|
Twitter is a silly source of data to analyse for polling purposes. There's an unknown number of bots, duplicate accounts, etc. There's also the huge potential for self-selection bias.
|
#
?
Nov 2, 2017 00:57
|
|
|
|
| # ? Jun 7, 2024 10:10 |
|
|
I mean it's not perfect, and the kind of people who are over 55 and on Twitter probably aren't representative of all people over 55, but it doesn't seem terrible. It would be good it it were written up as a proper article though.
|
|
#
?
Nov 2, 2017 00:59
|
|
|
That big data SSM study is a bit of a worry, I feel like if Australia votes no it will do one of those deeply disturbing people-are-not-like-I-thought-they-were things to my brain. Like when Trump won. I guess I live in a bubble and have too much faith in the people of this country.
|
|
#
?
Nov 2, 2017 00:59
|
|
|
Doesn't this imply that they know they have unlawfully elected MPs in the House and aren't addressing it?
|
|
#
?
Nov 2, 2017 01:00
|
|
|
Doctor Spaceman posted:The Conversation piece has slightly more detail loving mindblowing.
|
|
#
?
Nov 2, 2017 01:01
|
|
|
okay so this post will be hosed up because i'm not irony posting (i'm sorry) but we can't just dismiss research off hand because we don't a priori agree with it. We can critically evaluate it via methods (like where is the methodology/source research), context (money for grants/consulting, mad props for controversy, """political correctness"""), and pretext (worked once, but is it accurate this time), but i'm not super comfortable in just going "yeah this is poo poo because it's wrong" because it doesn't fit within what we really want. like (again i'm sorry this is really messed up) the conclusion of the article is that Australia is actually a really hosed up country where near the majority of voters don't think LGBTQIA people deserve equal human rights, which is fairly consistent with the stance of the current government on other "questionable" individuals, like First Australians, poor people, and refugees. I'm not sure about the margin of error here but I think that the conclusion of the article is probably within the realm of possibility, although (as they conclude): quote:So how accurate is our result? We will know on November 15 when the ABS announces the result of the postal vote. they are not really willing to stand behind their findings as scientists (for a variety of reasons). take that as you will
|
|
#
?
Nov 2, 2017 01:01
|
|
|
The Before Times posted:Twitter is a silly source of data to analyse for polling purposes. There's an unknown number of bots, duplicate accounts, etc. There's also the huge potential for self-selection bias. Seeing as you, a layperson not involved in the study, came up with those problems fairly easily, do you not think that people who do this for a living also came up with them, and tried to correct for those issues?
|
|
#
?
Nov 2, 2017 01:03
|
|
|
Teddybear posted:Doesn't this imply that they know they have unlawfully elected MPs in the House and aren't addressing it? Correct. They don't want an audit because they know they'll lose power as a result.
|
|
#
?
Nov 2, 2017 01:04
|
|
|
 uncanny
|
|
#
?
Nov 2, 2017 01:05
|
|
|
NPR Journalizard posted:Seeing as you, a layperson not involved in the study, came up with those problems fairly easily, do you not think that people who do this for a living also came up with them, and tried to correct for those issues? I'm just going to say that I and others in here to a greater extent have some grounding in analytics and quant methods and Twitter seems like an easily poisoned sample. I'm not saying I wouldn't read their actual research and scratch my head thinking about what it represents though.
|
|
#
?
Nov 2, 2017 01:06
|
|
|
NPR Journalizard posted:Seeing as you, a layperson not involved in the study, came up with those problems fairly easily, do you not think that people who do this for a living also came up with them, and tried to correct for those issues?
|
|
#
?
Nov 2, 2017 01:07
|
|
|
NPR Journalizard posted:Seeing as you, a layperson not involved in the study, came up with those problems fairly easily, do you not think that people who do this for a living also came up with them, and tried to correct for those issues? please don't do this!!!
|
|
#
?
Nov 2, 2017 01:09
|
|
|
CrazyTolradi posted:As Recoome says, they're not really standing by their results. Saying "well lets just wait two weeks to see if we're right" isn't really a vote of confidence into their methodology. What would be a great vote of confidence in their methodology then?
|
|
#
?
Nov 2, 2017 01:10
|
|
|
JBP posted:I'm just going to say that I and others in here to a greater extent have some grounding in analytics and quant methods and Twitter seems like an easily poisoned sample. NPR Journalizard posted:What would be a great vote of confidence in their methodology then? I already said SSM won't win because Australia so they're likely on the money.
|
|
#
?
