## Re: [Synoptic-L] Some numerical results

Expand Messages
• Dave, What surprises me about this whole discussion is that you have been trying to make sense of a set of results which represents a small proportion of the
Message 1 of 27 , Dec 2, 2001
Dave,
trying to make sense of a set of results which represents a small
proportion of the published data, and apparently (unless I've missed
something) without any attempt to check for overall consistency.
Before attempting to explain individual positive and negative
correlations, I would like to see a consistency check along the
following lines.
Divide the words up into three or four groups by first letter, e.g.
A-D, E-K, L-P, R-W, then carry out the analysis on each group, then
compare the correlations. Any disagreements among the correlations might
suggest there is something wrong with the method or with our implicit
generalization from a subset of the data. If they were all in agreement,
then we really would feel obliged to try hard to explain them.
If the whole alphabet is too much to tackle, carry out this procedure
with a subset of the alphabet divided into groups.

Ron Price

Weston-on-Trent, Derby, UK

e-mail: ron.price@...

Web site: http://homepage.virgin.net/ron.price/index.htm

Synoptic-L Homepage: http://www.bham.ac.uk/theology/synoptic-l
List Owner: Synoptic-L-Owner@...
• Hello Ron, Thanks for the suggestion. I was planning to do exactly that, once I got the next batch of data entered. I m not sure how soon that will be. Dave
Message 2 of 27 , Dec 2, 2001
Hello Ron,

Thanks for the suggestion. I was planning to do exactly that, once I got
the next
batch of data entered. I'm not sure how soon that will be.

Dave Gentile
Riverside, Illinois
M.S. Physics
PhD Management Science candidate

> Dave,
> trying to make sense of a set of results which represents a small
> proportion of the published data, and apparently (unless I've missed
> something) without any attempt to check for overall consistency.
> Before attempting to explain individual positive and negative
> correlations, I would like to see a consistency check along the
> following lines.
> Divide the words up into three or four groups by first letter, e.g.
> A-D, E-K, L-P, R-W, then carry out the analysis on each group, then
> compare the correlations. Any disagreements among the correlations might
> suggest there is something wrong with the method or with our implicit
> generalization from a subset of the data. If they were all in agreement,
> then we really would feel obliged to try hard to explain them.
> If the whole alphabet is too much to tackle, carry out this procedure
> with a subset of the alphabet divided into groups.
>
> Ron Price
>

Synoptic-L Homepage: http://www.bham.ac.uk/theology/synoptic-l
List Owner: Synoptic-L-Owner@...
• ... In this regard, it occurs to me that the 95% level of assessing significance is far too generous. 95% implies that one in twenty may be wrong, but in our
Message 3 of 27 , Dec 2, 2001
At 10:43 AM 12/2/2001 +0000, Ron Price wrote:
>trying to make sense of a set of results which represents a small
>proportion of the published data, and apparently (unless I've missed
>something) without any attempt to check for overall consistency.

In this regard, it occurs to me that the 95% level of assessing
significance is far too generous. 95% implies that one in twenty
may be wrong, but in our data set we have 19 total sections (222
through 002) for a total of 171 different pair-wise comparisons
(19*18/2). If 1 in 20 is wrong (inherent in the 95% level), then
about 9 of the "significant" correlations are really just to due
to random chance.

This is why in such analyses textbooks recommend using P < 0.05/C,
which, in this case, works out to P < 0.0003. Thus, only those
correlations at that level should be considered significant. And
any attempts to build a theory should only be concerned with
those correlations.

Stephen Carlson
--
Stephen C. Carlson mailto:scarlson@...
"Poetry speaks of aspirations, and songs chant the words." Shujing 2.35

Synoptic-L Homepage: http://www.bham.ac.uk/theology/synoptic-l
List Owner: Synoptic-L-Owner@...
• Hello again, I agree 95% is very generous. I used that in reporting Delta because there was little data. With Alpha-Delta I reported 99%. (And lost one
Message 4 of 27 , Dec 2, 2001
Hello again,

I agree 95% is very generous. I used that in reporting "Delta" because there
was little data.
With "Alpha-Delta" I reported 99%. (And lost one correlation that showed up
in the "Delta-only" test.)

