Step 1: A subset W of a vector space V is a subspace exactly when it contains the zero vector and is closed under addition and scalar multiplication. Since each condition here involves evaluating p or p' at a point, check whether the zero polynomial belongs to each set, and whether the defining condition is linear and homogeneous.
Step 2: Check \(W_1 = \{p : p(1/2) \notin \mathbb{Q}\}\). The zero polynomial has \(0(1/2) = 0 \in \mathbb{Q}\), so \(0 \notin W_1\). A subspace must contain the zero vector, so \(W_1\) is not a subspace.
Step 3: Check \(W_2 = \{p : p(1/2) = 1\}\). Again \(0(1/2) = 0 \neq 1\), so the zero polynomial is not in \(W_2\), and \(W_2\) is not a subspace. Also, if \(p,q \in W_2\) then \((p+q)(1/2) = 2 \neq 1\), so closure under addition fails too.
Step 4: Check \(W_3 = \{p : p(1/2) = p(1)\}\). Define \(L(p) = p(1/2) - p(1)\), a linear functional on V since evaluation at a point is linear in p. Then \(W_3 = \ker L\). Since \(L(0)=0\), \(0 \in W_3\); if \(p,q \in W_3\) then \(L(p+q)=L(p)+L(q)=0\), so \(p+q \in W_3\); and \(L(cp)=cL(p)=0\), so \(cp \in W_3\). All subspace axioms hold, so \(W_3\) is a subspace.
Step 5: Check \(W_4 = \{p : p'(1/2) = 1\}\). The zero polynomial has \(0'(1/2) = 0 \neq 1\), so \(0 \notin W_4\), and it is not a subspace.
Step 6: Only \(W_3\) satisfies all subspace conditions. \[\boxed{W_3 = \{p \in V : p(1/2) = p(1)\}}\]