What Happened to CSGnet?

[From Richard Pfau (2009.11.18.0135 EDST)

As a follow-up to the idea of someone involved in the modeling effort possibly contacting Dan Ariely to brief him on what you are doing, if contacted, you might ask him for copies of the original data that he, Drazen Prelec, and George Loewenstein collected from the 55 students in Prelec’s marketing research class at MIT’s Sloan School of Management. He or one of the others may still have the data sheets or data from each of the 55 students, including their individual 2-digit SS#s and bids for each of the 6 products bid upon.

If I understand your modeling approach, you might then calculate the gain for an individual student based on an individual student’s first bid or two, predict what his or her other bids would be, and check your predictions with the actual bids made.

Otherwise, perhaps someone with access to students could replicate the Arieby, Prelec, and Loewenstein experiment and obtain your own individual data with which to work.

For your info, Ariely’s book mentions that the chapter containing the summary data of the SS# and bid experiment was based on the following two articles (articles that I haven’t read, but may be worth looking at): (1) Dan Ariely, George Loewenstein, and Drazen Prelec, “Coherent Arbitrariness: Stable Demand Curves without Stable Preferences,” Quarterly Journal of Economics (2003), and (2) Dan Ariely, George Loewenstein, and Drazen Prelec, “Tom Sawyer and the Construction of Value,” Journal of Economic Behavior and Organization (2006).

With Regards,

Rich Pfau

RP: As a follow-up to the idea
of someone involved in the modeling effort possibly contacting Dan Ariely
to brief him on what you are doing, if contacted, you might ask him for
copies of the original data that he, Drazen Prelec, and George
Loewenstein collected from the 55 students in Prelec’s marketing research
class at MIT’s Sloan School of Management. He or one of the others
may still have the data sheets or data from each of the 55 students,
including their individual 2-digit SS#s and bids for each of the 6
products bid upon.
[From Bill Powers (2009.11.18.0745 MDT)]

Richard Pfau (2009.11.18.0135 EDST) –

BP: The problem is that the individual data won’t contain more than one
bid per item, so we can’t fit a model to the data for a single item with
different initial price numbers (SS#). I just think we’ll have to wait
for a different experiment to come along.

I see that my guesses about “A. Reilly” were a bit off the
mark. I mean totally.

As to contacting Dan Arielly, I’ll leave that to someone else. I have
received all the rejection I deserve, and it’s time to pass that
privilege on to someone else.

Best,

Bill P.

[From Dick Roberson,2009.11.18/1056CST]

[From Bill Powers (2009.11.18.0745 MDT)]
Richard Pfau (2009.11.18.0135 EDST) –

RP: As a follow-up to the idea of someone involved in the modeling effort possibly contacting Dan Ariely to brief him on what you are doing, if contacted, you might ask him for copies of the original data that he, Drazen Prelec, and George Loewenstein collected from the 55 students in Prelec’s marketing research class at MIT’s Sloan School of Management. He or one of the others may still have the data sheets or data from each of the 55 students, including their individual 2-digit SS#s and bids for each of the 6 products bid upon.

BP: The problem is that the individual data won’t contain more than one bid per item, so we can’t fit a model to the data for a single item with different initial price numbers (SS#). I just think we’ll have to wait for a different experiment to come along.

Could it work if the experiment were re-run with–after the first instructions (copying Ariely’s) – you said, "Assume you have learned the seller can be bargained with up to 5% to 20%, would that change your bid? If so, what would it be now?

As to contacting Dan Arielly, I’ll leave that to someone else. I have received all the rejection I deserve, and it’s time to pass that privilege on to someone else.

Richard P.: You seem to be in the best position to contact Ariely, given that you are farthest along in his book, and can cite the particular original studies. ??

Best,

Dick R
.

[From Bill Powers (2009.11.18.1300 MDT)]

Dick Robertson,2009.11.18/1056CST --

Could it work if the experiment were re-run with--after the first instructions (copying Ariely's) -- you said, "Assume you have learned the seller can be bargained with up to 5% to 20%, would that change your bid? If so, what would it be now?

