a reason, and SS Saunders

[From Dick Robertson, 2001.0426.0705CDT, recent 2001,0417.1600]

Sam:

Yes, I used Runkels' "Casting nets in an undergrad "experimental Psych
course." As ex-psy was viewed in our department it really was an
introduction to research methods. I started with Phil's book, stressing
the difference between casting nets and testing specimens. Then I used
one of my colleague's standard text to review some of the traditional
methods, pointing out that they mainly consisted of strategies for
"casting nets" types of research. Then I came back to the argument as to
why _basic_ research, in psych as in probably all fields, requires the
testing specimens approach because you are looking for universal
principles: the basic nature of the phenomena in question. In psych
that would mean every subject would show the "same" action. I then
illustrated that approach with a few instances: Psychophysics,
Ebbinghaus' studies of memory as a classic example, and my own paper on
Testing the Self as a control system--as a crude first step toward
developing a methodology for "personality" studies, or in HPCT terms,
higher order "behavior."

Some (a minority I believe) seemed really to get some insight from the
distinction between nets and specimens. However, our school does not
generally see many students who think of themselves as interested in
basic science. Being mainly focused on whatever in Exp-Psy would
help them in grad school, most didn't spend much time pondering the
significance of the issues posed for research in psychology.

Now, to Bruce N & [Chuck Tucker (2001.0424b.07:52)]

In a message dated 4/23/2001 2:47:57 PM Eastern Daylight Time,
bnevin@CISCO.COM writes:

<< But there it is: I would benefit greatly from description and
discussion of practical examples of designing experiments or case
studies and collecting data in a variety of naturalistic settings. (And
I don't know what a quasi-experiment is. So many gaps in my education!)

One of the major complaints I have (and have so protested in many
forums) with "social science research" is the lack of any useful
description of procedures in any research. It is impossible to
replicate such research for that very reason. Since PCT uses a different
method than is customary in human research (testing a specimen) it is
extremely important that procedures are describe in great detail.
Unfortunately, this is rarely done.

So what did you guys think of my reference to my attempt to see whether
I could check out student's work in a course as an instance of control--
controlling for a chosen grade? True, I didn't give as many of the
details as I have in the past, but did you get the idea of the attempt
to analogize that investigation with the control of a cursor as in a
tracking experiment? Some of the students did (the brighter ones, I
like to believe.) There was some evidence of increasing output to
counter increasing error -- with those students who acknowledged that
they were in fact attempting to control their grade to their chosen
reference value. But, there were too few data points to attempt any
decent kind of model. There could conceivably have been 85 or more data
points (17 tests each taken 5 or more times) but that would have been
for the slowest learners. In fact the best students, those who
consciously tried to control for an A (90% or better) also tended to
supply the fewest data points. No real surprise, huh? Bill pointed out
you can't make much of a model from that. But, what do you think of the
attempt? How could you do it better? Do you think it worth trying?

Best, Dick R.

[From Bruce Nevin (010427 22:38 EDT)]

Dick Robertson, 2001.0426.0705CDT, recent 2001,0417.1600–

So what did you guys think of my reference to
my attempt to see whether

I could check out student’s work in a course as an instance of
control–

controlling for a chosen grade? True, I didn’t give as many of
the

details as I have in the past, but did you get the idea of the
attempt

to analogize that investigation with the control of a cursor as in a

tracking experiment? […] How could you do it better?
Do you think it worth trying?

Sorry, Dick, I’ve got swallowed by work, and so I skimmed the first part
of this post and missed this in the second part until now.
I thought it was an interesting and it sounded like valuable approach to
pedagogy, but didn’t think of it as an experiment or a demonstration of
experimental methodology. I will re-read it and think some more about it,
perhaps this weekend.
I see that I need to return to Casting Nets and Testing Specimens
and reconsider that for answers to my questions. It’s been too long since
I read it, and my own understandings have shifted since then.

At the CSG meeting this summer I want to do a little demo of linguistic
work with an informant, and let those observing judge whether or not it
falls acceptably within PCT methodology. That’s assuming nothing steals
the time or the airfare from me.

    Bruce

Nevin

···

At 16:01 04/27/2001 -0500, Richard Robertson wrote:

[Chuck Tucker (2001.0428.08:05 EST)]

In a message dated 4/27/2001 4:59:16 PM Eastern Daylight Time,
r-robertson@NEIU.EDU writes:

<< There could conceivably have been 85 or more data
points (17 tests each taken 5 or more times) but that would have been
for the slowest learners. In fact the best students, those who
consciously tried to control for an A (90% or better) also tended to
supply the fewest data points. No real surprise, huh? Bill pointed out
you can't make much of a model from that. But, what do you think of the
attempt? How could you do it better? Do you think it worth trying? >>

I did not read you procedures carefully and I don't have a copy of them in my
files. Would you please send them to me? BTW, I have asked the students in
every class for the last 15 years this question:

"Given your ways of studying, your priority for this course and past academic
performance, circle the grade you will get in this course.

