As the byelection stakes continue to rise with every passing day, the two major parties are sparing little expense in their efforts to woo the voters of Canning (as my colleague Josh Taylor enumerates elsewhere in Crikey today).

Rail line extensions, new police stations and major road duplications are among the baubles being dangled before voters who appear to hold the immediate fate of a prime ministership in their hands.

However, such largesse comes at a substantial cost, with the residents of Armadale, Mandurah and surrounding localities suffering a daily avalanche of junk mail and disruptive congregations of politicians and journalists at their local shopping centres (this eloquent passing motorist, it may be assumed, spoke for many).

Worst of all, the phone won’t stop ringing.

Thanks largely to the phenomenon of low-cost robopolling, no fewer than seven opinion polls have been published since the ball got rolling on the byelection campaign four weeks ago.

Based on conservative assumptions about response rates, which are low and getting lower, these polls would have involved around 90,000 phone calls being made to an electorate of 112,000 voters, to elicit a combined total of 5410 completed responses — and that’s without factoring in the similarly voluminous private polling conducted by the political parties and other interested organisations.

Two published polls have appeared over the past 48 hours, and they have once again found the Liberals to be standing on the precipice of defeat, without quite toppling over it.

The Ipsos poll for today’s Fairfax papers and the Galaxy poll for the Sunday News Corp papers each credited Liberal candidate Andrew Hastie with a two-party lead over Labor’s Matt Keogh of 52-48, assuming minor party and independent preferences flowed as they did in 2013.

This would rattle Liberal nerves, given that both results are within the polls’ respective margins of error, of about 4% in Galaxy’s case and 3% for the larger-sample Ipsos poll.

However, not only do the two results corroborate each other, they are also of a piece with the five polls published earlier in the campaign, each of which gave the Liberals a similarly narrow two-party lead.

The Ipsos poll is the only one of the seven that was not a robopoll, having used live interviewers to target its sample of 1003 — with results all but identical to the Galaxy robopoll, conducted at around the same time from a sample of barely more than half the size.

Even without the impact of the weekend’s eruption of leadership speculation, which came too late to influence the polling, these figures are slightly at odds with the chatter that has been emerging about both sides’ assessments of their prospects, which has consistently credited Hastie with a little more breathing space.

Despite the noted uniformity of the headline results, Ipsos and Galaxy are also at odds with the three polls conducted earlier in the campaign by ReachTEL, which arrived at the same two-party destination by a somewhat different route.

ReachTEL’s polls have had the Liberal primary vote occupying a narrow band of 46.5% to 47.3%, whereas the other robopolls have ranged between 41% and 44% (Ipsos has it slightly higher, at 45%). But when ReachTEL probed minor party and independent voters for a next preference, less than a quarter favoured Liberal over Labor, compared with 48% at the 2013 election and around 40% in the Ipsos poll.

The explanation for the primary vote discrepancy most likely lies either in the pollsters’ demographic weighting methods — with ReachTEL, perhaps, calculating that older voters will constitute a higher share of the overall turnout — or the way they have designed their surveys.

On the latter count, ReachTEL appears to be an odd man out in not allowing for an uncommitted option. Those who do not wish to jump off the fence, either with respect to the primary vote or the follow-up question on preferences, can only do so by hanging up.

The primary vote variation allows at least some scope for pollsters to play for bragging rights once the true result is known.

However, it certainly can’t be ruled out that the entire polling industry will prove to have been wrong across the board. The narrow range of headline results brings to mind the “herding” phenomenon that afflicted polling before the British election in May, when every player in the game was shown to be singing false notes from the same song sheet.

Otherwise, it would appear that Andrew Hastie finished the penultimate week of the campaign with his nose in front of his Labor rival.

The week that remains, however, looks to be shaping up as the proverbial long time in politics — so much so that there is no guarantee that the byelection will even take place, should there be anything to speculation that the Prime Minister is considering pulling the rug from under a looming leadership putsch by rushing to an immediate double dissolution election.