The principle beneath the numbers

Any platform can claim to be unbiased. Few are willing to open the ledger and show their work. NorthVote705 exists on the conviction that credibility is not asserted — it is earned, formula by formula, in full view of the reader. What follows is the complete architecture behind every figure we publish: how it was gathered, how it was weighted, and precisely where our confidence gives way to estimation.

We ask you to read this page not as fine print, but as our founding document.

Two instruments, one purpose

NorthVote705 draws on two distinct analytical instruments, each built for a different circumstance.

The Aggregator assembles genuine, local polling data, riding by riding, into a single weighted figure — the clearest picture available when real evidence exists.

The Projector takes over where local evidence is silent, extrapolating from the broader currents of provincial and national opinion to estimate what a riding’s numbers likely are, absent a poll to confirm it.

The two are never confused with one another. A projection is never permitted to masquerade as a poll.

The Aggregator: how local polls are weighted

Not every poll deserves equal weight, and we do not pretend otherwise. Each poll entering our tracker is assigned a Final Weight, the product of three independent measures:

Final Weight =
Pollster Credibility × Recency × Sample Size

• Pollster Credibility — a rating from 0 to 1 reflecting a firm’s demonstrated accuracy and transparency of method. A pollster without an established record is treated with appropriate caution by default.
• Recency — confidence in a poll decays gently with time, on a 180-day half-life. A survey six months old carries roughly half the influence of one taken today; opinion, like all things, is perishable.
• Sample Size — larger, more statistically sound samples are given greater credence, though with diminishing returns beyond a certain scale, so that no single large poll can silently overwhelm the rest.

The riding-level figure you see is, in the end, a weighted average — every poll’s contribution proportional to how much it has earned our confidence.

The Projector: modelling what has not yet been measured

Where a riding lacks a recent local poll, NorthVote705 does not simply guess, nor does it leave the reader without a figure at all. Instead, we apply established methods of electoral swing analysis, extending a riding’s last confirmed election result forward in light of how province-wide or national opinion has since shifted.

We employ two competing methods, deliberately shown side by side rather than collapsed into false consensus:

• Uniform swing applies the same point-for-point shift to every riding, regardless of where it began.
• Proportional swing applies the same rate of change, scaled to each riding’s starting position — better behaved for smaller parties, though prone to overreach where a party began near zero.

When the two methods diverge, that divergence is itself the finding: it tells the reader precisely where genuine uncertainty lives, rather than concealing it behind a single confident number.

What our models cannot see

No formula can perceive what only a community can: the standing of a long-serving incumbent, the pull of a well-known independent, the texture of a local campaign fought door to door. Our projections are a floor, not a forecast — a reasoned estimate to be displaced the moment real local polling arrives. We say this plainly because a methodology that hides its own limits is not transparent; it is merely quiet about its blind spots.

Editorial independence

No sponsor, advertiser, or contributor is granted influence over our data, our weightings, or our written analysis, under any circumstance. Sponsored content is visibly and unmistakably distinguished from editorial content, never blended with it. Where a conflict of interest exists — including the founder’s other business interests or political affiliations — it will be disclosed here, plainly and without euphemism.

Corrections

We are not infallible, and we do not pretend to be. Where an error of fact is identified, it will be corrected publicly, with the nature of the correction disclosed rather than quietly erased. Read our full corrections policy →

A living document

Our methodology will be revisited after every election cycle, refined as pollster track records accumulate and as our own judgment is tested against results. This page will change accordingly — and every change will be dated and explained, not silently overwritten.