                                 A B+ B C+ C"

Of about 3500 students who have answered this question only about 10 have
selected a C+ or C about 90% of the students select "A." I can tell you that
my students do in fact "control" for the grades in my courses and most are
disappointed in their grade even though I do not "curve" my grades so it is
possible for everyone in the class to get an "A." It is rare for more that
20% of the students to get an "A" in my courses. I believe that with a
"normal" curve that only 10% would receive an "A."

Regards,
             Chuck

[From Dick Robertson, 2001.0429.1350CDT]

"Charles W. Tucker" wrote:

[Chuck Tucker (2001.0428.08:05 EST)]

In a message dated 4/27/2001 4:59:16 PM Eastern Daylight Time,
r-robertson@NEIU.EDU writes:

<< There could conceivably have been 85 or more data
points (17 tests each taken 5 or more times) but that would have been
for the slowest learners. In fact the best students, those who
consciously tried to control for an A (90% or better) also tended to
supply the fewest data points. No real surprise, huh? Bill pointed out
you can't make much of a model from that. But, what do you think of the
attempt? How could you do it better? Do you think it worth trying? >>

I did not read you procedures carefully and I don't have a copy of them in my
files. Would you please send them to me? BTW, I have asked the students in
every class for the last 15 years this question:

"Given your ways of studying, your priority for this course and past academic
performance, circle the grade you will get in this course.

                                 A B+ B C+ C"

Of about 3500 students who have answered this question only about 10 have
selected a C+ or C about 90% of the students select "A." I can tell you that
my students do in fact "control" for the grades in my courses and most are
disappointed in their grade even though I do not "curve" my grades so it is
possible for everyone in the class to get an "A." It is rare for more that
20% of the students to get an "A" in my courses. I believe that with a
"normal" curve that only 10% would receive an "A."

OK. You too sound convinced that students work for a grade can be viewed as an
effort to control their perception of it. You would only have had to have them
compute the difference between grade earned and reference grade, after each test,
to begin to resemble what I tried to set up. In addition, however, the big job
came for me in setting up a means for them to compute the error signal and show
any change in output to reduce error. That is where the multiple testing came
in, also the rule that they can't pass to the next chapter without passing the
previous one at a D level at least (But that would go into their running average
for the total course grade, so some retested a low-pass test even after moving
on.)

Let me repeat what I found in a general sense: A minority of students showed
clearly that they were controlling their perceived grade (throughout the course)
by raising (or even in a few cases lowering) amount of study time after passing a
chapter in regard to the magnitude of the last error signal. But many other
students revealed the working of a higher level in their hierarchy whereby they
lowered their reference grade to match the perceived grade as it was
developing--thus reducing error that way.

Sorry, I can't dig up all the procedures in detail. It's been over eight years
and we have moved in the meantime.

Best, Dick R.

···

[Chuck Tucker (2001.0502.12:09EDT)]

In a message dated 4/29/2001 2:55:18 PM Eastern Daylight Time,
r-robertson@NEIU.EDU writes:

<< You would only have had to have them compute the difference between grade
earned and reference grade, after each test, to begin to resemble what I
tried to set up. In addition, however, the big job came for me in setting up
a means for them to compute the error signal and show any change in output to
reduce error. That is where the multiple testing came in, also the rule that
they can't pass to the next chapter without passing the previous one at a D
level at least (But that would go into their running average for the total
course grade, so some retested a low-pass test even after moving on.)>>

My students seem to have a problem determining their current grade due to my
confusing (to them) grading system. The problem is that there are many
quizzes, exams and tests. There are daily quizzes which are worth 50% of
their grade, exams and tests worth 40% of their grades and a short paper
worth 10% of their grade. A student could get an "A" on all the quizzes and
exams but fail to do the paper and still receive an "A" in the course. If
the student scores less than perfect on all the quizzes and exams (the usual
case) then not doing the paper would result in less than an "A" grade. I
could give them what their current grade is but that would not be an adequate
"test" of a CV; they would have to do it on their own. Could you make a
suggestion as to how I make it possible for them to compute their current
grade given my current testing system?

<<Let me repeat what I found in a general sense: A minority of students
showed
clearly that they were controlling their perceived grade (throughout the
course)
by raising (or even in a few cases lowering) amount of study time after
passing a
chapter in regard to the magnitude of the last error signal. But many other
students revealed the working of a higher level in their hierarchy whereby
they
lowered their reference grade to match the perceived grade as it was
developing--thus reducing error that way. >>

Then you must have obtain information periodically about their study
activities. How did you get that information? How did you get the information
about them lowering their expected grade? I only know this if students tell
me that they expected a certain grade (usually an "A") and after several
tests decided that they would be satisfied with a "B" or "C."

Regards,
               Chuck

[Bruce Nevin (01.05.02 14:13 EDT)]

Chuck Tucker (2001.0502.12:09EDT)--

>Could you make a
>suggestion as to how I make it possible for them to compute their current
>grade given my current testing system?

