User talk:Centamia/sandbox

CHRISTIAN SCIENCE STAT JUNK

The arguments here are fascinating because Editors are arguing about apples and oranges. COLLECTORXXX has battled for statistical reasoning about ADHERENTS (as did I) while SlimVirginXXX wants the lowest number possible (thus MEMBERSHIP numbers or gross uncertainty.) It's so much like my statistical workaday world. Yuck!

TWO STATISTICAL PROBLEMS

There are two statistical problems operating in these arguments: (1) confidence/certainty/reliability and (2) unit of observation.

1) CERTAINTY/CONFIDENCE . . . and WIKIPEDIA-ESE RELIABILITY.

EDITORS are looking for an absolutely drop-dead certain number of Christian Scientists and you won't get it EVER. Christian Scientists are people and as people are not subject to the same mechanical rules of sheet metal. People have free will and can change their minds (they may not even know their minds!); sheet metal has no mind and cannot change what it is. COLLECTORXXX made that point weeks ago.

To make membership and adherent counts congruent requires cocktail napkin math. ???XXX recognized that church membership and adherent number estimates CANNOT BE MADE CONGRUENT without a bit of cocktail napkin math. It's clumsy, but if you must have such a number, this is what you get.

'METHODOLOGICAL REALITY: if you cannot put bounds on how sure you are, you really don't know what you're saying about a statistical estimate.'

(1) 'Adherents' counts are based on verifiable scientific method which allows for a measure of confidence.

(2) 'Membership' counts are based on cocktail napkin math which, insightful if that's all you have, offers no confidence in the estimate but only reliability because of the original source and your acceptance of the source's logic.

(3) 'Reliable sources' are considered reliable because they are considered prima facie reliable by a Wikipedia-ese consensus--it is dumb, but it is the Wikipedia-ese way.

(If you are a hard-core Bayesian, I think I tried, but please suggest at my sandbox how I might better have put this. If you can translate Bayes into Wikipedia-Speak, so much the better, my prior being that my skull is turning at all of this. Beware commentingas you'll quickly find yourself engaged in the statistical prior of Wikipedia-ease "reliability".)

2) UNIT OF OBSERVATION.

Editors are arguing about different units of observation. 2013 membership (whatever that EXACTLY really is) is not the same as 2013 adherents (whatever that EXACTLY really is). And, estimated observations in 1936 membership is only that--an estimate from more than seven decades ago (whoever responded is very likely long dead now, and that is common sense, not original research as XXXWOLFE so often likes to argue about anything that is common sense.)

STATISTICAL IMPLICATIONS.

"CARD-CARRYING" CHRISTIAN SCIENCE MEMBERS."

Per Church: 400,000 members - positive POV so may overestimate. Per Church critics: <100,000 members and declining rapidly -- negative POV so may underestimate.

"SELF-IDENTIFIED AS CHRISTIAN SCIENCE ADHERENTS."

Per ARIS: 339,000 self-identified - no POV but previous years (1990, 2008) suggest the possibility of overestimation for 2008.

'RECOMMENDATION.'

(1) In any case, card-carrying members is less than 1% of total church membership U.S. nationwide.

(2) In any case, the number of self-identified CS persons is far less than 1% of all Christian identification. [339/173,402 = .002 = 0.2% of Christians.]

As there is a strongly negative POV to this article, I suggest that SlimVirgin who rides herd on this article might like this: [339/229,182=0.0015 = 0.15% of U.S. religious identifieds.]

As a statistical practitioner, that's all I've got for Wikipedia.

XXXXcentamia

BIGGER ISSUES:WHAT IS RELIABLE

GETTING OUT OF THE WIKIPEDIA BOX: Journalistic standards versus Statistical standards.

Wikipedia is incestuous--buy your facts outside the Wikipedia house then come back and fight about what to do. In my seeking help from Wikipedia editors, one has sent me with a tongue-lashing to an appalling Wikipedia board. Let's think outside Wiki-Think (I should copyright that term.) There are a host of problems that super-recognized sources (if Wikipedia had such a term) worry over, and they are the result of bias manipulations as well as well as ignorance of bias. See, for example,

http://isites.harvard.edu/icb/icb.do?keyword=k70847&pageid=icb.page346376

JOURNALISTIC STANDARDS IN A NUTSHELL. http://www.pulitzer.org/2006_luncheon_steiger

Second topic: clarity of sourcing. One conclusion that an outside observer could confidently draw from watching the actions of this [Pultizer Prize] board over the last decade or so is that it consistently believes readers must be able to understand where information is coming from.

So, if you witnessed something personally, say so. - If you are recreating events by interviewing the parties to them, say so. - If you are borrowing someone else's facts or ideas, say so. - If you must grant some of your sources anonymity, say so.

The omniscient narrator has a well-founded role in fiction. When it appears in non-fiction or journalism, the work in question is unlikely to win a Pulitzer Prize.

STATISTICAL SAMPLING STANDARDS IN A NUTSHELL. http://www.scholarpedia.org/article/Sampling_bias

The primary concern is to eliminate bias in the sampling. If you have bias problems, then you must (ME: if you have any self-respect as a statistician) make adjustments or warn of the bias. Here is an easy vetted answer (ME: thus not derived by bullying or dumbed-down consensus):

LOGIC VS. WIKIPEDIA.

Other than these verbal or mathematical standards, I think this fairly covers what is legitimately logical. All other appeals are curious and could be logical fallacies. For example, the appeal to the NYT as accurate is nothing more than an appeal to authority. That is the core problem with Wiki-ese, the appeal to authority: if it's published it must be true without any other consideration. http://www.csun.edu/~dgw61315/fallacies.html#Introduction

There are a host of further analysis of fallacies in David Hackett Fischer's book, Historians' Fallacies.

Citation format:

^ "Table 75. Self-Described Religious Identification of of Adult Population 1990 to 2008" (PDF). United States Department of Commerce, United States Census Bureau. Retrieved 28/02/2013. — Preceding unsigned comment added by Centamia (talk • contribs) 09:08, 28 February 2013 (UTC)