(See also a related article.)This web page considers the extent to which vote-splitting among Ontario's three main center-left parties (NDP, Liberal, Green) contributed to seats won by the Conservatives. For this analysis, I used data from the Elections Ontario unofficial results to generate a table of results (scraped on June 8, 2018, by writing some computer scripts).
Specifically, I consider the question of how many additional seats would have been won if various of the center-left parties had combined their votes together. (Of course, this is purely hypothetical, since in reality a combined party might not keep all of the supporters of the original parties.)
The actual election seat counts were Conservatives 76, NDP 40, Liberals 7, and Greens 1, giving the Conservatives a majority.
If the NDP and Greens had combined their votes, they would have won a total of 50 seats (instead of 41). Specifically, they would have picked up 8 additional seats from the Conservatives (brampton-west, brantford-brant, cambridge, kitchener-conestoga, kitchener-south-hespeler, ottawa-west-nepean, sault-ste--marie, scarborough-rouge-park) plus one from the Liberals (thunder-bay-superior-north).
If the NDP and Liberals had combined their votes, they would have won a total of 82 seats (instead of 47), a clear majority. Specifically, they would have picked up 35 additional seats, all from the Conservatives (see list).
If the Liberals and Greens had combined their votes, they would have won a total of 12 seats (instead of 8). Specifically, they would have picked up two additional seats from the Conservatives (eglinton-lawrence and ottawa-west-nepean), plus two from the NDP (thunder-bay-atikokan, toronto-st--paul-s).
If the NDP and Liberals and Greens had all combined their votes, they would have won a total of 90 seats (instead of 48), an even larger majority. Specifically, in addition to the above 35 seats, they would have picked up an additional 7 seats from the Conservatives (barrie-springwater-oro-medonte, bay-of-quinte, kenora-rainy-river, markham-stouffville, newmarket-aurora, parry-sound-muskoka, simcoe-north).
CONCLUSION: In this Ontario 2018 election, using the hypothetical assumption that a combined party would keep all of the supporters of the original parties, vote-splitting among the three major center-left parties played a major role. Vote-splitting to the Greens affected 9 seats, but this was not sufficient to deny the Conservatives their majority. Vote-splitting between the NDP and Liberals was much more significant, affecting 35 seats, and a combined NDP-Liberal vote would have resulted in a large majority of 82 seats.
-- Jeffrey S. Rosenthal, June 8, 2018.
Addendum (June 9): Another perspective is: If the other parties' votes remained constant, and the NDP increased their votes by a uniform percentage in all ridings, then to win a majority of the seats, the NDP would have had to increase their votes by 42%. However, if the NDP and Green votes were combined, then to win a majority of the seats, they would only have had to increase their votes by 18% -- less than half as much.
Also, the pre-election aggregated poll predictions were quite accurate. For example, the CBC Poll Tracker predicted Con 38.7% (78), NDP 35.5% (45), Lib 19.6% (1), Grn 4.9% (0), and Calculated Politics predicted Con 38.4% (71), NDP 36.1% (44), Lib 19.5% (8), Grn 4.6% (1), both quite close to the final result of Con 40.49% (76), NDP 33.57% (40), Lib 19.59% (7), Grn 4.60% (1). Furthermore, the Conservative overperformance (relative to the polls) of about 2% was similar to their 3% overperformance at the federal level that I found in a previous paper.
See also a related article, and my recent interviews about election polls on CTV News, BNN Bloomberg TV, and U of T News.