[Synoptic-L] Fwd: The HBB analysis and the GH.
- In a message dated 3/16/2002 8:52:57 PM Eastern Standard Time,
<< Hello Leonard,
It seems this E-mail I.D. is currently blocked by the university host.
If you are not blocked as well, could you forward it to the list?
> Just those, I guess. Can you move now to a specific question, namely,
> how the results that you have reached so far seem to you to exclude theO.K. I'll move on. We can go back to the data if it comes up in the
> likelihood that Mark is a late gospel based on Matt and Lk? Thanks.
> Leonard Maluf
discussion of the interpretations.
A far as the analysis goes, I'll try to just mention this briefly, but I
want to be clear on what we can say with some degree of mathematical
confidence, before we get to interpretation.
The basic idea is to try to predict how often a specific Greek word will
occur in a specific category, and the compare it to the actual count. An
example may help to think about the idea behind it. Suppose we want to guess
how many times the word "the" occurs on page 234 of some book. We know that
there are 200,000 words in he book, 200 words on the page, and that the word
"the" occurs 4,000 times in the book. Without any other information, we
would guess there are 4 "the"s on page 234. ( 4,000 * 200 / 200,000 ).
That will probably not be a bad estimate if the book is fairly homogenous.
We can then find the actual number and compare it to our guess, and
determine how likely the actual result was, given our estimate. We do this
as a way of seeing how good our estimate was.
In the actual study the "book" is all categories. The "page" is an HBB
category, and the words are still words.
The final step in to see if additional information allows us to make better
estimates. We are then interested in whether or not these improved estimates
are enough better to be considered statistically significant.
Let's say we are trying to see how may times ABBRAM occurs in 202.
We find that it occurs 18 times in all categories, that there are 1621
words in category 202 and 25843 in all categories. We would expect about
1.12 "ABBRAM"s based only on that information. (1621*18/25843)
We find that there are 3.
The question is would more information help us make a better guess.
200 has 3 occurrences. If we assume 200 is more similar to 202 than to the
total of all categories that knowledge of the frequency in 200 would help us
make a better guess about the frequency in 202.
We repeat this for all 800+ words.
That gives us our certain statement. If we find a significant relation
between to categories, that means that if we know the frequency in category
"A" and assume that "B" is related to "A" we can improve our estimate of the
frequency in "B". If the relationship is significant, we have a high level
of confidence that this is not due to random chance.
The specific results we'll probably need to concentrate on are:
221-121 9E-16 confidence level
221-021 .0003 "
221-020 .0075 "
122-121 .0001 "
221-211 not significant
122 -112 not significant
221-200 not significant
Next we get to interpretation, so I'll start a new post.
Ph.D. Management Science candidate