Talk:2024 Sri Lankan presidential election

Errors/possible misunderstanding of reported opinion polls by Institute for Health Policy
IHP's voting intent estimates, which are reported in the Opinion polls section, are produced using multi-level regression and post-stratification (MRP) polling, which is a modeling approach to tracking public opinion. See Multilevel regression with poststratification for Wiki's write up, and the citation in that article to a New Scientist article discussing the YouGov use of MRP (https://www.newscientist.com/article/2134144-how-yougovs-experimental-poll-correctly-called-the-uk-election/#ixzz63vulf5ZP).

As the IHP reports, linked in the table, indicate, the sample sizes for the MRP estimates are the full sample over several months, so typically 10,000+. Wiki authors should consider this when reporting sample sizes. The actual full samples used in its MRP estimations are given in the linked documents.

As MRP polling is different to standard sample polling, this may make comparisons with other standard polls misleading. But this seems to be a generic issue in how polls are reported for elections in Wiki. Maybe a footnote should be added to the table to identify and elaborate on which of the polls use MRP modeling, with additional details. 49.45.134.55 (talk) 09:46, 10 February 2024 (UTC)