance Table
Response: resp.flow
Df Sum Sq Mean Sq F value Pr(>F)
Treatment 1 1.8 1.8 0.0569 0.8149
sex 1 6626.9 6626.9 209.6419 8.117e-10 ***
age 1 10264.8 10264.8 324.7247 4.396e-11 ***
height 1 8346.9 8346.9 264.0536 1.758e-10 ***
body.weight 1 5786.0 5786.0 183.0373 1.978e-09 ***
Residuals 14 442.6 31.6
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
a) Identify three aspects of the design and/or analysis of this study that reduce your confidence in the conclusion that pets dont influence respiratory flow.
b) List four assumptions of the model used in this analysis.
c) If the assumptions aren’t met, what are three approaches you can take to analyse the data?
d) Which variable in their analysis explained the most variance in respiratory flow?
e) In the ANOVA table below, which letters (A-E) represent numbers that would NOT change if the sample size of people increased? (1 mark)
Response: resp.flow
Df Sum Sq Mean Sq F value Pr(>F)
Treatment A B C D E
sex 1 6626.9 6626.9 209.6419 8.117e-10 ***
age 1 10264.8 10264.8 324.7247 4.396e-11 ***
height 1 8346.9 8346.9 264.0536 1.758e-10 ***
body.weight 1 5786.0 5786.0 183.0373 1.978e-09 ***
Residuals 14 442.6 31.6