Advertisements
Advertisements
Question
For the certain bivariate data on 5 pairs of observations given
∑x = 20, ∑y = 20, ∑x2 = 90, ∑y2 = 90, ∑xy = 76
Calculate:
- cov(x, y)
- byx and bxy
- r
Solution
Given, ∑x = 20, ∑y = 20, ∑x2 = 90, ∑y2 = 90, ∑xy = 76, n = 5
Now,
`bar x = (sumx)/n = 20/5` = 4
`bar y = (sumy)/n = 20/5` = 4
(i) cov(x, y) = `1/n sumxy - barx bary`
= `1/5 xx 76 - 4 xx 4`
= 15.2 – 16
= - 0.8
(ii) byx = `(sumxy - nbarxbary)/(sumx^2 - nbarx^2)`
= `(76 - 5 xx 4 xx 4)/(90 - 5(4)^2)`
= `(76 - 80)/(90 - 80)`
= `-(4)/10`
= - 0.4
bxy = `(sumxy - nbarxbary)/(sumy^2 - nbary^2)`
= `(76 - 5 xx 4 xx 4)/(90 - 5(4)^2)`
= `(76 - 80)/(90 - 80)`
= - 0.4
(iii) r = `+-sqrt(b_xy. b_yx)`
= `+- sqrt((-0.4)(-0.4))`
= `+- sqrt(0.16)`
= ± 0.4
Since byx and bxy both are negative,
∴ r is negative.
∴ r = - 0.4
RELATED QUESTIONS
Choose the correct alternative:
The regression line is obtained by
Following table shows the all India infant mortality rates (per '000) for years 1980 to 2010:
Year | 1980 | 1985 | 1990 | 1995 | 2000 | 2005 | 2010 |
IMR | 10 | 7 | 5 | 4 | 3 | 1 | 0 |
Fit the trend line to the above data by the method of least squares.