In a study of the relationship between the permeability (y) of human skin and its electrical resistance (x), the data presented in the following table were obtained for 50 skin specimens, each 2.54 cm² in area. Here permeability is measured in 𝜇m/h and resistance is measured in kΩ. Using a linear model, find a 95% confidence interval for the mean permeability for skin specimens with resistance 25 kΩ. (From the article “Multi-Species Assessment of Electrical Resistance as a Skin Integrity Marker for In Vitro Percutaneous Absorption Studies,” D. J. Davies, R. J. Ward, and J. R. Heylings, Toxicology in Vitro, 2004:351–358; values obtained by digitizing a graph.)
Resistance | Permeability | Resistance | Permeability | Resistance | Permeability |
10.09 | 11.58 | 18.67 | 9.73 | 25.98 | 7.01 |
11.37 | 13.89 | 20.28 | 14.33 | 26.37 | 6.66 |
12.08 | 11.77 | 20.17 | 7.52 | 26.42 | 5.35 |
12.25 | 9.02 | 20.17 | 5.96 | 26.75 | 4.05 |
13.08 | 9.65 | 19.94 | 8.10 | 26.92 | 7.16 |
13.52 | 9.91 | 21.72 | 10.44 | 27.80 | 7.07 |
13.75 | 12.42 | 20.94 | 7.30 | 27.80 | 6.47 |
14.19 | 9.93 | 21.44 | 7.56 | 28.63 | 6.50 |
15.13 | 10.08 | 22.05 | 7.58 | 28.47 | 5.30 |
15.13 | 5.42 | 21.66 | 6.49 | 28.19 | 4.93 |
16.07 | 12.99 | 21.72 | 5.90 | 28.97 | 4.36 |
16.51 | 10.49 | 22.66 | 7.01 | 29.85 | 4.28 |
17.18 | 8.13 | 22.10 | 9.14 | 30.02 | 4.88 |
18.34 | 5.78 | 22.82 | 8.69 | 31.79 | 6.02 |
18.84 | 7.47 | 23.99 | 4.66 | 34.28 | 4.67 |
18.34 | 7.93 | 24.82 | 8.88 | 34.61 | 6.12 |
18.17 | 9.95 | 25.70 | 5.92 |
We calculate the following quantities:
\overline{x} = 21.7548 \overline{y} = 7.849 \sum\limits_{i=1}^{n} (x_{i}\ −\ \overline{x})^{2} = 1886.48 \sum\limits_{i=1}^{n} (y_{i}\ −\ \overline{y} )^{2} =325.993
\sum\limits_{i=1}^{n}(x_{i}\ −\ \overline{x})(y_{i}\ −\ \overline{ y}) = −566.121 \hat{\beta}_{0} = 14.3775 \hat{\beta}_{1} = −0.300094 s = 1.80337
The estimate of the mean permeability for skin specimens with a resistance of 25 kΩ is
ŷ = 14.3775 − 0.300094(25) = 6.875
The standard deviation of ŷ is estimated to be
s_{\hat{y}} = s\sqrt{\frac{1}{n} + \frac{(x\ −\ \overline{x})^{2}}{ ∑^{n}_{i=1} (x_{i}\ −\ \overline{x})^{2}}}
= 1.80337\sqrt{\frac{1}{50} + \frac{(25\ −\ 21.7548)^{2} }{1886.48}}
= 0.28844
There are n − 2 = 50 − 2 = 48 degrees of freedom. The t value is therefore t_{48,0.25} = 2.011. (This value is not found in Table A.3 but can be obtained on many calculators or with computer software. Alternatively, since there are more than 30 degrees of freedom, one could use z = 1.96.) The 95% confidence interval is
6.785 ± (2.011)(0.28844) = (6.295, 7.455)