User:Bondegezou

I created Jewish Indian theory in 2022 and that also made DYK on 16 September 2022.

Things I'm interested in

... but haven't edited (much):

Other things I may want to reference: Wikipedia:Templates_for_discussion/Log/2012_October_8#Template:DWspinoff; Wikipedia:Categories for discussion/Log/2018 March 11; Wikipedia:Articles for deletion/List of people with coronavirus disease 2019

Sue Gardner's blog on expertise and transgendered issues: [2]

Article on

Belle Gibson

Wikipedia:Categories_for_discussion/Log/2021_May_6#Category:People's_Alliance_of_Tower_Hamlets_councillors

The

Wikipedia_talk:WikiProject_Elections_and_Referendums#RfC_on_leader_since_election_infobox_parameter

Health data analysis teaching

Florence Nightingale

Ida B. Wells and her use of statistics in The Red Record: Tabulated Statistics and Alleged Causes of Lynching in the United States

Notes on 'margin of error' in opinion polls

The following notes were inspired by a discussion on

Margin of error is a familiar term in opinion polling, but often a misunderstood one. Statisticians usually talk about a related concept, the confidence interval. A poll is based on a sample from a population: we use the proportion in the poll supporting party X (or whatever) to estimate the proportion doing so in the population. But that estimate won't be spot on because of sampling error. Assuming we have sampled well (meaning randomly), the sample proportion is an estimate of the population proportion, with the confidence interval/margin of error describing the precision of that estimate. A 95% confidence interval is defined such that the confidence interval for the sample estimate will include the true (population) value 95% of the time. (We can also calculate 99% confidence intervals, or 80% confidence intervals, or whatever, but we usually just use 95%.) The margin of error is then half the confidence interval, so a value may be reported as 40% ± 3% (margin of error), meaning the confidence interval is 37%-43%. This means the true value could be outside that range, but usually (95% of the time), the interval will capture the true value. The only problem is we can't tell from a single poll whether it's one where we were unlucky and the confidence interval doesn't capture the true result.

The precise calculation of confidence intervals is a large topic. However, if we presume a proper sampling technique, for large samples, of large populations, the margin of error for a proportion can be easily calculated: see margin of error for details.

The margin of error and confidence interval vary with the sample size. The details depend on what you are doing, but generally the margin of error is inversely proportional to the square root of the sample size. (Thus, to get half the margin of error, you need four times the sample size.) Most political opinion polls are done on a similar number of people (so as to achieve a suitably small margin of error), but some are bigger or smaller, so you need to see what margin of error is reported for each poll, or do the maths yourself.

The margin of error for a proportion, like vote share, also depends on the vote share. We get better estimates of proportions near 0% and 100% than in the middle. The margin of error is smaller for a party with a vote share of 10% (or 90%) than one on 50%. The margin of error quoted is, thus, usually the maximum margin of error at 50%.

So, consider the following examples. At

Now let's consider the 21-2 Dec 2011 YouGov poll of 1721 people [4]. This puts the Conservatives on 40% and the Liberal Democrats on 9%. I note the YouGov pdf doesn't even give the margin of error, but we can work it out. Thus, the maximum margin of error is 2.36%, but the margin of error for the Conservative vote is 2.31% and for the Liberal Democrat vote, a mere 1.42%.

With me so far? Good. Because it gets more complicated. Often, we're interested in the difference between two parties, e.g. the lead Labour may have over the Conservatives. We could look at the margin of error for each party, but the margin of error for the difference requires a separate calculation: see Margin of error#Comparing percentages. So, let's take the 18-9 Dec 2011 YouGov poll, again on 1721 people, so we know the maximum margin of error is 2.36%. The Conservatives are on 38% and Labour are on 42%, so the difference between them is 4%, larger than the maximum margin of error. But the margin of error of the difference between the two parties is 4.22%, bigger than the difference between them.

Often, we have more than one poll, at which point we can consider combining multiple polls. When we do that, we're effectively making the sample size bigger, so the margin of error falls. If we have, say, 5 polls within a few days and they all show two parties within the margin of error, the combination of all 5, by reducing the margin of error, may show the two parties as being significantly different. See [5] and [6] for details.

This all, of course, presumes the polling was done well. The margin of error only considers the sampling error under certain assumptions. In practice, polls are usually less accurate because of methodological challenges in doing them right: see [7] for discussion. Some polling strategies appear to be generally more reliable than others. Thus, in the UK, phone polls appear to perform better than online polls.

Doctor Who

Useful cite: [8]

There are many long-standing and deep-rooted issues with Wikipedia coverage of articles pertaining to Doctor Who. I have been guilty of these as well. There is much interesting and entertaining material that would appear better suited to something like the TARDIS wikia, but which does not meet Wikipedia's policy and standards. Here are some developing notes on what I see as the core matters.

• WP:OR
  tendency). Egregious examples include the chronology article and the list of Dr Who companions.
• reliable sources
  , no content beyond basic production details + plot, and no evidence of notability. Wikipedia is not a list of indiscriminate information. Many of these could be moved to tables in articles describing series.
• WP:V (e.g. not wishing to call Jenna Coleman
  "Jenna" rather than "Jenna-Louise" until the day she is credited so in an episode).
• The
  WP:RS
  .

Notes: Wikipedia:Articles for deletion/List of Doctor Who historical characters

Election notes

This article may support a campaign section on 2015 general election article, noting extensive discussion of hung Parliament as possibility and resultant "The Fear" campaigning by the Tories. Plus this.

This user has publicly declared that they have a conflict of interest regarding the Wikipedia article Caroline Deys.

This user has a pet cat.

KC	This user is a member of WikiProject King Crimson.

This user enjoys reading The Signpost.

This user writes for The Signpost.

<6	This user is a scientist with an Erdős number of 5 (or less).

ORCİD

This user has the ORCID identifier: 0000-0002-6200-8804