I think P < 0.0003 might be a bit too restrictive at the moment. If I'm not
mistaken, that would be what we want, if we want to be 95% sure that ALL of
them are real.

Actually, I'm looking at the confidence of each one in interpreting it. The
most information is gained by reporting each one with its confidence level,
then you know which ones to look at more closely. At 99% I expect 1 or 2 to
be false, but probably only 1 since a lot are well above 99%. My candidate
for that is 012 with 210.

Most of the controversial ones were at high levels.
222 - 202 > 99.99%
200 - 202 > 99.99%
122 - 112 > 99.9%

Again, I'm glad you understand the results.

Thanks,

Dave Gentile
Riverside, Illinois
M.S. Physics
PhD Management Science candidate

-------------------------------------------------------------------------
"When you have eliminated the impossible,
whatever remains, however improbable, must be the truth."
- Sherlock Holmes,
in The Sign of Four, by Sir Arthur Conan Doyle.

"Why sometimes I've believed as many as
six impossible things before breakfast."
- The Red Queen,
in Through the Looking Glass, by Lewis Carroll

----- Original Message -----
From: "Stephen C. Carlson" <scarlson@...>
To: "Synoptic-L" <Synoptic-L@...>
Sent: Sunday, December 02, 2001 9:29 AM
Subject: Re: [Synoptic-L] Some numerical results

> At 10:43 AM 12/2/2001 +0000, Ron Price wrote:
> >trying to make sense of a set of results which represents a small
> >proportion of the published data, and apparently (unless I've missed
> >something) without any attempt to check for overall consistency.
>
> In this regard, it occurs to me that the 95% level of assessing
> significance is far too generous. 95% implies that one in twenty
> may be wrong, but in our data set we have 19 total sections (222
> through 002) for a total of 171 different pair-wise comparisons
> (19*18/2). If 1 in 20 is wrong (inherent in the 95% level), then
> about 9 of the "significant" correlations are really just to due
> to random chance.
>
> This is why in such analyses textbooks recommend using P < 0.05/C,
> which, in this case, works out to P < 0.0003. Thus, only those
> correlations at that level should be considered significant. And
> any attempts to build a theory should only be concerned with
> those correlations.
>
> Stephen Carlson
> --
> Stephen C. Carlson mailto:scarlson@...
> "Poetry speaks of aspirations, and songs chant the words." Shujing 2.35
>
>
> Synoptic-L Homepage: http://www.bham.ac.uk/theology/synoptic-l
> List Owner: Synoptic-L-Owner@...

Synoptic-L Homepage: http://www.bham.ac.uk/theology/synoptic-l
List Owner: Synoptic-L-Owner@...
• Brian Wilson wrote -- ... Stephen Carlson replied -- ... Stephen, But the aim of the game is to discover the relationship between the synoptic gospels, not the
Message 5 of 27 , Dec 2, 2001
Brian Wilson wrote --
>
>The more I look at the HHB categorization the more I wonder whether
>analysing the stats reveals the decisions of the editors of the
>Concordance rather than the relationship between the synoptic gospels.
>
Stephen Carlson replied --
>
>Of course, it reveals the decisions of the editors about the
>relationship between the synoptic gospels.
>
Stephen,
But the aim of the game is to discover the relationship between
the synoptic gospels, not the editors' decisions.

