Peppermint Patty Posted November 7, 2013 Posted November 7, 2013 I've never heard of Bayesian calculations. I'm trying to figure this all out, and I humbly submit this is giving me a massive headache. Can someone with a statistical background shine some light on exactly what this means: http://www.mormoninterpreter.com/a-bayesian-cease-fire-in-the-late-war-on-the-book-of-mormon/?utm_source=feedburner&utm_medium=feed&utm_campaign=Feed%3A+mormoninterpreter+%28Interpreter%3A+A+Journal+of+Mormon+Scripture+%28RSS%29%29 Hence P(H|E) increases at most to around 0.6 from 0.5. That is, the occurrence of E ( = LW has the highest similarity score of all 5000 books in the corpus) increases the likelihood of H ( = a direct influence of LW on the Book of Mormon) from a 50-50 situation to a 60-40 situation. No slam dunk has occurred. Does this mean the author feels that there is a 60% likelihood that The Book of Mormon used the LW as a direct influence? In my opinion, P(E|H) ( = the probability of the evidence assuming that LW is a direct source for the Book of Mormon) is at most 0.8. Even if LW is a source for the Book of Mormon, a book other than LW may well be identified as more similar to the Book of Mormon for several reasons What does the 0.8 mean? Is this the same as the 60% probablility?
Peppermint Patty Posted November 7, 2013 Author Posted November 7, 2013 Okay. My husband is trying to figure this out with me. I think the with the 0.8 probablitly, the author is stating he feels at most, Bayesian calculations show there is an 80% probability LW was a direct source for The Book of Mormon. However, the author's own opinion is that there is a 60% probability the LW was a direct source for The Book of Mormon. Do I have it right?
cdowis Posted November 7, 2013 Posted November 7, 2013 (edited) I think the conclusion is about "similarity", not whether it is a "source". The former is a statistical conclusion (objective), while the latter is subjective. Edited November 7, 2013 by cdowis
SeekingUnderstanding Posted November 7, 2013 Posted November 7, 2013 Okay. My husband is trying to figure this out with me. I think the with the 0.8 probablitly, the author is stating he feels at most, Bayesian calculations show there is an 80% probability LW was a direct source for The Book of Mormon. However, the author's own opinion is that there is a 60% probability the LW was a direct source for The Book of Mormon. Do I have it right? No. The author is proposing that if the LW was a direct source, there is an 80% chance that of the 5000 books surveyed the LW would match the Book of Mormon the best. The author also proposes that there is a 50% potential for a false positive. That is even if the LW (or any other book for that matter) topped the list, there is still a 50% chance it was not a direct source for the Book of Mormon. If we started with 50/50 odds (before conducting the word match test), after applying the new fact of the match between the LW and the BOM, we end up with a 60% probability the LW was a direct source. 2
cdowis Posted November 7, 2013 Posted November 7, 2013 seeking, "source" Correlation does not prove causality. Please show me where I am mistaken.
SeekingUnderstanding Posted November 7, 2013 Posted November 7, 2013 (edited) seeking, "source" Correlation does not prove causality. Please show me where I am mistaken.Your mistake is that nowhere in my very short post did I claim correlation = causation. I (attempted to) explain the author's math in response to a request from Peppermint Patty. In addition, the article linked also has no claim that correlation = causation. The author of the article used Bayes theorem and some assumptions to calculate the increased probability of the LW being a direct source for the Book of Mormon given the fact that it was the best match for the BOM in a recent study. His answer was not very much. Edited November 7, 2013 by SeekingUnderstanding 1
cdowis Posted November 7, 2013 Posted November 7, 2013 Your mistake is that nowhere in my very short post did I claim correlation = causation. I (attempted to) explain the author's math in response to a request from Peppermint Patty. In addition, the article linked also has no claim that correlation = causation. The author of the article used Bayes theorem and some assumptions to calculate the increased probability of the LW being a direct source for the Book of Mormon given the fact that it was the best match for the BOM in a recent study. His answer was not very much. My point is that the author's underlying assumption behind the analysis is flawed. "Source" specifically implies causality in this case. The analysis will only demonstrate similarity. Sorry, but I thought that was obvious.