Nov 2, 2017 01:11
|
|
|
CrazyTolradi posted:As Recoome says, they're not really standing by their results. Saying "well lets just wait two weeks to see if we're right" isn't really a vote of confidence into their methodology. What do you want them to say? They're publishing their findings and accepting that they aren't guaranteed to be accurate. I'd be more skeptical if they were more confident.
|
|
#
?
Nov 2, 2017 01:13
|
|
|
NPR Journalizard posted:What would be a great vote of confidence in their methodology then? If you can't say with any measure of certainty or margin of error without relying on the actual results of the SSM survey, the conclusion is ultimately meaningless. You could read tea leaves and say, "oh yes, this says no/yes will win by X% but we'll just have to wait and see". You might as well just poll SA accounts.
|
|
#
?
Nov 2, 2017 01:13
|
|
|
NPR Journalizard posted:What would be a great vote of confidence in their methodology then? Honestly they give a point estimate without any measure of variability, so we don't know and can't infer what the true mean really is. We cant tell, and they don't offer a measure, how accurate their prediction is. It's inferential statistics 101 so really depending on what you look at support is anywhere (and different) from 49%-72%. For the 49% prediction, it is possible that the actual mean is still above 50%. We just don't know because they don't tell us. If they had that I'd be more confident in the methodology
|
|
#
?
Nov 2, 2017 01:15
|
|
|
NPR Journalizard posted:Seeing as you, a layperson not involved in the study, came up with those problems fairly easily, do you not think that people who do this for a living also came up with them, and tried to correct for those issues? Claims they predicted the 2016 Australian election have no supporting information either. Note that polling was very accurate for the final 2PP result as well. It's barely worth calling an analysis.
|
|
#
?
Nov 2, 2017 01:16
|
|
|
Has anyone asked that octopus what it thinks the result will be? Seems like that would be just as reliable as the results of that study. E: Someone get the NT times on this, I sure they can wrangle up a croc or wild pig for a real aussie flavoured prediction. Bonus points if the Croc has to choose between eating a federal labor or LNP member to predict the result. Periphery fucked around with this message at 01:19 on Nov 2, 2017 |
|
#
?
Nov 2, 2017 01:17
|
|
|
the accuracy of literally all polls is determined by what the results are on the day. we develop polls to be as accurate as they can but in the end you’re relying on human-reported data which can be inherently inaccurate.
|
|
#
?
Nov 2, 2017 01:18
|
|
|
I'd also give this thing more credit if it was actually peer-reviewed, but it's not so lol. "Methodology? Where we're going, we don't need no methodology!"
|
|
#
?
Nov 2, 2017 01:18
|
|
|
Periphery posted:Has anyone asked that octopus what it thinks the result will be? Seems like that would be just as reliable as the results of that study. I think the octopus is dead  (((
|
|
#
?
Nov 2, 2017 01:20
|
|
|
I only listen to polls conducted by crocodiles in the northern territory news.
|
|
#
?
Nov 2, 2017 01:24
|
|
|
I would blow Dane Cook posted:I only listen to polls conducted by crocodiles in the northern territory news.
|
|
#
?
Nov 2, 2017 01:24
|
|
|
I actually wrote an article (which I didn't publish) where I basically argued that the way we teach measures of central tendency (mean, median, and mode) are all well and good, and we love averages, but we don't really cover variability in-depth at high-school (at least not my high school?). I'd argue that variability/accuracy is just as important as the mean/average itself, because averages themselves are "forced" into being, and don't reflect that the underlying data itself is varied and chaotic. The result is that we have a large bunch of people running around without being able to critically evaluate estimations/numbers. It's ironic that variance is often the first thing to be sidelined when examining statistics/predictions when it is actually one of, if not ~the~ foundation on which statistics is built on.
|
|
#
?
Nov 2, 2017 01:40
|
|
|
Recoome posted:It's ironic that variance is often the first thing to be sidelined when examining statistics/predictions when it is actually one of, if not ~the~ foundation on which statistics is built on. Really? Variance is an important part of literally every course I've seen in statistics.
|
|
#
?
Nov 2, 2017 01:53
|
|
|
EoinCannon posted:That big data SSM study is a bit of a worry, I feel like if Australia votes no it will do one of those deeply disturbing people-are-not-like-I-thought-they-were things to my brain. Like when Trump won. Auspol is certainly the place to be if you want to dissuade yourself of this opinion.
|
|
#
?
Nov 2, 2017 01:54
|
|
|
Zenithe posted:Really? Variance is an important part of literally every course I've seen in statistics. the person taking a stats course probably isn't the demographic I'm worried about. The large majority of people in Australia (even the world?) won't have taken a stats course so we can't really assume that everyone will understand variability (it also becomes intensely theoretical topic). Like I'm sure I covered standard deviation in high school or something but I don't remember it ever being linked to anything practical, which I think is a problem.
|
|
#
?