I would suggest that a computer could be programmed to categorize every
word of the synoptic gospels into 222, 221, 220, and so on, without
assuming the Two Document Hypothesis (as do the editors of the HHB
Concordance), or any other supposed documentary relationship between the
synoptic gospels. The program might be designed first to search for
strings of words in one synoptic gospel containing at least a stated
number of words (or word-roots) in the same order as strings of words of
the same size in another synoptic gospel. This could be extended to find
similarities of wording in the same order in all three synoptic gospels
within stated parameters. The program would establish triple parallel,
double parallel, and Sondergut pieces of material. The size of each unit
of triple, double, or special, could be, say, at least 8 word-roots in
the same order, though of course a triple, double or special could be
longer than this minimum. Whatever minimum size is laid down would be
arbitrary, but at least the procedure using it would be independent of
any assumed documentary hypothesis. On this basis, the categorization
into 222, 221, 220 and so on could be carried out without making any
assumptions of the documentary relationship between the synoptic
gospels. 222 would represent a word-root present in each synoptic gospel
in a computer-generated triple parallel, and so on. This would give
"clean data", to use Dave Gentile's terminology. It would produce
results that would not be dependent on the decisions of the editors of
the HHB Concordance. I think that ideally this is where the analysis
might have started.
>
>The main question, however, is whether and to what extent the editors'
>judgment is biased to a particular solution to the synoptic problem.
>
I agree. That is one question I have been asking. The other question I
have been raising is to what extent the editors' have applied their
criteria inconsistently.

What is really needed is computer-produced data that is not dependent on
any synoptic documentary hypothesis and that applies criteria
consistently. The HHB Concordance does not provide such data.

Best wishes,
BRIAN WILSON

>HOMEPAGE http://www.twonh.demon.co.uk/

Rev B.E.Wilson,10 York Close,Godmanchester,Huntingdon,Cambs,PE29 2EB,UK
> "What can be said at all can be said clearly; and whereof one cannot
> speak thereof one must be silent." Ludwig Wittgenstein, "Tractatus".
_

Synoptic-L Homepage: http://www.bham.ac.uk/theology/synoptic-l
List Owner: Synoptic-L-Owner@...
• ... If you (or anyone else) could produce such a computer program, I d be very impressed. However, one cannot even identify a parallel without supposing some
Message 6 of 27 , Dec 2, 2001
At 06:07 PM 12/2/2001 +0000, Brian E. Wilson wrote:
>I would suggest that a computer could be programmed to categorize every
>word of the synoptic gospels into 222, 221, 220, and so on, without
>assuming the Two Document Hypothesis (as do the editors of the HHB
>Concordance), or any other supposed documentary relationship between the
>synoptic gospels.

If you (or anyone else) could produce such a computer program,
I'd be very impressed. However, one cannot even identify a parallel
without supposing some documentary relationship between the synoptic
gospels. For example, the decision to identify Mark's healing of
the paralytic as a parallel to Matthew's healing of the paralytic --
but not to Matthew's genealogy -- supposes that there is a documentary
relationship with the former but not the latter. Of course, most
decisions about parallelization will not be controversial, but
they will be for not only for the so-called Mark-Q overlap texts and
for doublets in one gospel where there is only one instance in the
others (which of the two in the doublet is "the" parallel?). Even
in noncontroversial parallels, there is discretion in deciding which
of two KAIs in Mark correspond to the one KAI in Matthew. Frankly,
I doubt it can be done objectively, because the theories and hypotheses
we hold influence our perception of the data. This extends to the
programmer of the computer as well.

>>The main question, however, is whether and to what extent the editors'
>>judgment is biased to a particular solution to the synoptic problem.
>>
>I agree. That is one question I have been asking. The other question I
>have been raising is to what extent the editors' have applied their
>criteria inconsistently.

The editors admit their bias on the Mark-Q overlaps, but does that
bias extend to other portions? You've pointed out some inconsistencies,
but unless one can show they somehow favor the 2ST, it may be better
to attribute the inconsistencies to noise rather than bias.

>What is really needed is computer-produced data that is not dependent on
>any synoptic documentary hypothesis and that applies criteria
>consistently.

