Fortunately, we have many seasons in which we have both Hart Trophy voting and All-Star voting results. This will shed a great deal of light on this. We know the highest-rated Hart goalie for 1934, 1937, 1939, 1940 and every season from 1942 to 1957 and beyond. But for now we'll stop at 1957, giving us 20 seasons to work with.
In these 20 seasons, only six times is the highest-rated Hart goalie the same as the First Team All-Star (Frank Brimsek in 1942, Bill Durnan in 1944, 1945, 1946 and 1949, and Harry Lumley in 1955). It's the Second All-Star goalie six times (Roy Worters in 1934, Earl Robertson in 1939, Brimsek in 1943 and 1948, Chuck Rayner in 1950 and Jim Henry in 1952), and eight times the Hartiest netminder was third in the All-Star voting (Tiny Thompson in 1937, Robertson in 1940, Turk Broda in 1947, Al Rollins in 1951, 1953 and 1954, Gump Worsley in 1956 and Terry Sawchuk in 1957). Clearly, different standards are at play between the voting for the two awards. You might not expect 100% agreement between the two if the standards were the same, but with this degree of difference, where the third-best goaltender was the most likely to be seen as the most valuable, there's clearly something going on. And it we look at the results in detail, it's pretty clear what that difference is.
|1934||WORTERS, Roy||.472||GARDINER, Charlie||.531|
|1937||THOMPSON, Tiny||.552||SMITH, Norm||.615|
|1939||ROBERTSON, Earl||.478||BRIMSEK, Frank||.779|
|1940||ROBERTSON, Earl||.354||KERR, Dave||.667|
|1943||BRIMSEK, Frank||.570||MOWERS, Johnny||.610|
|1947||BRODA, Turk||.600||DURNAN, Bill||.650|
|1948||BRIMSEK, Frank||.491||BRODA, Turk||.642|
|1950||RAYNER, Chuck||.486||DURNAN, Bill||.539|
|1951||ROLLINS, Al||.775||SAWCHUK, Terry||.721|
|1952||HENRY, Jim||.471||SAWCHUK, Terry||.714|
|1953||ROLLINS, Al||.493||SAWCHUK, Terry||.635|
|1954||ROLLINS, Al||.235||LUMLEY, Harry||.558|
|1956||WORSLEY, Gump||.529||PLANTE, Jacques||.734|
|1957||SAWCHUK, Terry||.618||HALL, Glenn||.629|
In every season but one, when the All-Star Goaltender (ASG) is a different man than the Most Valuable Goaltender (MVG), the (ASG) was a winning percentage (W%) that is higher than that of the MVG, and sometimes the difference is enormous. The voters of the time clearly had no problem giving Hart Trophy votes to goaltenders on bad teams (see 1940 and 1954 in particular), but not All-Star votes. The only season in which the MVG has a higher W% than the ASG is 1951, when Al Rollins did not play a full season but was still considered MVG. So MVP votes could go to a netminder who did not play a full season, but All-Star votes would not; see also in 1957 (when the difference in W% is the smallest) when Sawchuk played only 34 games. So in every season in which both the MVG and the ASG played a full season, the ASG always had a significantly higher W%; he played on a significantly better team.
It seems absolutely clear to me, then, that the voters were applying the following standards to these votes, in general:
1. The All-Star voting was for the goaltender who was the most proficient at his position.
2. The Hart Trophy voting was for the goaltender who contributed the greatest proportion of his team's success.
So if you have two goaltenders, one worth 10 points on a team that recorded 40 points, and another worth 15 points on a team that recorded 70, the latter would do better in the All-Star voting while the former would do better in the Hart voting. The former was more valuable to him team in the sense that he contributed a greater proportion of his team's success. This is the only way that the voting results for Earl Robertson in 1939 and 1940, Jim Henry in 1952 and especially Al Rollins in 1954 make sense. For purposes of the Hart Trophy, it was often most valuable in the sense of "imagine how bad they would have been without him." This is not the modern attitude, when generally a player cannot be seen as "valuable" unless his team did well, but it clearly a prevalent attitude in the time period we're looking at.
Now, the best Hart goalies in the pre-All-Star seasons were John Roach in 1924 (.435 W%), Clint Benedict in 1925 (.333), Roy Worters in 1926 (.529), 1927 (.375), 1928 (.523) and 1929 (.553), and Charlie Gardiner in 1930 (.534). It's possible that some of these men would have been voted First Team All-Stars if there had been such a thing at the time, but given this analysis, it seems that at least in 1924, 1925 and 1927 these netminders would not have been named the best, rather than most valuable in the sense defined above.
So it seems that Roy Worters should not be considered the best goaltender of the late 1920s based on his Hart voting records. These votes were not intended to say he was the best goaltender, but that he was the best relative to his team. And since his teams were usually not good, that doesn't tell us much about his absolute standing among his brethren netminders.