Here's one excerpt, on polling during the election, which the author titles, "the carnival mirror":
Funnelling the bewildering cacophony into a rubric of empirical order, 76 national polls ranking the relative standing of political parties and leaders were published during the 37-day campaign period.Must read.
CTV News and the Globe and Mail hired Nanos Research to conduct almost daily tracking, and pollsters like EKOS President Frank Graves were featured in regular appearances on current affairs shows like CBC’s Power and Politics. Increasing competition among polling firms, differences in methodology, and sampling problems produced wild variations during the campaign — on at least three occasions polls conducted during overlapping periods produced a 10-point spread in their estimation of Conservative support. Reporting on these results was occasionally submitted to scrutiny, but print and television reports routinely omitted information required to properly interpret these polls, such as the chasm of difference between national and regional margins of error, giving the impression of a race where there was none. Amidst the fervour to forecast the election (as if it were an inevitable weather system), seat projection sites like threehundredeight.com rose in popularity, utilizing models later devalued by actual election results.
Through their ostensible representative function, metrics of public opinion can serve democratic purposes. Nik Nanos, president of Nanos Research, argues that daily surveys conducted by his firm gave voters a voice by allowing them to identify the issue of greatest concern, setting a trajectory for public discussion. Yet the overwhelming aim of polling during the election was to gauge the relative standing of parties and party leaders. “I’m totally depressed,” complained an exasperated Graves, “[public opinion firms are] trying to explain the underlying social forces that are producing political change, and the media, despite protestations to the contrary, are much more interested in the horse-race side of things.” More than a “snapshot,” horse-race polls have become a ubiquitous electoral hermeneutic, reducing an intricate tapestry of collaborative visions into a sterile, competitive ordinal. A carnival mirror, this imagery entertains and distorts, crowding our limited repertoire of political images, warping the conception of our role in the democratic process from contemplative agents to nominal units.