Quantitative & Qualitative Modeling using traditional stats & methods

[DTSDTO]

<<<<<<For those of us who believe in the PCT model "Behavior" can only be
explained by a full understanding of the interactions of the control loops
used to control the variables we need to control, so we can maintain our
reference levels at an acceptable level. In turn, control loops have
functions that accomplish these tasks. Given this basic understanding, if I
am intereted in understanding why behavior takes place or what constitutes
behavior I need to understand how a single control loop works and I would
need to model multiple loops and levels to understand how "interactions"
take place between people. We know that we can and do control many hundreds
( thousands? ) of variables simultaneously. Some we are aware of, others we
are not. The point here is that if we wanted to model an interaction between
two individuals with the current quantifiable techniques it would be a
daunting task for all except the most basic interactions for the most
sophisticated modelers >>>>>

When I am counseling someone using MOL - If you want to TEST for some CV in a
qualitative or quantitative way, look for the expression of emotional
distress or contentment or happiness the person expresses.

You will find the emotional expression is a difference of
"THEIR INTENTIONS/EXPECTIONS" and "WHAT HAPPENED" = THE EMOTION
"WHAT THEY WANT" and "WHAT THEY ARE GETTING" = THE EMOTION
"WHERE THEY ARE" and WHERE THE WANT TO BE" = THE EMOTION

("perception" minus "reference" = "error")

These are emotional states based on what the person is perceiving about their
past present and future. Depending on the exact CV determines whether or not
it is Quantitative or Qualitative. Either what it is testable and easy
enough to create an empirical study using real numbers. Don't forget to
keep in mind if in your methodology you are casting nets or testing specimens.

I don't see the need to model the 100's or 1000's closed loops -- to have a
successful experiment. Just the ones that the person expresses distress
over. I know this is also subjective in the person's relative expression of
error. But it can be observed and measured.

My assumption in counseling is, "ANXIETY AND DEPRESSION ARE GOOD FOR YOU"
without them you would not KNOW something was wrong, nor would you have the
motivation to change.

We all have various degrees of "acceptable error" in each level of the
Hierarchy of perceptions. (This would be a great experiment -- What is the
NORMAL range of acceptable error per LEVEL in the Hierarchy of perceptions.)
You could use all kinds of traditional statistical measures. This would
definitely be casting nets.

As for a test for the individual - you could test for their personal
integration factors.
An integration factor is the rate of change in behavioral output for
perceptual error sensed, by the person performing a task or trying to reach a
goal. This factor helps determine the ease or difficulty with which
cooperation is achieved.

What I find interesting is the "integration factors" with in each level of
the Hierarchy vary.

Another hypothesis-Would people with different integration factors cooperate
at different speeds and ease.

Giving instructions vs. receiving instructions

High I.F. paired with High I.F.
High I.F. paired with Low I.F.
Low I.F. paired with High I.F.
Low I.F. paired with Low I.F.

OR a third party giving instruction to a Dyad of

High I.F. paired with High I.F.
High I.F. paired with Low I.F.
Low I.F. paired with Low I.F.

Define cooperation as: The point when actions are taken by two or more
people, to control mutually-desired perceptions of environmental variables.
To collaborate and achieve mutually-desired perceptions, each person adopts a
subordinate goal of the desired superordinate result. Each person
establishes his or her own reference or idea of how to perform the task and
then begins to carry out the subordinate strategy to achieve the perceived
superordinate goal.

I can see many ways to empirically test for the above examples using
traditional stats and methods - the BIG differences between Traditional
Psychology and PCT will be seen in the Results, Discussion and Conclusion
Sections of research papers.

Mark Lazare

[Marc_Abrams2]

from [ Marc Abrams (990531.1832) ]

A quick note: The organization and explanation of my SD MOL sessions are
taking a bit longer then I anticipated. They are forthcoming

from Mark Lazare sometime on 5/31, No heading on the post.

When I am counseling someone using MOL - If you want to TEST for some CV

in a

qualitative or quantitative way, look for the expression of emotional
distress or contentment or happiness the person expresses.

Mark, my interest is not in a theraputic ( counseling ) vein. The test for a
CV is always qualitative. I am not sure which of my posts you are responding
to, or if in fact you are responding to any of them, with the exception of
the last quote of my last post used in this post by you.

Have you done a SD MOL? From this post it seems like you haven't. You seem
to speak more in theoretic terms then in experiential terms about it.

