What would have been the results if the 2003 Ontario Election had been held under the proposed Mixed-Member Proportional system? Back in 2003, the Liberal candidates won in 72 ridings (out of 103), PC candidates - in 24 ridings and NDP - in 7. Under MMP there would be only 90 constituencies. Of course, it would be quite difficult to calculate the exact vote transposition, especially if the new constituencies haven't yet been drawn. So let's just assume that each party wins a similar share of the constituencies. Thus, the Liberals would have won 63 direct seats, the PCs - 21 and the NDP - 6.
What about the party list vote? Many just assume that it would mirror the local vote, to keep the things simple. With the small party vote discarded, the Liberals end up with 48.5% of the "qualified" vote, PCs - with 36.2%, NDP - with 15.3%; which entitles the parties to 63, 46 and 20 seats respectively. Since the Liberals already won their 63 seats in local ridings, they get no list seats. The PCs that only won in 21 ridings, get 25 list seats to bring the total to 46. The NDP with their 6 local seats receive the remaining 14 list seats to bring the total to 20.
But what if the Green party passed the 3% threshold? (It was just 0.18% short.) The total number of "qualified votes" would go up, so the Liberal share (~47%, rather than 48.5%) would only entitle them to 61 seats, while they've already won 63. Of course, the extra seats can't be taken away. That leaves the opposition slightly underrepresented: The PCs and the Greens each would end up one seat short (PC: 44, Greens: 3 instead of 45 and 4), while the Liberals would be 2 seats overrepresented. However this slight over-representation can't even be compared to a 70% majority, which the Liberals won under FPTP with the same number of votes.
But what if the vote splitting had been much bigger? After all, those who voted strategically on a local ballot could still vote their conscience on a party list ballot. In the last election, there were many NDP and Green party supporters who voted Liberal, to block a PC candidate from winning the riding. With strategic voting unnecessary on a party ballot, the Green party could have won about twice the number of votes and the NDP share would have been higher too. Let's assume, it was 5.75% for the Green and 17% for the NDP, leaving the Liberals with 41.7%. (The numbers for the local ballot don't change, so the Liberals still have 63 local seats.)
At the other end of the spectrum, there was the FCP. It only ran 51 candidate and got about 0.8% of the vote. Let's assume that FCP would have got 1.5% if they had run full slate and their share too would have doubled (just like the Green party) if it hadn't been for strategic voting. Let's give them 3.05%, with the remaining 32.5% going to the PCs.
41.7% of the vote would entitle Liberals to only 54 seats. Since they've already won 63, the opposition would end up with 9 less seats; not to accommodate the smaller parties, but to compensate for the 9 extra seats the Liberals won in the local constituencies. (Fully proportional results would have been: Lib: 54, PC: 42, NDP: 22, Green: 7 or 8, FCP: 3 or 4.)
I guess, if that had been the actual outcome of the 2003 election, it would have encouraged smaller parties to put more emphasis on a local campaign, rather than let the people vote strategically on a local ballot and count on a party ballot alone to deliver 100% proportionality.