Talk:Ascertainment bias

From Wikipedia, the free encyclopedia
Jump to: navigation, search

older entries[edit]

Oh, dash it all! The entry was created from a search for the phrase "ascertainment bias" and somehow the quote marks got into the title! Does anyone know how to remove them? Thanks. —The preceding unsigned comment was added by Jedwards01 (talkcontribs).

  • That problem seems to have been resolved Leevanjackson 01:58, 26 November 2006 (UTC)

Classroom example[edit]

I'm not sure what is meant by stating that the sample would be biased "by families with only a single [...] child (themselves).". The sample is also biased by the larger families who will have at least one female family member, so the average number of women per family will be - for all family-sizes - 0.5 too high. The problem is the same for small families as for big families (although the bias will be larger in a classroom with only small families). IMO the part "and is biased by families with only a single, female child (themselves)" adds nothing to the clarity of the example and should be removed. Elmarj 22:44, 15 July 2007 (UTC)

Merge with "Selection Bias"?[edit]

The term 'Ascertainment bias' seems to be the term that is used in the medical field, for something that is more widely known (at least within the social sciences) as Selection Bias. At least the article should refer to that entry. Elmarj 23:00, 15 July 2007 (UTC)

I just noticed that content is also overlapping with the Biased sample article. Elmarj 12:28, 16 July 2007 (UTC)

I wikilinked the related and associated terms; ascertainment bias is the term used in medicine, and the medical article is needed, separate from selection bias. SandyGeorgia (Talk) 01:13, 10 August 2007 (UTC)

Biased results[edit]

I'm bothered by the original line: "false results are produced by non-random sampling and conclusions made about an entire group are based on a distorted or nontypical sample." In fact, if you know there is an ascertainment bias, then you modify the statistical test to account for it (if you can). There is still ascertainment bias, but the results are not false since you have accounted for them. (talk) 18:47, 5 February 2008 (UTC)

Sources of bias[edit]

Is the all too human need for getting paid a possible source of bias?

Regarding Bisphenol A:

  • "It would be easy to conclude that the industry-funded scientists are all biased, whereas government-funded scientists are not, but it is not necessarily so clear-cut. Government-funded researchers could by biased by the need to secure research funds. If it looked like there was nothing to the BPA story, funds dedicated to investigating it would dry up." [1]

Do we have any articles on Sources of bias or on Scientific bias in general? --Uncle Ed (talk) 18:31, 28 May 2009 (UTC)

How is ascertainment different from exclusion?[edit]

From biased sample: "Exclusion bias is bias due to any exclusion of any particular group or type of subjects from the sample." With the exception of the caveman effect, the article appears to be about exclusion bias. The only reference to the literature so far is about exclusion bias. I suppose that at least some WP:RS define ascertainment in the broader sense of nonprobability sampling, but this needs to be sourced. Paradoctor (talk) 17:47, 19 November 2009 (UTC)