I am not really interested in trying to guess what someone else might be
controlling for at any one point or interval in time, Chances are there are
many such CV's and I am not quite sure what knowing about any one of them at
any point in time would do for me, considering how quickly they change and
how complex they are.

In a theraputic setting ( counseling ) you ( the facilitator ) really have
no clue as to _what_ is happening internally to that individual. It's the
individual that understands the significance of the connections., and it's
the individual that _must_ "drive the truck". You only have _very_ small
glimpses into what is taking place. As a therapist, that's not a bad thing.
The client feels better ( hopefully ) and your main job as the facilitator
is to give a gentle helping hand while you _stay out of the way_.

Facilitating an MOL session and _experiencing_ one are not the same. The
research I have in mind is _not_ for the sole purpose of identifying CV's.
though you will ) But in trying to understand how the hierarchy works and
is organized.

You will find the emotional expression is a difference of
"THEIR INTENTIONS/EXPECTIONS" and "WHAT HAPPENED" = THE EMOTION
"WHAT THEY WANT" and "WHAT THEY ARE GETTING" = THE EMOTION
"WHERE THEY ARE" and WHERE THE WANT TO BE" = THE EMOTION

Sorry Mark. You lost me here. Aren't these all examples of Comparisons
between Perceptions and Reference levels? What part does "emotion" play with
these things?

("perception" minus "reference" = "error")

These are emotional states based on what the person is perceiving about

their

past present and future. Depending on the exact CV determines whether or

not

it is Quantitative or Qualitative.

Sounds interesting, but it doesn't sound like PCT. If you do a SDMOL I think
you will find that there is no such thing as an "exact CV". We control many
variables simultaneously, and the "importance", "priority" and "components"
of these CV's can and do change very rapidly.

What makes something Quantitative is the ability to _measure_ it.
preferably with numbers that have certain properties ) What does past,
present, and future emotional states have to do with qualitative or
quantitative CV's.?. On second thought, what does quantitative or
qualitative have to do with CV's

Either what it is testable and easy enough to create an empirical study

using real numbers. >Don't forget to keep in mind if in your methodology
you are casting nets or testing specimens.

I am casting nets by testing specimens.

I don't see the need to model the 100's or 1000's closed loops --

Neither do I. I simply mentioned that at any point in time there could be
thousands of active CV;s

to have a successful experiment. Just the ones that the person expresses

distress

over.

What would constitute a "successful experiment" ?

How would you know you had the "right" CV's?

I know this is also subjective in the person's relative expression of
error. But it can be observed and measured.

How? What are you measuring?

My assumption in counseling is, "ANXIETY AND DEPRESSION ARE GOOD FOR YOU"
without them you would not KNOW something was wrong, nor would you have

the

motivation to change.

An interesting position I disagree with strongly with but is irrelevant to
the SDMOL. The SDMOL is not about therapy. It's about trying to have a
better understanding of how the hierarchy works.
I think some interesting things might come out of this, that might help the
theraputic MOL method. But it is not my main thrust. Maybe you or Dick, or
Dave will find some ways of utilizing it for those purposes

We all have various degrees of "acceptable error" in each level of the
Hierarchy of perceptions.

Does "acceptable error" reside in the "level" or in the loops of each level?
Or possibly both. The model right now religates error to each control loop
only. I have no reason to believe it to be otherwise, but research could
shed a bit more light on this.

(This would be a great experiment -- What is the

NORMAL range of acceptable error per LEVEL in the Hierarchy of

perceptions.)

You could use all kinds of traditional statistical measures. This would
definitely be casting nets.

Mark, you speak of the levels in the hierarchy as if they are well defined
and well understood. They aren't. You speak of defining "ranges" of normalcy
by methods ( stats ) we argue against ( at least for the individual ) Who
gets to decide what "normal" is? I want in on that committee
What does "acceptable error" mean?

As for a test for the individual - you could test for their personal

integration factors.

An integration factor is the rate of change in behavioral output for
perceptual error sensed, by the person performing a task or trying to

reach a

goal.

How do you connect a _specific_ output with a _specific_ error? How do you
come up with the "integration factor"?

This factor helps determine the ease or difficulty with which cooperation

is achieved.

Cooperation between what? Two individuals, two internal control systems?,
etc

What I find interesting is the "integration factors" with in each level of
the Hierarchy vary.

What do your "integration factors" represent?

Another hypothesis-Would people with different integration factors

cooperate

at different speeds and ease.

Giving instructions vs. receiving instructions

High I.F. paired with High I.F.
High I.F. paired with Low I.F.
Low I.F. paired with High I.F.
Low I.F. paired with Low I.F.

