If Statscan knows that certain census tracts have commute mode shares that fall within a certain range, that suggests they know they exact number. I don't know where I can find that data though, census tract profiles only have commute mode share by place of residence, not place of work.