SeekingUnderstanding Posted November 7, 2013 Posted November 7, 2013 My point is that the author's underlying assumption behind the analysis is flawed. "Source" specifically implies causality in this case. The analysis will only demonstrate similarity. Sorry, but I thought that was obvious.cdowis, Do you have a basic understanding of Bayes theorem? If not then further conversation with you on this subject is impossible.
Bruce Schaalje Posted November 7, 2013 Posted November 7, 2013 No. The author is proposing that if the LW was a direct source, there is an 80% chance that of the 5000 books surveyed the LW would match the Book of Mormon the best. The author also proposes that there is a 50% potential for a false positive. That is even if the LW (or any other book for that matter) topped the list, there is still a 50% chance it was not a direct source for the Book of Mormon. If we started with 50/50 odds (before conducting the word match test), after applying the new fact of the match between the LW and the BOM, we end up with a 60% probability the LW was a direct source. Seeking and Patty, Sorry to give you a headache, but thanks for grappling with the post. What I'm saying is that if we assumed: 1. 50/50 initial odds (just to be generous to the critics) that LW influenced the Book of Mormon,2. 80/20 odds of a true positive (that the Johnson brothers' technique would identify LW as a major influence on the Book of Mormon if it truly was), and3. 50/50 odds of a false positive (that the Johnson brothers' technique would identify LW as a major influence on the Book of Mormon if it truly was not, then, the observation that LW is most similar (using the Johnson similarity measure) to the Book of Mormon out of 5000 random books, increases the odds that LW influenced the Book of Mormon to about 60/40. I deliberately chose values for the odds in 1, 2, and 3 above to skew things towards the critic's position so that they could not say I simply chose values for the odds that would give the answer I wanted. The 50/50 to 60/40 change is not dramatic even if we choose these worst-case (from our point of view) values for the odds. So no slam dunk, no destruction of Mormonism based on these results. Bruce 3
cdowis Posted November 7, 2013 Posted November 7, 2013 (edited) cdowis, Do you have a basic understanding of Bayes theorem? If not then further conversation with you on this subject is impossible. Yes I do, and you are probably (75%) correct. Can I assume that you understand that Baysian analysis is merely a mathmatical tool. If your logic is flawed, it will not save you from faulty analysis. It will only make you look impressive. Edited November 7, 2013 by cdowis
Peppermint Patty Posted November 7, 2013 Author Posted November 7, 2013 Mr. Schaalje, Thank you so much for explaining this to me. As you can probably tell, math and statistic are not my strong points.
SeekingUnderstanding Posted November 7, 2013 Posted November 7, 2013 Yes I do, and you are probably (75%) correct.Then please explain where exactly in the referenced article that the author conflated correlation with causation. Bayes theorem allows us to assign probabilities to correlation and causation. That's precisely what the author does. It sounds like you'd assign a much higher false positive probability than the author, but since the author has stated he was intentionally being generous to the critics, I really don't understand your point.