I think the best route would be to give the subject many trials, each one with a different random number between 10 and 90 in steps of 20, with the instruction "This number is the first asking price for this item in a bargaining session. Assume you want to buy it. Write down how much you would offer as your first bid. Remember that your bids could only increase if the bargaining went on for this item."

An alternative would be to conduct actual bargaining sessions, with items costing from 10 cents to 90 cents. A model could be fit to both parties. Give the buyer $5.00 to start.

If we could actually get a model to fit a real bargaining session, so that after one or two rounds we could accurately deduce the buyer's and seller's reference levels, would that be the end of bargaining forever? How about the stock market?

Best,

Bill P.

[From Rick Marken (2009.11.18.1330)]

Bill Powers (2009.11.18.1300 MDT)–

Dick Robertson,2009.11.18/1056CST –

Could it work if the experiment were re-run with–after the first instructions (copying Ariely’s) – you said, "Assume you have learned the seller can be bargained with up to 5% to 20%, would that change your bid? If so, what would it be now?

I think the best route would be to give the subject many trials, each one with a different random number between 10 and 90 in steps of 20, with the instruction “This number is the first asking price for this item in a bargaining session. Assume you want to buy it. Write down how much you would offer as your first bid. Remember that your bids could only increase if the bargaining went on for this item.”

An alternative would be to conduct actual bargaining sessions, with items costing from 10 cents to 90 cents. A model could be fit to both parties. Give the buyer $5.00 to start.

If we could actually get a model to fit a real bargaining session, so that after one or two rounds we could accurately deduce the buyer’s and seller’s reference levels, would that be the end of bargaining forever? How about the stock market?

Why not do it as a test for the controlled variable? I suppose that’s what you would implicitly be doing after fitting the model to the data. If the model fit nearly perfectly for one individual but not another, then what?

Best

Rick

···


Richard S. Marken PhD
rsmarken@gmail.com
www.mindreadings.com

[Martin Taylor 2009.11.18.17.42]

[From Bill Powers (2009.11.18.1300 MDT)]

Dick Robertson,2009.11.18/1056CST –

Could it work if the experiment were re-run

with–after the first instructions (copying Ariely’s) – you said,
"Assume you have learned the seller can be bargained with up to 5% to
20%, would that change your bid? If so, what would it be now?

I think the best route would be to give the subject many trials, each
one with a different random number between 10 and 90 in steps of 20,
with the instruction “This number is the first asking price for this
item in a bargaining session. Assume you want to buy it. Write down how
much you would offer as your first bid. Remember that your bids could
only increase if the bargaining went on for this item.”

I think you would run into a problem similar to the one we are already
confronted with (pace Winston). No matter how you did the
bargaining, you couldn’t use the same object twice with the same
subject, so you couldn’t directly test within subjects by altering the
disturbance (the random number), unless you previously knew that
several objects were perceived by that subject to have the same value.
To do that, you would have to run paired comparison tests of the kind
“Would you swap this object for that as a fair trade?” You couldn’t ask
them to value the objects using numerical values beforehand. If you
didn’t do some such preliminary experiment, you would necessarily wind
up with different subjects evaluating any one object with the different
random numbers, and that’s what we have with the current experiment.

Martin

[From Bill Powers (2009.11.18.1710 MDT)]

Martin Taylor 2009.11.18.17.42 –

BP: earlier: I think the best
route would be to give the subject many trials, each one with a different
random number between 10 and 90 in steps of 20, with the instruction
“This number is the first asking price for this item in a bargaining
session. Assume you want to buy it. Write down how much you would offer
as your first bid. Remember that your bids could only increase if the
bargaining went on for this item.”

MT: I think you would run into a problem similar to the one we are
already confronted with (pace Winston). No matter how you did the
bargaining, you couldn’t use the same object twice with the same subject,
so you couldn’t directly test within subjects by altering the disturbance
(the random number), unless you previously knew that several objects were
perceived by that subject to have the same value.