OR a third party giving instruction to a Dyad of

High I.F. paired with High I.F.
High I.F. paired with Low I.F.
Low I.F. paired with Low I.F.

This is a great example of conventional research. What does it tell us about
control?

Define cooperation as: The point when actions are taken by two or more
people, to control mutually-desired perceptions of environmental

variables.

To collaborate and achieve mutually-desired perceptions, each person

adopts a

subordinate goal of the desired superordinate result. Each person
establishes his or her own reference or idea of how to perform the task

and

then begins to carry out the subordinate strategy to achieve the perceived
superordinate goal.

This is in "PCT speak" but contributes little to our understanding of the
phenomenon you describe from a PCT perspective. for example:

What are "mutually desired perceptions"? How would we know we had one if we
perceived it.

Do the "subordinate goals" need to be similiar. If so. Who decides. in what
ways?

I can see many ways to empirically test for the above examples using
traditional stats and methods - the BIG differences between Traditional
Psychology and PCT will be seen in the Results, Discussion and Conclusion
Sections of research papers.

Sorry Mark :-), the big differences will be in that the conventional
researchers will _think_ that they have hit upon something important. The
traditional methods and stats you speak of will not give us ( PCT ) the
answers to questions such as What kinds of levels exist in the hierarchy?
What is the nature of the interaction among the levels and of the loops in
the levels?

There are _many_ questions dealing with the nature of the hierarchy ( what
kind of hierarchy exists in conventional Psych ) The nature of the loops
that comprise the hierarchy ( what kind of feedback loops do other schools
of Psych use )What _exactly_ is a CV? ( hate to be boring, but what other
schools talk of CV's ) On it goes.

Marc

[Isaac_Kurtzer1]

i.kurtzer (990531.2130)

thumbs up mark!!
There is a RTP meeting in Phoenix soon. Tim Carey, a very cool fellow and drinking
buddy, will be there and so should Dr. B.

i.

[Isaac_Kurtzer1]

i.kurtzer (990531.2130)

>From [ Marc Abrams (990531.1832) ]

>From Mark Lazare sometime on 5/31, No heading on the post.

The test for a
CV is always qualitative.

You mean in the therapeutic context right? The test for the CV is also
formalized and the stability of the CV is in principle a quantifiable measure.

> You will find the emotional expression is a difference of
> "THEIR INTENTIONS/EXPECTIONS" and "WHAT HAPPENED" = THE EMOTION
> "WHAT THEY WANT" and "WHAT THEY ARE GETTING" = THE EMOTION
> "WHERE THEY ARE" and WHERE THE WANT TO BE" = THE EMOTION

Sorry Mark. You lost me here. Aren't these all examples of Comparisons
between Perceptions and Reference levels? What part does "emotion" play with
these things?

I think Mark was trying to say that emotion is a corollary of a error signal. I
would suggest particularly that it is the corollary of a changing error signal
either further or closer to zero, for "negative" and "positive" affects,
respectively. This is neither unfamiliar to psychology in general i.e. the
James/Lange theory of emotion nor alien to PCT conceptualizations i.e. Powers
"On Emotion" LCSII.

On second thought, what does quantitative or
qualitative have to do with CV's

The difference between anecdote and modelable data.

>Either what it is testable and easy enough to create an empirical study
using real numbers. >Don't forget to keep in mind if in your methodology
you are casting nets or testing specimens.

I am casting nets by testing specimens.

Could you please explain?

Also, I am unfamiliar with SDMOL. Could you briefly reiterate what this is and
in particular how it is different from MOL?

i.

[Marc_Abrams2]

from [ Marc Abrams (990601.0211) ]

i.kurtzer (990531.2130)

> The test for a
> CV is always qualitative.

You mean in the therapeutic context right?

Yes.

The test for the CV is also formalized and the stability of the CV is in

principle a quantifiable >measure.

In Principle. But _identifying_ a CV it is always a qualitative measure.
That might be quantifiable.

> > You will find the emotional expression is a difference of
> > "THEIR INTENTIONS/EXPECTIONS" and "WHAT HAPPENED" = THE EMOTION
> > "WHAT THEY WANT" and "WHAT THEY ARE GETTING" = THE EMOTION
> > "WHERE THEY ARE" and WHERE THE WANT TO BE" = THE EMOTION
>
> Sorry Mark. You lost me here. Aren't these all examples of Comparisons
> between Perceptions and Reference levels? What part does "emotion" play

with

> these things?
>

I think Mark was trying to say that emotion is a corollary of a error

signal.

