Test the significant difference between correlation coefficient.
Bicycle helmet use. Table lists data from a cross-sectional survey of bicycle safety. The explanatory variable is a measure of neighborhood socioeconomic status (variable_RFM). The response variable is percent of bicycle riders wearing a helmet (P_Helm).
Percent of school children receiving free or reduced-fee lunches at school (variable P_RFM) and percent of bicycle riders wearing a helmet (variable P_HELM). Data for this study was recorded by field observers in October of 1994.
|
I (i)
|
School
|
P_RFM
|
P-HELM
|
|
|
|
|
|
|
1
|
Fair Oaks
|
50
|
22.1
|
|
2
|
Strandwood
|
11
|
35.9
|
|
3
|
Walnut Acres
|
2
|
57.9
|
|
4
|
Disc Bay
|
19
|
22.2
|
|
5
|
Belshaw
|
26
|
42.4
|
|
6
|
Kennedy
|
73
|
5.8
|
|
7
|
Cassell
|
81
|
3.6
|
|
8
|
Miner
|
51
|
21.4
|
|
9
|
Sedgewick
|
11
|
55.2
|
|
10
|
Sakamoto
|
2
|
33.3
|
|
11
|
Toyon
|
19
|
32.4
|
|
12
|
Lietz
|
25
|
38.4
|
|
13
|
Los Arboles
|
84
|
46.6
|
In practice, the next step in the analysis would be to identify the cause of the outlier. Suppose we determine that Los Arboles had a special program in place to encourage helmet use. In this sense, it is from a different population, so we decide to exclude it from further analyses. Remove this outlier and recalculate r. To what extent did removal of the outlier improve the fit of the Test Ho: p=0 (excluding outlying observation 13).