Suppose the statistics values are given as follows: $\sum_{i=1}^{n}x_i, \sum_{i=1}^{n}y_i, \sum_{i=1}^{n}x_iy_i, \sum_{i=1}^{n}x_i^2,\sum_{i=1}^{n}y_i^2$

Firstly, we can compute the regression coefficient $\beta_0, \beta_1$ such that: $Y =\beta_0 + \beta_1 X$

An ANOVA (F-test) can also be carried out to see whether X and Y are dependent. Suppose $F<F_{critical}$ , p-value $> 0.05$ such that there is no strong evidence that $\beta_1 \ne 0$. i.e. X and Y are independent.

However, what if we were told that $\sum_{i=1}^{n}(x_i-\bar x)^2, \sum_{i=1}^{n}(y_i-\bar y)^2, \sum_{i=1}^{n}(x_i-\bar x)(y_i-\bar y)$ were doubled, X and Y are dependent by ANOVA, and find the minimum sample size $n$.

My approach: I am new to ANOVA and statistics test, the only I can think of is to start it with F-test formula , and if X and Y are dependent, it should fulfill $F>F_{critical}$ but $F_{critical}$ is unknown due to unknown sample size $n$.

Besides, the new model has the same $\beta_1$ as before. from doubling up $\sum_{i=1}^{n}(y_i-\bar y)^2$. We can also find out the new $SS_{TOTAL}$

Apologies for the unorganised ideas above. I have no idea of the right direction to solve this problem.

I would like to know how to solve this please. Thank you very much.



Your Answer

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

Browse other questions tagged or ask your own question.