Tuning into Scotland’s Voice with Sub-Polls

Sub-samples from polls have been misunderstood for years. Afterall what credibility can anyone lend to a 100-person Scottish subsample from a UK-wide poll, wobbling with a ±10% margin of error?

It’s like hearing a single note of a song—an earworm that teases but doesn’t deliver. Everyone says they’re too noisy, too unreliable. In newsrooms and the podcast cutting floors the idea of never give weight to a sub-poll is rife.

But what if those fleeting odd sounding notes hold the opening bars of a symphony? Sub-polls aren’t the problem—it’s how we listen to them that's all wrong.

As I'll cover shortly, in 2014 the sub-polls were humming a tune and many in the media and the political classes ignored it until a 'full' Scottish poll confirmed the reality.

But rather than simply ignore the sub-polls as too wee and too random to matter maybe there is a way to weave those scattered beats into a weekly symphony of Scottish opinion. This is why it's time to rethink sub-polls.

A primer Let’s start with the basics why do we regard sub-polls as so shaky—it’s all about sampling error.

Imagine testing a coin to see if it’s fair, 50/50 for heads or tails.

Toss it 10 times, and you might get 7 heads, that's just a wild guess telling us nothing. It's a tiny sample, useless for certainty. Toss it 100 times, maybe 55 heads, and you’re closer but still wobbly, with a ±10% margin of error.

That’s your typical sub-poll: 100 Scots in a UK poll, too small to pin down the truth.

Now toss that coin 1,000 times, and you’re at 505 heads, razor-sharp with just ±2.5% error. Bigger samples mean more precision, like hearing more notes of a song to catch its tune.

Single sub-polls rightly get dismissed because their small size makes them too noisy, but that doesn’t mean they’re worthless—they’re raw beats, waiting to be mixed into a symphony.

And those beats can carry a signal.

The 2014 signal

Rewind to 2014, just after Scotland’s independence referendum delivered a No vote. UK-wide polls’ Scottish sub-samples, each a mere 100 people, started humming an unexpected earworm: the SNP was surging.

One sub-poll showed it—easy to shrug off as noise. Two? Still dubious.

But when that tune (the SNP's risin') kept looping, sub-poll after sub-poll, week after week, it was clearly no fluke. It was the melody of a seismic shift in Scottish politics.

But nobody really took it seriously. We waited for a 'proper' Scottish poll, with its ~1,000 respondents and ±3% error, to pump up the volume on the speakers and confirm the SNP’s rise—a wave that reshaped Scotland.

Those sub-polls were singing the truth all along, an earworm we ignored because we didn’t know how to tune into in. They weren’t junk; they were vibrant snippets of Scottish sentiment, packed with a beat we could’ve caught earlier.

The mixing desk

So why do we keep ignoring sub-polls? Each one’s a sample, a raw loop of what Scots are thinking. Glance at enough of them, and you might spot a pattern, like the SNP’s 2014 climb. But glancing isn’t enough—those sub-polls are a pulsing weekly signal Scotland overlooks while we wait for a rare Scottish poll to drop.

There’s is perhaps a better way to hear the full track. Imagine a DJ at the decks, grabbing those choppy samples—say four UK polls a week, each with ~100 Scots—and mixing them together (circa 400) then layering them over four weeks into a seamless blend of circa 1,600 voices. That’s the #SubPollSeries: a method to turn fleeting riffs into a clear, harmonious symphony, delivering a weekly barometer of Scottish political opinion.

Let’s go back to that coin toss to see why this can work.

A single Scottish poll, with ~1,000 respondents, is like tossing a coin 1,000 times—pretty good, with a ±3% margin of error. But even then, 1 in 20 polls might be a rogue, skewing the picture.

Now imagine four people, each tossing that coin 100 times, every week for four weeks—16 sets, 1,600 tosses total. Each weekly result per person isn't accurate enough, but blend them together, and you get a ±2.5% margin of error, tight and clear.

I would argue that there is more, that averaging those sets smooths out the flukes. One person’s bad toss gets balanced by the rest, making a wild miss far less likely than in a single big toss. That’s the power of chaining sub-polls in this way: it’s not just a potentially bigger sample, it’s a steadier beat, weaving four UK polls a week into 1,600 Scottish voices.

So there is value here, indeed there is really potentially explosive value here. While Scotland hangs on for one poll to drop (maybe once a month), those sub-polls give us vibrant data every week. Chained sub-polls taps into that data, spinning a weekly barometer that’s (arguably) as reliable as any one off Scottish poll.

Using the #SubPollSeries at the last General election I predicted the SNP would be on 14 seats and Labour 37. Whilst not exactly spot on it was far more accurate than many individual polls for Scotland were predicting.

Take Sub-Polls Seriously

Clearly monthly polls are valuable, giving us a clear snapshot with their ~1,000 respondents. My point is not that Sub-polls are better, but simply that we should not ignore them. When mixed like the #SubPollSeries they complement our existing polls whilst offering a more frequent pulse of what Scots are thinking.

Why wait a month for one picture when you can tune into a weekly rhythm? My series is just one approach, born from a bit of time and effort to gather those sub-poll beats. Others can find their own ways to mix the data—anyone with a knack for it can be that statistical DJ.

So can we stop dismissing sub-polls as noise. They’re not perfect alone, but they’re far from useless. In 2014, they hummed the SNP’s rise before anyone else heard it. Today, they’re a weekly signal worth listening to (because Reform are on the rise and it was obvious before the Scottish polls came out) and it's capable of being spun into a symphony of Scottish opinion.

The #SubPollSeries shows one way to do it. But it’s not the only way. The data’s there, pulsing in every UK poll. Grab it, mix it, and hear what Scotland’s saying—not once a month, but every single week. Who’s ready to start spinning the decks with me?