Your data is not relevant to the hypothesis. The surveys were conducted in different time-frames (Canada: 2009-2014, USA 2005-2010), which I shouldn't need to tell anyone why that dataset is not useful for rigorous comparison, but I will anyway: one surveys people who were just recently hit by a recession and tracks them throughout the economy's recovery; while the other surveys people who recently experienced massive economic growth (exuberance) and tracks them throughout their economy crashing. Both, Canda's and the USA's, economies were intertwined during the 2007-2009 financial crisis, and this should be accounted for -- but it isn't.
Furthermore, the data DOES hold up to the GP's assertion: that as the unemployment rate goes up, the gap in life satisfaction between employed and unemployed approaches zero (chart 2). That is, 20% nationwide unemployment, unemployed and employed Canadians have the same life satisfaction (chart 1).
Lastly, this is social sciences. We cannot conclude anything from the data without experimentation. At best, this data is observational reason to pursue deeper inquiry, at worst it's idealogical fodder to drive other people's agends by lying with statistics. Everyone does it. Statistics don't really mean anything on their own, and lay-people give them too much weight!