cdowis Posted November 7, 2013 Posted November 7, 2013 (edited) Do you want to quibble over whether "source" is just another flavor of "cause". Not interested in that discussion. "We have found the source of the BOM text" is a mental trap that we need to avoid, regardless of the analysis. Edited November 7, 2013 by cdowis
SeekingUnderstanding Posted November 7, 2013 Posted November 7, 2013 (edited) Do you want to quibble over whether "source" is just another flavor of "cause". Not interested in that discussion. "We have found the source of the BOM text" is a mental trap that we need to avoid, regardless of the analysis.The entire point of the article is that even if we make generous assumptions in favor of the critics, the fact that the BOM and the LW are correlated doesn't tell us very much about whether the LW is a source for the BOM. That is, the author's conclusion is that correlation does not equal causation. I give up. Edited November 7, 2013 by SeekingUnderstanding
wenglund Posted November 7, 2013 Posted November 7, 2013 Seeking and Patty, Sorry to give you a headache, but thanks for grappling with the post. What I'm saying is that if we assumed: 1. 50/50 initial odds (just to be generous to the critics) that LW influenced the Book of Mormon,2. 80/20 odds of a true positive (that the Johnson brothers' technique would identify LW as a major influence on the Book of Mormon if it truly was), and3. 50/50 odds of a false positive (that the Johnson brothers' technique would identify LW as a major influence on the Book of Mormon if it truly was not, then, the observation that LW is most similar (using the Johnson similarity measure) to the Book of Mormon out of 5000 random books, increases the odds that LW influenced the Book of Mormon to about 60/40. I deliberately chose values for the odds in 1, 2, and 3 above to skew things towards the critic's position so that they could not say I simply chose values for the odds that would give the answer I wanted. The 50/50 to 60/40 change is not dramatic even if we choose these worst-case (from our point of view) values for the odds. So no slam dunk, no destruction of Mormonism based on these results. Bruce Just curious, using the same formula. what would be the change in odds for the other 5000 books? Thanks, -Wade Englund-
Freedom Posted November 7, 2013 Posted November 7, 2013 The example they gave in the article is a good one. A woman over 40 has a 1 in 10,000 chance of developing breast cancer. The mammogram is highly effective in diagnosing breast cancer, but there is a probability that it may be wrong in its conclusions. There is also the matter of what you are comparing. Does the fact that two people with the same height, hair color, number of digits, and placement of internal organs prove that they are siblings? If the New testament and Harry Potter novels both quote from the savior prove that the Harry Potter novels are scripture? So, are the factors being compared relevant to the conclusions being drawn. did Joseph Smith get his grammatical constructs from 'The late War' or did both authors get their grammatical constructs from the same source, the english language of the time. 2
poulsenll Posted November 7, 2013 Posted November 7, 2013 Little's little book of statistics teaches basic statistics by selecting two sets of data and uses them to teach basic statistical methods including correlation coefficients. After showing a 99 percent correlation between the 10 year production of tobacco and I think it was corn, he then suggests that one might propose that corn husks were being used to adulterate tobacco. He goes on to explain that this is a common misuse of statistics and is usually the result of both sets of data having a relationship with a third set of data. In this case it was the gross national product over the same 10 year period. Freedom is right on the mark. These efforts to show a statistical relationship for word usage and the BofM as proof that Joseph Smith used these sources as a basis for writing the BofM are doomed to failure by the simple fact that they all wrote in the language of the time and Joseph in all likelihood translated the BofM using the language of the time. Larry P 3
BCSpace Posted November 7, 2013 Posted November 7, 2013 I think there is no statistical relationship at all in the authorship sense because the context is being overlooked. What you have is someone trying to write an historical work in a way society recognizes as scripture and you have JS translating through the gift and power of God into something that will be recognized by society as scripture, and yes, an historical work as well. Hence there will be many similarities, but no indication whatsoever that the one copied from the other. The only thing one has in common is society's common recognition of a scriptural mode of writing. 1
Kenngo1969 Posted November 7, 2013 Posted November 7, 2013 Mr. Schaalje, Thank you so much for explaining this to me. As you can probably tell, math and statistic are not my strong points.Join the vast majority of the rest of the world!
Calm Posted November 7, 2013 Posted November 7, 2013 Join the vast majority of the rest of the world! Well, they used to be mine...but now I probably feel as lost reading that kind of stuff as many others not trained it in. Like all skills, if one doesn't continue to use and improve upon it, math and stats fade over time...and 35 years of not using something even if one used to live and breathe it will cause a lot of fading and falling behind. No one should ever feel bad about not understanding something because they chose to devote their lives to developing other good things, they should only feel bad if it is a lack of motivation to stir out of their comfort zone that caused them to avoid learning about something. I can see I am feeling way too serious at the moment. Perhaps I should go clean some toilets to remind myself it is better to enjoy life than to look on it as work.
Recommended Posts