I don’t understand. I’m proposing to have each subject make an opening
bid for all the items, bidding on the same item five times, each time
with a random number between 1 and 100 in place of the SS# used by
Arielly. The items could be scrambled so the person bids on a different
item each time, eventually covering 5 random prices for each item. It
will be explained that each time it’s like starting the bargaining over
with a new person selling.

To do that, you would have
to run paired comparison tests of the kind “Would you swap this
object for that as a fair trade?” You couldn’t ask them to value the
objects using numerical values beforehand.

Sure you could. Arielly does it right now: makes the subject write down a
random number (SS#) to be thought of as a price for that item, and then
asks how much the person would offer to buy that item.

If you didn’t do some such
preliminary experiment, you would necessarily wind up with different
subjects evaluating any one object with the different random numbers, and
that’s what we have with the current experiment.

You probably know some technical reason for doing this. What is
it?.

It would be better to have actual bargaining sessions. Try this: I offer
to sell you my 2001 Subaru Forester in good condition for $28,000. What’s
your counteroffer?

If we keep this up, we will either end up with no deal, or I will have a
series of prices proposed by you, and I will have to reconsider my offer
after each new price.

I must say that the correlations in my model are a lot higher than I
thought they would be. It looks possible that the population
characteristics show up as a virtual control system. Whether or not it
represents any individual, it gives the seller a good view of the
composite buyer as the crowd strolls past the items with the price-tags
on them. The buyers with reference levels below the price go on past; the
others buy it.

The main reason for the very high correlations in the “output”
pairings is that the RealBid - Asked (RealOutput) price is being
correlated with the modelBid - Asked (ModelOutput) price. The same
variable, the Asked price, contributes to both measures. If control were
perfect, the difference between RealBid[i] and ModelBid[i] would be zero,
and the correlation would be 1.000…

This sounds like cheating, but that’s just the way control systems work.
The controlled variable is the sum of the disturbance and the output: B =
A + O. Therefore O = B - A. If you correlate B(real) - A(real) with
B(model) - A(model) you get a very high correlation between A and
O.

The output data also shows the correlations between ReadBids and
ModelBids, which are not as high, but still higher than I thought they
would be: 0.862 to 0.974.

Best,

Bill P.

[Martin Taylor 2009.11.18.23.54]

[From Bill Powers (2009.11.18.1710 MDT)]

Martin Taylor 2009.11.18.17.42 –

BP: earlier: I think
the best
route would be to give the subject many trials, each one with a
different
random number between 10 and 90 in steps of 20, with the instruction
“This number is the first asking price for this item in a bargaining
session. Assume you want to buy it. Write down how much you would offer
as your first bid. Remember that your bids could only increase if the
bargaining went on for this item.”

MT: I think you would run into a problem similar to the one we are
already confronted with (pace Winston). No matter how you did
the
bargaining, you couldn’t use the same object twice with the same
subject,
so you couldn’t directly test within subjects by altering the
disturbance
(the random number), unless you previously knew that several objects
were
perceived by that subject to have the same value.

I don’t understand. I’m proposing to have each subject make an opening
bid for all the items, bidding on the same item five times, each time
with a random number between 1 and 100 in place of the SS# used by
Arielly. The items could be scrambled so the person bids on a different
item each time, eventually covering 5 random prices for each item. It
will be explained that each time it’s like starting the bargaining over
with a new person selling.

The problem is that after the first time a person has bid on the
object, he has established a mental price for that object. We are
asking the question of whether a prior suggestion of a price will
affect the price the person sets. We want that prior suggestion to be
just the random number, but if the person already has a price in mind,
it’s unlikely that the prior suggestion will be only the random number.
It’s quite likely that the price bid the last time will have a much
stronger influence than the random number, perhaps to the extent of
acting as a reference value for the price to be bid. The first time you
do it, the random number will very probably have a much bigger
influence than on subsequent occasions. Even if you scramble a very
large number of objects, you will not be able to guarantee that the
person has forgotten what price he set for that object on a previous
occasion.

To ask someone to ignore the previous bids is likely to work about as
well as a judge asking a jury to ignore some prejudicial information
the prosecution “inadvertently” let slip about the defendant.

