Can anybody give me a hint how to solve the following problem?
*One way to construct preferences over lotteries with monetary prizes is by evaluating each lottery L on the basis of two numbers, Ex(L), the expectation of L and var(L), L’s variance. Such a procedure may or may not be consistent with vNM assumptions.
Show that u(L) = Ex(L) − (1/4)var(L) induces a preference relation that is not consistent with the vNM assumptions. (For example, consider the mixtures of each of the lotteries  and 0.5 ⊕ 0.5 with the lottery 0.5 ⊕ 0.5.)*
I am especially not sure about how to compute the variance in this case.