Reno and Chicago had about 21% non-attending voting members. Montreal was about 15%. However, the trend isn't a straight line: Boston in 2004 had 19% non-attending voting members, and while Philadelphia in 2001 had 17%, San Jose the following year had merely 7%.
Here's how it looks for the years for which I have data. The first figure after the year and site is the number of attending voting members (including YA and any other voting class that could attend the convention if applicable). The second is the number of supporting members. The third is the percentage that is supporting. Non-voting members aren't including in any figures.
2001 Philadelphia: 4440 899 17%
2002 San José: 4498 340 7%
2003 Toronto: 3969 614 13%
2004 Boston: 4868 1107 19%
2005 Glasgow: 3884 394 9%
2006 Anaheim: 3937 532 12%
2008 Denver: 3524 630 15%
2009 Montréal: 3238 574 15%
2010 Melbourne: 1999 964 33%
2011 Reno: 3697 991 21%
2012 Chicago: 4249 1112 21%
2014 London: 4987 2397 32%
I welcome those new supporting members. I want them to become attending members when there's a Worldcon that they consider they can afford, and because the convention moves around, for most of them (freely admitting not everyone), there should be one at least once a decade within striking distance.