We are given \( f(x) = 2x \), \( 0<x<1 \), and \( n = 3 \). Let the order statistics be \( X_{(1)} \leq X_{(2)} \leq X_{(3)} \).
We need to find:
\[
p = P(X_{(3)} \geq 2X_{(1)}).
\]
The joint pdf of \( X_{(1)} \) and \( X_{(3)} \) for a sample of size \( n = 3 \) is:
\[
f_{X_{(1)}, X_{(3)}}(x_1, x_3) = 6 [F(x_3) - F(x_1)]^{1} f(x_1) f(x_3), \quad 0<x_1<x_3<1.
\]
Since \( F(x) = x^2 \) and \( f(x) = 2x \), we get:
\[
f_{X_{(1)}, X_{(3)}}(x_1, x_3) = 6 [x_3^2 - x_1^2] (2x_1)(2x_3) = 24x_1x_3(x_3^2 - x_1^2).
\]
Now, the required probability is:
\[
p = \int_0^{0.5} \int_{2x_1}^{1} 24x_1x_3(x_3^2 - x_1^2) \, dx_3 \, dx_1.
\]
Evaluating this integral gives approximately:
\[
p = 0.4377.
\]
Thus, the value of \( p \) is \( \boxed{0.438} \).