regression analysis

Need some help in forming the ANOVA table without observed values. I only managed to find SST with the first line of hint, but do not know how to proceed to find SSTreatment or SSError. Any hint/solution will be greatly appreciated.


First set up the error funcion

$$F(\beta_0, \beta_1) =\sum_{i=1}^{N}[y_i-\beta_0-\beta_1x_i]^2.$$

Differentiate with respect to $\beta_0$ and $\beta_1$ and set the derivatives equal to $0$ to obtain.

$$\sum_{i=1}^Ny_i-N\beta_0-\left[\sum_{i=1}^{N}x_i\right]\beta_1=0$$ $$\sum_{i=1}^Ny_ix_i-\left[\sum_{i=1}^{N}x_i\right]\beta_0-\left[\sum_{i=1}^{N}x^2_i\right]\beta_1=0.$$

The first equation can be rewritten as

$$N\bar{y}-N\beta_0-N\bar{x}\beta_1=0 \implies \beta_0 = \bar{y}-2\bar{x}=0.$$

From the second equaiton we can obtain

$$150-15\cdot 2\cdot 0 - \left[\sum_{i=1}^{15}x^2_i\right]\beta_1=0$$ $$\implies \sum_{i=1}^{15}x^2_i=37.5$$

We know that the correlation $r$ is given by

$$r= \dfrac{1/N\sum_{i=1}^{15}x_iy_i-\bar{x}\bar{y}}{s_x s_y}$$

We have to determine $s_x$ in order to calculate $r$: $$s_x = \dfrac{1}{15-1}\sum_{i=1}^{15}(x_i-\bar{x})^2$$ $$= \dfrac{1}{15-1}\left[\sum_{i=1}^{15}x^2_i-2\bar{x}\sum_{i=1}^{15}x_i+\sum_{i=1}^{15}\bar{x}^2\right]$$ $$= \dfrac{1}{15-1}\left[\sum_{i=1}^{15}x^2_i-2\cdot 15\cdot\bar{x}^2+15\bar{x}^2\right]$$ $$= \dfrac{1}{15-1}\left[\sum_{i=1}^{15}x^2_i-15\cdot\bar{x}^2\right].$$

Finally, we know that $\text{SSTotal}=\sum_{i=1}^{15}\left[y_i-\bar{y}\right]^2$ and

$$r^2 = \dfrac{\text{SSTreatment}}{\text{SSTotal}}=\dfrac{\text{SSTreatment}}{\sum_{i=1}^{15}\left[y_i-\bar{y}\right]^2}=\dfrac{\text{SSTreatment}}{(15-1)s^2_y}$$ $$\implies \text{SSTreatment} =(15-1)\,r^2\,s^2_y$$


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.