Nov 2, 2017 01:59
|
|
|
Losing a referendum, a moral victory Tony Abbott has criticised Malcolm Turnbull’s cabinet for refusing to campaign against marriage equality and warned that the Australian Conservatives, led by Cory Bernardi, will be the beneficiary of new “activated” conservative campaigners. In a speech delivered in New York to the anti-gay Alliance Defending Freedom group, which is classed as a hate group by the Southern Poverty Law Centre, Abbott has claimed that a 40% no vote in the marriage law postal survey would be a “moral victory”. Abbott claimed that polling for the no campaign showed that support for marriage equality had fallen to about 50%, with about 40% opposed and 10% undecided. The former prime minister said the no campaign was a “nucleus of an organisation” that could represent 40% of Australians and become a counterweight to the progressive campaign organisation GetUp.
|
|
#
?
Nov 2, 2017 02:00
|
|
|
Hello madam, I'm from the Alliance Defending Freedom. Papers please.
|
|
#
?
Nov 2, 2017 02:07
|
|
|
Hm I love how our failed ultra conservative politicians are able to freely make speeches to actual hate groups but look union corruption/bikies
|
|
#
?
Nov 2, 2017 02:08
|
|
|
“Win, lose, or draw, though, starting from scratch two months ago, the campaign for marriage in my country has mobilised thousands of new activists; and created a network that could be deployed to defend western civilisation more broadly and the Judeo-Christian ethic against all that’s been undermining it,’’ he said.
|
|
#
?
Nov 2, 2017 02:16
|
|
|
Lid posted:“Win, lose, or draw, though, starting from scratch two months ago, the campaign for marriage in my country has mobilised thousands of new activists; and created a network that could be deployed to defend western civilisation more broadly and the Judeo-Christian ethic against all that’s been undermining it,’’ he said. no loving way
|
|
#
?
Nov 2, 2017 02:17
|
|
|
Every journalist should have a statistics degree. Not just one or two subjects, but at least the six or so necessary to actually get it intuitively, and that include having to actually do non routine stuff with data.
|
|
#
?
Nov 2, 2017 02:18
|
|
|
Do these clowns know that Twitter is full of botted right wing troll swarms controlled by Russia/the FSB in order to make right wing views appear more mainstream? Because this is a thing that I recall from the US primaries/elections.
|
|
#
?
Nov 2, 2017 02:20
|
|
|
I think we can throw statistics onto the pile with bike helmets and bottled water. For example: As only my bottom, mouth and hands typically engage in gay sex I can confidently state that I, personally, are only around 20% in favour of gay sex. -/- Has nobody in the LNP twigged that the ALP is behind the current dual citizenship fiasco? I keep hearing LNP pundits confidently prognosticate that there must be problems in the ALP too. Well apart from a very unlikely error or possibly two the ALP have nothing to fear from an audit and know this very well.  Still I guess they too enjoy watching the muppets in charge keeping digging the hole deeper.
|
|
#
?
Nov 2, 2017 02:22
|
|
|
a baby walked up to me and said "i'm gay" which means yes is going to win because the truth comes out of the mouths of babes and not from pseudoscience involving Twitter Dot Com you absolute imbeciles
|
|
#
?
Nov 2, 2017 02:22
|
|
|
Cartoon posted:Has nobody in the LNP twigged that the ALP is behind the current dual citizenship fiasco? I keep hearing LNP pundits confidently prognosticate that there must be problems in the ALP too. Well apart from a very unlikely error or possibly two the ALP have nothing to fear from an audit and know this very well. As if Federal MP Tanya Plibersek is going to say "yes we should have an audit."
|
|
#
?
Nov 2, 2017 02:24
|
|
|
It's pretty clear that the ALP have some hand in this mechanically given that Dastyari engaged Iranians to go and hang out at the citizenship office to take some photos of them with his paperwork and asking questions. If this is all some underhanded Labor game, all I can say is N O I C E
|
|
#
?
Nov 2, 2017 02:29
|
|
|
|
| # ? Jun 7, 2024 10:10 |
|
|
Birdstrike posted:a baby walked up to me and said "i'm gay" which means yes is going to win because the truth comes out of the mouths of babes and not from pseudoscience involving Twitter Dot Com you absolute imbeciles "i don't like thinking because i am a big dumb babby who is smells and likes to smell my own bbutt" ~ Forums Poster Birdstrike, anno domini two thousand and seventeen
|
|
#
?
Nov 2, 2017 02:31
|
|