Guy Norman
@guynormanuk.bsky.social
Just someone in Norwich interested in politics and birdwatching and ridiculously exhausting adventures with backpack. And some other stuff, probably.
If that's what you've decided to think, fine. Have a good day.
August 23, 2024 at 5:34 PM
If that's what you've decided to think, fine. Have a good day.
Patrick: I've given up any attempt at reasoned debate on this thread, and deleted all of my posts. Like you, I abhor Israel's crimes in Gaza, and I will leave it at that.
August 23, 2024 at 5:25 PM
Patrick: I've given up any attempt at reasoned debate on this thread, and deleted all of my posts. Like you, I abhor Israel's crimes in Gaza, and I will leave it at that.
Adam, you were right, I was wrong! bsky.app/profile/awme...
This is a great question! The answer is both no and yes depending on your philosophical approach to statistics. So the MOE that pollsters report (e.g. +/- 4 points) is roughly equivalent to a 95% confidence interval around the level of support for a specific candidate. 1/
Loving your cat Andrew! But sneaky question... imagine poll in State X puts DT ahead of KH by 3%, with +/- 4% MOE. Checking my understanding: true value is most likely 3%, and with p=95% TV lies between -1% and 7%; but p of being TV declines from centre (3%) to extremes of MOE. Right? Thank you!
August 18, 2024 at 3:45 PM
Adam, you were right, I was wrong! bsky.app/profile/awme...
I'd like to say Pancho grateful too, but he doesn't give a monkey's!
August 18, 2024 at 3:33 PM
I'd like to say Pancho grateful too, but he doesn't give a monkey's!
... and that one poll is a ring-throw which should fall around the true value, but no reason to suppose the true value will lie at the centre of the ring. Understood I think. Again, thank you!
August 18, 2024 at 3:26 PM
... and that one poll is a ring-throw which should fall around the true value, but no reason to suppose the true value will lie at the centre of the ring. Understood I think. Again, thank you!
Blimey, that's a massively helpful and crystal-clear response, thanks so much Andrew! I see that I was wrong on at least two counts: I now get why the MOE of the difference is double the poll's MOE...
August 18, 2024 at 3:25 PM
Blimey, that's a massively helpful and crystal-clear response, thanks so much Andrew! I see that I was wrong on at least two counts: I now get why the MOE of the difference is double the poll's MOE...
[TV within the limitations of the survey, etc etc. Long response not required: just checking that "declines" aka "drops off" is correct understanding.]
August 18, 2024 at 8:53 AM
[TV within the limitations of the survey, etc etc. Long response not required: just checking that "declines" aka "drops off" is correct understanding.]
Loving your cat Andrew! But sneaky question... imagine poll in State X puts DT ahead of KH by 3%, with +/- 4% MOE. Checking my understanding: true value is most likely 3%, and with p=95% TV lies between -1% and 7%; but p of being TV declines from centre (3%) to extremes of MOE. Right? Thank you!
August 18, 2024 at 8:50 AM
Loving your cat Andrew! But sneaky question... imagine poll in State X puts DT ahead of KH by 3%, with +/- 4% MOE. Checking my understanding: true value is most likely 3%, and with p=95% TV lies between -1% and 7%; but p of being TV declines from centre (3%) to extremes of MOE. Right? Thank you!
I mean yeah, it depends what you're using the stats for, I think: are you centrally interested in the most likely estimate (as here), or are you looking for definitive proof (by convention, <5% prob of being wrong)? But even in the former case, I agree, MOE needs to be taken into account.
August 17, 2024 at 5:51 PM
I mean yeah, it depends what you're using the stats for, I think: are you centrally interested in the most likely estimate (as here), or are you looking for definitive proof (by convention, <5% prob of being wrong)? But even in the former case, I agree, MOE needs to be taken into account.
Here, assuming margin of error +/- 4%, I would understand that Trump's (current) lead in Georgia, per this poll, is with 95% probability in the range 0-8%. [Or could be 90% probability, not sure which cut-off they use.] But certainly true value more likely to be 4% than 0%. As I understand it!
August 17, 2024 at 5:37 PM
Here, assuming margin of error +/- 4%, I would understand that Trump's (current) lead in Georgia, per this poll, is with 95% probability in the range 0-8%. [Or could be 90% probability, not sure which cut-off they use.] But certainly true value more likely to be 4% than 0%. As I understand it!
I'm definitely not a statistics expert Adam, but as I understand it no: here, the most likely value is "Trump 4% lead in Georgia"... 3% or 5% are less likely, but more likely than 2% or 6%. Bell curve, basically. Actual stats experts feel free to correct me if I'm wrong.
August 17, 2024 at 5:29 PM
I'm definitely not a statistics expert Adam, but as I understand it no: here, the most likely value is "Trump 4% lead in Georgia"... 3% or 5% are less likely, but more likely than 2% or 6%. Bell curve, basically. Actual stats experts feel free to correct me if I'm wrong.