Always?

I would suggest particularly that it is the corollary of a changing error

signal

either further or closer to zero, for "negative" and "positive" affects,
respectively. This is neither unfamiliar to psychology in general i.e.

the

James/Lange theory of emotion nor alien to PCT conceptualizations i.e.

Powers

"On Emotion" LCSII.

What is the effect of this on control? Is it true for all people? What kinds
of "negative" and "positive" affects are you attributing to this? Sorry for
my lack of familiarity with the work of James/Lange.

> On second thought, what does quantitative or
> qualitative have to do with CV's

The difference between anecdote and modelable data.

How is that the case? Most work in the social sciences ( using ordinal
numbers ) are quantified, but meaningless. ( at least the numbers and what
they purport to represent ) There are a number of "qualitative" modeling
techniques that are much less anecdotal then some "quantitative" techniques
I have seen. To say that something that is quantifiable is modelable and to
say that if it's not quantifiable it's not modelable is nonsense.

> >Either what it is testable and easy enough to create an empirical study
> using real numbers. >Don't forget to keep in mind if in your

methodology

> you are casting nets or testing specimens.
>
> I am casting nets by testing specimens.
>

Could you please explain?

Glad to. In Phil's book he speaks of using Statistical methods to get
proportions of the population with regard to a set of attributes of interest
in that _population_. These proportions are fine and dandy for understanding
_proportions_ ( groups of people in the population ) They are useless in
telling you exactly which indiviuals comprise each of the proportional
groups and cannot tell you exactly which 40 of the 100 in the population who
will come down with such and such a disease or some other attribute of
interest to the survey takers. Phil calls this _Casting the Net_. You get
good info on proportions of the entire population, but nada on individuals.
On the other hand Phil talks about _Action Research_ and PCT as methods of
understanding what attributes _individuals_ might have. The Self Directed
MOL uses a technique that is familiar with Action Science researchers,
called the _Left Hand Column Method_, It is used as a "reflection" tool.
When combined with the MOL as introduced by Bill to CSGnet about a year ago.
It becomes SDMOL ( Self Directed MOL ) It is a way for anyone to
self-reflect using the simple notion of "going up a level". But unlike the
Action Science version. It doesn't stop after one iteration. It continues on
and on. For as long as the participant wants to. And unlike the MOL as
practiced theraputically, the SDMOL can become quite tedious and
frustrating. When this happens to a client it usually provides the _ending_
for a session ( for good reason ) But with the SDMOL this is a
_critical_ point and needs to be dealt with in any number of ways. ( one, is
to stop )
This provides us with a way of examing our "levels" qualitatively. It also w
ill provide some _very_ interesting insights into what we are controlling,
_Guarenteed_ ) The "data" comes from each individual that participates.
This data is _not_ summarized. We try and see if there are similarities in
_when_ people go up a level. Can we actually "name" a level? Can we identify
CV's? I'd even venture to say that we don't even really know all we might
find out.
So by gathering data from individuals ( testing the individual ) we may be
able to decern larger patterns in the population ( casting the net )

Also, I am unfamiliar with SDMOL. Could you briefly reiterate what this

is and

in particular how it is different from MOL?

I will be coming out with the technique on CSGnet in a few days. The SDMOL
in my mind is a potential way to explore the upper levels in the hierarchy
and CV's and to explore how we interact with our environment. It's a top
down, qualitative approach. It could lead to a more precise quantitative
approach once we actually understand what it is we are dealing with. It is
not intended to replace, improve, change, or in anyway impact on the current
MOL useage.

One of the things that is holding me up is some late breaking software
issues. I ran across some tools that are free or shareware and would help
considerably in cutting down the amount of paper and organization currently
needed to do SDMOL. The SDMOL takes time and _everything_ must be recorded

Hope I answered your questions I.

Marc

[Bill_Powers1]

[From Bill Powers (990601.0319 MDT)]

Marc Abrams (990601.0211) ]

The test for a CV is always qualitative.

Isaac:

The test for the CV is also formalized and the stability of the CV is in

principle a quantifiable >measure.

Marc:

In Principle. But _identifying_ a CV it is always a qualitative measure.
That might be quantifiable.