I'm not holding my breath. Perhaps a better approach is to ask
the proponents of the various solutions to produce their own
"partisan" synopses and concordances that sets forth the data
in such a manner that renders their solution in the most favorable
light. Then we can compare them to see how well they handle the
data. Maybe Dungan was right after all that there is no neutral
synopsis.

Stephen Carlson
--
Stephen C. Carlson mailto:scarlson@...
"Poetry speaks of aspirations, and songs chant the words." Shujing 2.35

Synoptic-L Homepage: http://www.bham.ac.uk/theology/synoptic-l
List Owner: Synoptic-L-Owner@...
• Brian Wilson wrote -- ... Stephen Carlson replied -- ... Stephen, I do not understand your argument here. The occurrence of parallels between Mark s healing of
Message 7 of 27 , Dec 3, 2001
Brian Wilson wrote --
>
>I would suggest that a computer could be programmed to categorize every
>word of the synoptic gospels into 222, 221, 220, and so on, without
>assuming the Two Document Hypothesis (as do the editors of the HHB
>Concordance), or any other supposed documentary relationship between
>the synoptic gospels. The program might be designed first to search for
>strings of words in one synoptic gospel containing at least a stated
>number of words (or word-roots) in the same order as strings of words
>of the same size in another synoptic gospel. This could be extended to
>find similarities of wording in the same order in all three synoptic
>gospels within stated parameters. The program would establish triple
>parallel, double parallel, and Sondergut pieces of material. The size
>of each unit of triple, double, or special, could be, say, at least 8
>word-roots in the same order, though of course a triple, double or
>special could be longer than this minimum. Whatever minimum size is
>laid down would be arbitrary, but at least the procedure using it would
>be independent of any assumed documentary hypothesis. On this basis,
>the categorization into 222, 221, 220 and so on could be carried out
>without making any assumptions of the documentary relationship between
>the synoptic gospels. 222 would represent a word-root present in each
>synoptic gospel in a computer-generated triple parallel, and so on.
>
Stephen Carlson replied --
>
>If you (or anyone else) could produce such a computer program,
>I'd be very impressed. However, one cannot even identify a parallel
>without supposing some documentary relationship between the synoptic
>gospels. For example, the decision to identify Mark's healing of
>the paralytic as a parallel to Matthew's healing of the paralytic --
>but not to Matthew's genealogy -- supposes that there is a documentary
>relationship with the former but not the latter.
>
Stephen,
I do not understand your argument here. The occurrence of
parallels between Mark's healing of the paralytic and Matthew's healing
of the paralytic but not between Matthew's healing of the paralytic and
his Genealogy, is precisely the sort of result such a computer program
would produce without assuming a documentary hypothesis such as the 2DH,
FH or GH. There are many instances of strings of, say, no more than 20
words in the Markan healing that contain at least 8 word-roots the same,
and in the same order, in a string of no more than 20 words in the
Matthean healing. On the other hand, there is no string of no more than
20 words in the Matthaean healing that contains at least 8 word-roots
the same and in the same order in a corresponding string of no more than
20 words in the Genealogy in Matthew. The criterion would establish that
the two healing stories include parallels in the sense of having many
word roots the same and in the same order within strings of specified
lengths, but that the Matthaean healing and the Genealogy in Matthew do
not include parallels in this sense. It would be easy to set up a
computer program to show this.

Indeed, we could carry out such a check by hand, without using a
computer program to spot parallels. I have just done this. There are no
pairs of strings that meet the criteria above between the healing of the
paralytic in Mt and the Genealogy in Mt. However, the following string
of 20 words is found in the Matthaean healing of the paralytic --

KAI IDWN O IHSOUJ THN PISTIN AUTWN EIPEN TW PARALUTIKW QARSEI TEKNON
AFIENTAI SOU AI AMARTIAI KAI IDOU TINEJ TWN

And it has 17 word roots the same and in the same order as the string of
words in the Markan healing of the paralytic --

