How to calculate a combined effect size? I have got several significant variables in multiple regression analysis. Standardized coefficients of two of the significant variables are:
Education of Son: -.487
Education of Daughter: .428
I also got few more variables. Now I would like to group these variables such as: children education, economic condition etc. and compare the effect size of those in graph. 
Can I add the effect size of the variables within the same group. For example, can I say children education effect size is: (-.487+.428)= -.059?
 A: Easiest way would be to create a new variable that is the education of the children, ignoring their gender, and add that new variable to your model instead of son's and daughter's education. You will need to decide what to do when there are multiple children, e.g. constrain all their effects to be equal.
However, that would be a bad idea in your case. What strikes me is that the effects of sons and daughters are fairly similar except that they have the opposite sign. So ignoring the childrens gender would mean that the effects would more or less cancel each other out, and a very small effect would be left. This would hide the quite substantial effect you found.

Edit: 
Alternatively, if you want to combine the effects of multiple variables you could look into sheaf coefficients.
A: That's not quite right....  The effect of having a single kid would be a weighted average of the two (depending on the probabilities of each event occurring).
Let $y$ be some outcome variable. Let $x_1$ be an indicator variable for category 1 and $x_2$ be an indicator variable for category 2. So the linear regression gives you a conditional expectation function:
$$ E[y  \mid  x_1, x_2] = \beta_0 + \beta_1 x_1 + \beta_2 x_2 $$
Let $p_1 = P(x_1 = 1, x_2 = 0)$ and $p_2 = P(x_1 = 0, x_2 = 1)$. We can alway condition down:
$$ \begin{align*}
E[y  \mid  x_1 + x_2 = 1] &= E[E[y  \mid  x_1, x_2] \mid \ x_1 + x_2 = 1] \quad \quad \text{by law of iterated expectations}\\
&=\beta_0 + \frac{p_1}{p_1 + p_2} \beta_1 + \frac{p_2}{p_1 + p_2} \beta_2 
\end{align*}$$
