Talk:2017 Montana's at-large congressional district special election

Google Consumer Surveys

I feel like the Google Consumer Surveys polls should be removed. A) because I don't see the margin of error listed anywhere, B) because I'm not sure it is reliable, and C) because the numbers on the link don't correspond to the results listed on our article which is why one of the polls adds up to 101% on our article. Prcc27 (talk) 12:12, 16 May 2017 (UTC)

Oppose. Margin of error is listed for each tail for each candidate. Also, the reason it adds up to 101% is because of rounding. 66.234.203.99 (talk) 00:17, 18 May 2017 (UTC)

Oppose. To address your concerns (and reiterate some points made by 66.234.203.99): A) margins of error are listed on the polls themselves. B) I'm inclined to be lenient—I'm no polling expert but I am encouraged because they do include the 'crosstabs' and allow sorting by demographic and the methodology of Google Consumer Surveys is reasonably understood, i.e., asking people within a geographic region before reading news articles. Its perhaps not as mature as the science of traditional phone polling, but Internet polls are an emerging field and discounting them altogether would be excessive. C) One of the Emerson polls sums to 101, and one of the Gravis polls sums to 99; it's just a function of rounding—I'm actually in favor of throwing the aesthetics of integers out and inputting the numbers as they actually are, unrounded. Specifically, it seems the only case where "the numbers on the link don't correspond to the results listed on our article" is the May 8-9 poll. However, in that instance they do if the numbers are recalculated discounting the "I am not likely to vote in this election" option—essentially just making it a poll of LVs. One way around this problem would be including an "abstentions" column like the article on polling for the French presidential election—however, I am disinclined to truly endorse this as I have yet to run across a polling section for an American election that includes abstentions as most American polling approximately confronts that question as either RV or LV polling.

Support Margin of error is shown and rounding isn't an issue. However, Google Survey Polls are not taken seriously. Real Clear Politics for example does not include them in their polling. Internet polling tends to be unscientific and does not take a good sample, hence why the results of those polls are so divergent. If more evidence of this is needed, looking at the polling data for 5/8 - 5/9, more than 2/3 of the sample was female. This is clearly a poor sample. MattCloudy (talk) 01:24, 25 May 2017 (UTC)

Support Google Consumer Surveys don't meet the standard of a true political poll or a reliable source. Google Surveys are a market research product. Market research is different doing political surveys. There is no science behind them and no rigorous method associated with them. The method of collection in this case is extremely sensitive to false and biased responses. As well, the number of google surveys presented tends over-weight them within the article as a source. Articles should use reliable sources whose names stand behind the polls they conduct. 75.17.126.24 (talk) 02:00, 25 May 2017 (UTC)

Oppose this moment but I might change my mind if their results turn out to be very divergent from today's result. As far as I can tell, the other pollsters don't seem to have had much of a track record polling them either and in any case, polling for special elections is an erratic business. Gravis have polled only one previous one and got the winner right but the margin was way out. Anywikiuser (talk) 16:45, 25 May 2017 (UTC)

Support Google Consumer Surveys are too informal and erratic to be treated as serious election polls. 12.12.144.130 (talk) 17:40, 25 May 2017 (UTC)

Oppose/Keep - it's a poll? yes. It's widely reported? yes. That's all that matters. Everything else is original research. Rami R 18:59, 25 May 2017 (UTC)

A Longer Comment

FiveThirtyEight referenced the Google surveys in their article on this race today, but said they have "some methodological issues". [1]. There is every possibility of a "house effect" in these polls. Google Surveys is well-calibrated nationally but not always on the state level; sub-populations with low computer use (such as farmers or ranchers) are often under-sampled, and they tend to be more conservative in states like Montana. There's also a second, independent issue that there may be withheld polls (as anyone can commission these polls privately), allowing the file-drawer effect to cause bias. Power~enwiki (talk) 19:25, 25 May 2017 (UTC)

FiveThirtyEight also mention that Gravis (the only other recent polster) were off by 20 points the last time they did a special election survey.[2] If we take FiveThirtyEight as gospel, we should just scrap the poll section entirely. Rami R 19:40, 25 May 2017 (UTC)

As a political analyst, based on the late-breaking news I would advise ignoring the polls entirely at this point. For an encyclopedia it may be worth keeping them. Power~enwiki (talk) 21:07, 25 May 2017 (UTC)

Google Surveys are *not* polls in any sense of the word. As far as Gravis, they are not perfect, but they still predicted the correct result even though their numbers were off by 20 points. On the other hand, the google survey results in this election were consistently and wildly wrong in one direction. Six out of seven results predicting a vastly incorrect outcome at odds with all the other polling being done suggests serious problems with their method. The worse problem is that attributing them as "google" surveys in the article is distorting what these surveys are and who is responsible for them. 198.45.174.250 (talk) 05:53, 26 May 2017 (UTC)

References

Page title

The page title seems unwieldy. Montana's at-large congressional district special election, 2017.... do we really need to say "at large"? Montana's congressional district special election, 2017 covers this just fine, since there is no other Montana congressional district. Valenciano (talk) 22:20, 13 June 2017 (UTC)

@Valenciano: The official name is "Montana's at-large congressional district." Precedent is to use the full title of the district. See Rhode Island's at-large congressional district special election, 1808, for example. MB298 (talk) 23:31, 13 June 2017 (UTC)

Comment about early voting

Gianforte was charged with misdemeanor assault, as noted in the article. Some people wanted to turn against him in this election, but those who had already sent in early voting could do nothing. Be aware of such pitfall if you wish to vote early. Carlm0404 (talk) 17:56, 24 October 2020 (UTC)