Well yes and no, I think. The indicated lead is likely right, but there's a small chance it's wrong. So the lead is a Real Thing, but not big enough to be a Certain Thing. Right? [Their "3 to 4 point margin of error" for a poll of this type presumably corresponds to 95% or 90% confidence limits.]
August 17, 2024 at 4:18 PM
Well yes and no, I think. The indicated lead is likely right, but there's a small chance it's wrong. So the lead is a Real Thing, but not big enough to be a Certain Thing. Right? [Their "3 to 4 point margin of error" for a poll of this type presumably corresponds to 95% or 90% confidence limits.]
(except Georgia, disappointing)
August 17, 2024 at 1:52 PM
(except Georgia, disappointing)
Your headline seems misleading Mark, unless I'm misunderstanding... NYT just published suggests Harris 5% lead in Arizona, 2% lead in NC, 4% behind in Georgia, 1% behind in Nevada. But certainly positive.
August 17, 2024 at 1:49 PM
Your headline seems misleading Mark, unless I'm misunderstanding... NYT just published suggests Harris 5% lead in Arizona, 2% lead in NC, 4% behind in Georgia, 1% behind in Nevada. But certainly positive.
Methinks Kay & Scaramucci under-credited there! But all four are excellent (Campbell & Stewart, Kay & Scaramucci, and the unaffiliated News Agents, UK and USA). BTW, I wish my surname was Scaramucci, so unfair!
August 16, 2024 at 5:37 PM
Methinks Kay & Scaramucci under-credited there! But all four are excellent (Campbell & Stewart, Kay & Scaramucci, and the unaffiliated News Agents, UK and USA). BTW, I wish my surname was Scaramucci, so unfair!
I'm not sure what all that tech-speak means Jon, but can't government tweeters simply use Hootsuite or similar to simultaneously parallel-post everything to (say) Bluesky, Threads and X?
August 16, 2024 at 1:32 PM
I'm not sure what all that tech-speak means Jon, but can't government tweeters simply use Hootsuite or similar to simultaneously parallel-post everything to (say) Bluesky, Threads and X?
August 16, 2024 at 1:19 PM
I just used a free Android app called Buffer to try parallel-posting on Bluesky and X. It worked. But I'm no tech expert, there are multiple possible issues around security, compatibility, ease-of-use. Not endorsing Buffer. I'd guess that pay-for Hootsuite offers more sophisticated functionality.
August 16, 2024 at 7:53 AM
I just used a free Android app called Buffer to try parallel-posting on Bluesky and X. It worked. But I'm no tech expert, there are multiple possible issues around security, compatibility, ease-of-use. Not endorsing Buffer. I'd guess that pay-for Hootsuite offers more sophisticated functionality.
Exacto! That is the scientifically correct analysis.
August 15, 2024 at 11:31 PM
Exacto! That is the scientifically correct analysis.
A nice approach would be to post primarily here and explicitly "mirror" to Twitter 😁
August 15, 2024 at 9:44 PM
A nice approach would be to post primarily here and explicitly "mirror" to Twitter 😁
Reality check: few major voices will leave Twitter until there's an alternative with analogous reach. What people (journalists, UK government, etc) could very easily do is systematically parallel-post to Bluesky and Threads (...), alongside Twitter. That could help erode Musk's monopoly.
August 15, 2024 at 9:39 PM
Reality check: few major voices will leave Twitter until there's an alternative with analogous reach. What people (journalists, UK government, etc) could very easily do is systematically parallel-post to Bluesky and Threads (...), alongside Twitter. That could help erode Musk's monopoly.
Bit harsh. It's Norfolk Rd, Norwich NR4 7TJ.
August 15, 2024 at 9:30 PM
Bit harsh. It's Norfolk Rd, Norwich NR4 7TJ.
Your article offers a very smart solution, I think: UK government and institutions should parallel-post to Bluesky and Threads (etc) alongside X. Meets their need to reach big audience, and at the same time can gradually erode X's toxic monopoly. 👍🏼
August 15, 2024 at 1:15 PM
Your article offers a very smart solution, I think: UK government and institutions should parallel-post to Bluesky and Threads (etc) alongside X. Meets their need to reach big audience, and at the same time can gradually erode X's toxic monopoly. 👍🏼
I don't generally favour the Oxford comma. It can sometimes help separate the elements of a complex list (of multi-word premodifiers, or of noun phrases). But in simple cases, I don't think it reflects prosody. So AP seems quite sensible on this one.
August 15, 2024 at 8:36 AM
I don't generally favour the Oxford comma. It can sometimes help separate the elements of a complex list (of multi-word premodifiers, or of noun phrases). But in simple cases, I don't think it reflects prosody. So AP seems quite sensible on this one.