I agree with both Marc and Isaac. Every variable has a qualitative aspect,
by which it is distinguished from other variables, and a quantitative
aspect, by which its state is measured. "Area" is a concept distinguished
qualitatively from other variables such as "volume" or "length." But all
three of these different qualities can be measured to produce a number
indicating how much area, how much volume, or how much length, which is the
quantitative aspect of each. A "circle" is a qualitative concept, with
qualitative aspects of radius and location of center. The radius can be
measured to yield a number, and the location to yield three numbers (its
position in space). The circleness, however, can't be measured; it is
strictly a qualitative term referring to a kind of shape but not to an
amount of something. One could quantify circleness by breaking it down to,
say, two orthogonal measures of diameter, with the requirement that
differences in the two (or more) quantitative measures of diameter be the
same within close limits.

To _identify_ a controlled quantity means to find a _kind_ of variable that
is possibly being controlled. But the Test then requires, among other
things, a quantitative measure of the _amount_ of effect that a disturbance
has on it, as opposed to the amount of effect we can predict that the
disturbance would have if there were no control. The _quantitative_ aspect
of the Test is what enables us to accept or reject the qualitative
identification of a controlled variable.

Some qualitative variables, like honesty, are hard to measure in
quantitative terms. This is because they are defined primarily by unaided
human perceptions, and their nature depends on agreements which are seldom
spelled out formally. But such formal agreements can be spelled out, and
human observers can be used to report their quantitative judgments of the
degree to which a given quality of experience is present. Psychologists
have used this approach for a long time, although they tend to use informal
identifications of variables as if everyone ought to know what words
referring to them mean. The concept indicated by the term "Inter-rater
reliability" is nevertheless useful, and it can be made more useful by
paying more attention to formally identifying variables before assuming
that different raters are rating the same variable.

Strictly speaking, there is no such thing as a qualitative test for a
controlled variable. The Test is inherently quantitative, and it's the
quantitative measure that is the basis for accepting or rejecting the
qualitative identification. However, measurement scales can be more or less
precise, and they can be logical as well as numerical. When we speak of
"qualitative PCT" I think we're talking about the precision and nature of
the measurement scales, not the difference between quantity and quality. To
identify a controlled variable, we must still assure that it must be
perceived to be controlled, that the actions of the controller demonstrably
affect it, and that disturbances affect it markedly less than they ought to
if there were no control. The judgment as to the validity of an
identification of a CV ultimately rests on what we agree to mean by
"markedly less."

Best,

Bill P.

[Isaac_Kurtzer1]

i.kurtzer (990601.1100)

>From [ Marc Abrams (990601.0211) ]

> I would suggest particularly that it is the corollary of a changing error
signal
> either further or closer to zero, for "negative" and "positive" affects,
> respectively. This is neither unfamiliar to psychology in general i.e.
the
> James/Lange theory of emotion nor alien to PCT conceptualizations i.e.
Powers
> "On Emotion" LCSII.

What is the effect of this on control? Is it true for all people? What kinds
of "negative" and "positive" affects are you attributing to this? Sorry for
my lack of familiarity with the work of James/Lange.

unknown, unknown, sadness and happiness for one, don't worry.these are all
interesting points. But this is not my field of research so I'll let someone
else do the dirty work.

> > On second thought, what does quantitative or
> > qualitative have to do with CV's

> The difference between anecdote and modelable data.

How is that the case? Most work in the social sciences ( using ordinal
numbers ) are quantified, but meaningless.

how about necessary but not sufficient?

i.

[Marc_Abrams2]

from [ Marc Abrams (990601.1404) ]

[From Bill Powers (990601.0319 MDT)]

The Test is inherently quantitative, and it's the
quantitative measure that is the basis for accepting or rejecting the
qualitative identification. However, measurement scales can be more or

less

precise, and they can be logical as well as numerical. When we speak of
"qualitative PCT" I think we're talking about the precision and nature of
the measurement scales, not the difference between quantity and quality.

To

identify a controlled variable, we must still assure that it must be
perceived to be controlled, that the actions of the controller

demonstrably

affect it, and that disturbances affect it markedly less than they ought

to

if there were no control. The judgment as to the validity of an
identification of a CV ultimately rests on what we agree to mean by
"markedly less."

Thanks for the clarity.

Marc

[Marc_Abrams2]

from [ Marc Abrams (990601.1447) ]

i.kurtzer (990601.1100)

> > > On second thought, what does quantitative or
> > > qualitative have to do with CV's
>
> > The difference between anecdote and modelable data.
>
> How is that the case? Most work in the social sciences ( using ordinal
> numbers ) are quantified, but meaningless.

how about necessary but not sufficient?

Good point. I agree.

Marc