KAI IDWN O IHSOUJ THN PISTIN AUTWN LEGEI TW PARALUTIKW TEKNON AFI/ENTAI
SOU AI AMARTIAI HSAN DE TINEJ TWN

Similarly, in the same two passages, the following are pairs of strings,
one string from each passage, of no more than 20 words each such that in
each pair there are at least 8 word roots the same and in the same order
as in a string of no more than 20 words the same and in the same order
in the other member of the pair --

EN TAIJ KARDIAIJ UMWN TI GAR ESTIN EUKOPWTERON EIPEIN AFIENTAI SOU AI
AMARTIAI H EIPEIN EGEIRE KAI

EN TAIJ KARDIAIJ UMWN TI ESTIN EUKOPWTERON EIPEI=N TW PARALUTIKW
AFI/ENTAI SOU AI AMARTIAI H EIPEIN EGEIRE KAI

===

PERIPATEI INA DE EIDHTE OTI ECOUSIAN EXEI O UIOJ TOU ANQRWPOU EPI THJ
GHJ AFIENAI AMARTIAJ TOTE LEGEI TW PARALUTIKW

PERIPATEI INA DE EIDHTE OTI ECOUSIAN EXEI O UIOJ TOU ANQRWPOU AFIENAI
AMARTIAJ EPI THJ GHJ LEGEI TW PARALUTIKW

===

EGERQEIJ ARON SOU THN KLINHN KAI UPAGE EIJ TON OIKON SOU KAI EGERQEIJ

EGEIRE ARON TON KRABATTON SOU KAI UPAGE EIJ TON OIKON SOU KAI HGERQH

===

Each of these strings is a pair of parallels between Mt and Mk, and many
words within these strings (after doing the same sort of comparison with
Lk) could be categorized 220, 221, and so on. These are by no means all
the parallels of this type that can be listed. These parallels, and
others, between Mt and Mk could have been spotted by a computer
systematically comparing every string of 20 words in Mt with every
string of 20 words in Mk, and looking for at least 8 word-roots the same
and in the same order in each pair of strings compared.

The point is that it is totally unnecessary to assume a synoptic
documentary hypothesis such as the 2DH, FH or GH, to establish that two
pieces of material, one in Mt and the other in Mk, are parallels, or
that two pieces of material in Mt are not parallels.

You also wrote --
>
>Maybe Dungan was right after all that there is no neutral synopsis.
>
I am sure he was right. The HHB Concordance, however, is not a synopsis,
and the question is whether such a Concordance can be produced without
assuming a documentary hypothesis such as the 2DH, FH or GH. The answer
is that it can.

In the same way, a colour-coded text showing similarities of wording
between the synoptic gospels, like W. R. Farmer's SYNOPTICON, is not a
synopsis. Such a colour-coded text of the synoptic gospels can be
constructed without assuming any particular solution to the synoptic
problem, and without using any synopsis.

Best wishes,
BRIAN WILSON

>HOMEPAGE http://www.twonh.demon.co.uk/

Rev B.E.Wilson,10 York Close,Godmanchester,Huntingdon,Cambs,PE29 2EB,UK
> "What can be said at all can be said clearly; and whereof one cannot
> speak thereof one must be silent." Ludwig Wittgenstein, "Tractatus".
_

Synoptic-L Homepage: http://www.bham.ac.uk/theology/synoptic-l
List Owner: Synoptic-L-Owner@...
• ... I see some different biases in an automatic process : - Brian pointed out that a gospel may be selected, and the priority of this gospel (typicallly :
Message 8 of 27 , Dec 4, 2001
"Stephen C. Carlson" wrote :
>
> Brian E. Wilson wrote:
> >I would suggest that a computer could be programmed to categorize every
> >word of the synoptic gospels into 222, 221, 220, and so on, without
> >assuming the Two Document Hypothesis (as do the editors of the HHB
> >Concordance), or any other supposed documentary relationship between the
> >synoptic gospels.
>
> If you (or anyone else) could produce such a computer program,
> I'd be very impressed. However, one cannot even identify a parallel
> without supposing some documentary relationship between the synoptic
> gospels. [...] Even
> in noncontroversial parallels, there is discretion in deciding which
> of two KAIs in Mark correspond to the one KAI in Matthew. Frankly,
> I doubt it can be done objectively, because the theories and hypotheses
> we hold influence our perception of the data. This extends to the
> programmer of the computer as well.

