Omitted Variable Bias

[xtr0v3r7]

Hi folks,

I have just completed a program that illustrates the effects of Omitted Variable Bias. Within the program, I am altering the following things :
1)sample size
2)the true value of the omitted beta value
3)and the value of 'a'.

'a' is used in the following way - series X2=!c+!a*x1+@nrnd*(10^2/12)^.5.
This way I am able to compare sample correlation between x1 and x2 compared to the true value, which works out to be (a/(sqrt(a^2+1))).

My question is hopefully a simple one:
What factors should I be studying when evaluating the consequences of omitting a variable?

Thank you in advance for your attention and effort

xtr0v3r7