Get a student web maven to write up a little calculator for your class web page. A prof in computer science might assign this to someone, for example. This should be an HTML form in which each element (quizzes, exams, paper) has a blank that they fill in with their actual (or, if desired, future) grade. Display the calculated result for them. You could have it access current grades from a file that you keep current. (There could be another form available only to you by which you update that database after each quiz, etc.) The program would have to use existing security mechanisms as for email to restrict the student to only her own data.

You might find others interested in this after you get it implemented and working smoothly.

  Bruce

···

At 12:13 05/02/2001 -0400, Charles W. Tucker wrote:

In a message dated 5/2/2001 2:16:00 PM Eastern Daylight Time,
bnevin@CISCO.COM writes:

<< You might find others interested in this after you get it implemented and
working smoothly. >>

Thanks for the suggestion; I will look into it.

Regards,
              Chuck

[From Dick Robertson, 2001.0502.1735CDT]

"Charles W. Tucker" wrote:

[Chuck Tucker (2001.0502.12:09EDT)]

In a message dated 4/29/2001 2:55:18 PM Eastern Daylight Time,
r-robertson@NEIU.EDU writes:

<< You would only have had to have them compute the difference between grade
earned and reference grade, after each test, to begin to resemble what I
tried to set up. In addition, however, the big job came for me in setting up
a means for them to compute the error signal and show any change in output to
reduce error. That is where the multiple testing came in, also the rule that
they can't pass to the next chapter without passing the previous one at a D
level at least (But that would go into their running average for the total
course grade, so some retested a low-pass test even after moving on.)>>

My students seem to have a problem determining their current grade due to my
confusing (to them) grading system. The problem is that there are many
quizzes, exams and tests. There are daily quizzes which are worth 50% of
their grade, exams and tests worth 40% of their grades and a short paper
worth 10% of their grade. A student could get an "A" on all the quizzes and
exams but fail to do the paper and still receive an "A" in the course. If
the student scores less than perfect on all the quizzes and exams (the usual
case) then not doing the paper would result in less than an "A" grade. I
could give them what their current grade is but that would not be an adequate
"test" of a CV; they would have to do it on their own. Could you make a
suggestion as to how I make it possible for them to compute their current
grade given my current testing system?

No, I'm afraid not, given your current grading system. I think you would have to
do something like I did, that is to have a student repeat his desired course
grade before taking each test, have him aware of his running average, score his
test immediately after taking it, have him recompute his running average and
subtract that from his desired grade ( I.e. say his desired grade is A {=.90 -
1.00, theoretically}, and his running average is now now .87, then his RS -PS, or
error is -.03. In a true control situation he show study harder and retake this
test, or do enough bettter on the next test to get his running average >.90.

<<Let me repeat what I found in a general sense: A minority of students
showed
clearly that they were controlling their perceived grade (throughout the
course)
by raising (or even in a few cases lowering) amount of study time after
passing a
chapter in regard to the magnitude of the last error signal. But many other
students revealed the working of a higher level in their hierarchy whereby
they
lowered their reference grade to match the perceived grade as it was
developing--thus reducing error that way. >>

Then you must have obtain information periodically about their study
activities. How did you get that information?

They were asked to indicate their study time since taking the last test.
Imperfect, sure, but I hoped a beginning approach to quantification.

How did you get the information about them lowering their expected grade?

Since they had to indicate their desired grade (RS) before taking each test, they
either confirmed the previously announced RS or changed it.

···

I only know this if students tell
me that they expected a certain grade (usually an "A") and after several
tests decided that they would be satisfied with a "B" or "C."

Best, Dick R.

[From Dick Robertson, 2001.0502.1745CDT]

Bruce Nevin wrote:

[Bruce Nevin (01.05.02 14:13 EDT)]

Chuck Tucker (2001.0502.12:09EDT)--

>Could you make a
>suggestion as to how I make it possible for them to compute their current
>grade given my current testing system?

Get a student web maven to write up a little calculator for your class web
page. A prof in computer science might assign this to someone, for example.
This should be an HTML form in which each element (quizzes, exams, paper)
has a blank that they fill in with their actual (or, if desired, future)
grade. Display the calculated result for them. You could have it access
current grades from a file that you keep current. (There could be another
form available only to you by which you update that database after each
quiz, etc.) The program would have to use existing security mechanisms as
for email to restrict the student to only her own data.

Right. A somewhat different procedure, but it looks like it would correspond
reasonably well. (If I comprehend it as Bruce intended.)

You might find others interested in this after you get it implemented and
working smoothly.

I would love to see that.

Best, Dick R.

···

At 12:13 05/02/2001 -0400, Charles W. Tucker wrote:

In a message dated 5/2/2001 6:44:17 PM Eastern Daylight Time,
r-robertson@NEIU.EDU writes:

<< No, I'm afraid not, given your current grading system. >>

Thanks for the suggestion. I might be able to incorporate it into what Bruce
N. mentioned in another post.

Regards,
               Chuck