Large aggregate snowflakes an inch or so across formed from dendritic ice crystals sticking together are not uncommon, and according to research summarized by the New York Times, "Snowflakes as Big as Frisbees?" may be extremely unlikely but not impossible.
Larger aggregates fall faster, so as they fall they will sweep out and agglomerate smaller snowflakes, growing ever larger until broken up by wind or they land. To get the largest aggregates, one needs temperatures around freezing so the snowflakes are sticky, a very large density of snowflakes, and relatively still air. There is no simple theoretical upper limit on their size.
The Montana report is definitely sketchy, but William Pike's 1988 article on "Unusually-Large Snowflakes" noted that there are reports from "reliable observers" of 2-6 inch diameter aggregate snowflakes. (Pike was inspired to do the research after he observed 2-3 inch snowflakes himself in Vancouver, Canada.)
As noted in the New York Times article, the problem with both forming and documenting large snowflakes is that they are so fragile. Large aggregates are weakly bound, can be easily broken by air movements, and disappear on landing. This can be contrasted with large hailstones which can be very robust, so they stick around after landing and are easily photographed. Large snowflakes need to be imaged as they fall.
There are lots of videos online showing 1-2 inch aggregates falling. Interestingly, some of the best (such as this or this) do come from Montana. The apparent lack of videos or images of larger aggregate snowflakes shows that they are clearly very rare.