I see some different biases in an automatic process :
- Brian pointed out that a gospel may be selected, and the priority
of this gospel (typicallly : Mark) is assumed, biasing the result.
In fact, in the treatment, the input gospels should be processed
with equity.
- Even if synoptic gospels are processed in equity, some a priori
may influence the result. I mean particularly to the pattern
of redaction process that will be seen as most probable : a
first small document that is step by step increased, or a large
first document that as been cut toward our gospel. the balance
between deletion and completion in synoptic process is a global
a priori in synoptic study.
- Then, an automatic process would hardly integrate all the other
information, particularly the links between Luke and John.

> >What is really needed is computer-produced data that is not dependent on
> >any synoptic documentary hypothesis and that applies criteria
> >consistently.
>
> I'm not holding my breath. Perhaps a better approach is to ask
> the proponents of the various solutions to produce their own
> "partisan" synopses and concordances that sets forth the data
> in such a manner that renders their solution in the most favorable
> light. Then we can compare them to see how well they handle the
> data. Maybe Dungan was right after all that there is no neutral
> synopsis.

Are there any referenced biases in Boismard's Synopse ?
Do you know about a comparison of different existing synopses ?

a+
manu

Synoptic-L Homepage: http://www.bham.ac.uk/theology/synoptic-l
List Owner: Synoptic-L-Owner@...
• ... Dave, did you actually mean: why does 221 look like both 121 and 211, but 122 look only like 121 and not 112 , or have I actually got the wrong data? In
Message 9 of 27 , Dec 19, 2001
Way back on Dec 1 Dave Gentile wrote:

> But perhaps the real question is, why does
> 122 look like both 121 and 112, but 221 look only like 121 and not 211?

Dave, did you actually mean: "why does 221 look like both 121 and 211, but
122 look only like 121 and not 112", or have I actually got the wrong data?
In any case, have you decided why?

Dave Inglis
david@...
3538 O'Connor Drive
Lafayette, CA, USA

Synoptic-L Homepage: http://www.bham.ac.uk/theology/synoptic-l
List Owner: Synoptic-L-Owner@...
• Hi Dave, The case reversed itself with more data. It was 121 - 122 now its 121 - 221 . These appear to be redactor effects / chance , they vanish in the macro
Message 10 of 27 , Dec 20, 2001
Hi Dave,

The case reversed itself with more data. It was 121 - 122 now its 121 - 221
. These appear to be redactor effects / chance , they vanish in the macro
categories.

Dave Gentile
Riverside, Illinois
M.S. Physics
Ph.D. Management Science candidate

----- Original Message -----
From: "David Inglis" <david@...>
To: <Synoptic-L@...>
Sent: Thursday, December 20, 2001 12:54 AM
Subject: Re: [Synoptic-L] Some numerical results

> Way back on Dec 1 Dave Gentile wrote:
>
> > But perhaps the real question is, why does
> > 122 look like both 121 and 112, but 221 look only like 121 and not 211?
>
> Dave, did you actually mean: "why does 221 look like both 121 and 211, but
> 122 look only like 121 and not 112", or have I actually got the wrong
data?
> In any case, have you decided why?
>
> Dave Inglis

Synoptic-L Homepage: http://www.bham.ac.uk/theology/synoptic-l
List Owner: Synoptic-L-Owner@...
Your message has been successfully submitted and would be delivered to recipients shortly.