Martin

[From Bill Powers (2009.11.18.2335 MDT)]

Martin Taylor 2009.11.18.23.54

The problem is that after the
first time a person has bid on the object, he has established a mental
price for that object. We are asking the question of whether a prior
suggestion of a price will affect the price the person
sets.

I’m not assuming that people work that way. In my model, the (composite)
person establishes a reference price that remains the same throughout.
What the person changes is the bid price, with those changes being like
an attempt to bring the price down to the reference price. The higher the
asked price is relative to the reference level, the harder the person
tries to lower the price, but that increased effort is caused by an
increased error signal when the asked price goes up. The bid price
increases, but not as much as it would if the buyer didn’t resist what he
sees as the seller’s excessive asked price.
This would be easier to understand if we included more than just this one
control system. If you ask why the person doesn’t just pay the asking
price whatever it is, I think the reason is simply that the person has a
limited budget and needs to spend money on other things, too. So spending
the money on the item disturbs other control processes like buying food,
keeping warm, running a car, and so on. If it weren’t for the need to
spend the money on other things, the person would just pay the asking
price, the way we do when we shop in stores.
It often happens that we want or need something, but find that the price
is more than we wanted to spend on it. This is like one round of
bargaining. The asked price for milk is $3.50 per gallon instead of the
$2.85 we anticipated paying, so we have to decide whether to buy it
anyway – that is, to offer more than our reference price for it. Maybe
our gain for keeping the price down is so high that the price we’re
willing to pay is only $3.00 when the asked price is $3.50. In that case
we don’t buy the milk, but perhaps go shopping at a different store, or
decide to buy skimmed milk which is closer to $2.85 per gallon, though we
don’t want it as much.
I’m challenging the assumption that subtle effects of perceptions on
other perceptions are enough to cause significant changes in behavior.
I’m saying there is another explanation for the observed apparent
effects in the Arielly experiment. The control-system model fits the data
well enough so my challenge isn’t thrown out automatically, which leaves
us to ask how well another theory can plausibly explain the same
observations. I don’t know if we could come up with a plausible
control-system model in all such cases, but it seems to work quite well
in this case.

Of course we could just fit y = ax + b to the data, but that doesn’t
suggest any explanation for why that should be observed. In fact, that is
the observation that needs to be explained. Aside from the control-system
model, what other explanation is there, beside “That’s just the way
it happens”?

We want that prior
suggestion to be just the random number, but if the person already has a
price in mind, it’s unlikely that the prior suggestion will be only the
random number. It’s quite likely that the price bid the last time will
have a much stronger influence than the random number, perhaps to the
extent of acting as a reference value for the price to be
bid.

I don’t think prior bids have any influence on the behavior, nor do the
social security numbers nor would the random numbers. Of course I could
be wrong, but you’d have to show me a model indicating the mechanism from
which these influences could be deduced, as we deduce the observed
effects from the control system model. Otherwise just saying
“influences” doesn’t tell us anything (certainly nothing
quantitative), and all we really have is curve-fitting.

The first time you do it,
the random number will very probably have a much bigger influence than on
subsequent occasions.

What model are you assuming that makes that seem probable? If you’re just
telling me you have a hunch that this will occur, you’re not making any
quantitative prediction or describing a method for constructing one. Any
amount of influence from the minimum observable to the maximum measurable
would make your statement true: an effect of one millionth of a cent
change in the bid price is clearly much larger than an effect of one
nanocent, so you can hardly go wrong with such a prediction. But how
about a real test?

Even if you scramble a
very large number of objects, you will not be able to guarantee that the
person has forgotten what price he set for that object on a previous
occasion.

So your model says that memory has something to do with it. Maybe it does
– when you work out the model enough so we can determine parameters and
fit it to the observations, we can compare it with the control system
model and see which does better. If your prediction about what will
happen under my proposed experiment turns out to be correct, then we will
find my model doesn’t predict individual behavior as well as it does for
the virtual population control system (if at all), and we will know more
than we know now. But I don’t have to guarantee anything to justify doing
the experiment as I propose. You’re just saying that under your model, if
the person remembers his previous bid, that will influence his next bid.
I am saying that his next bid will be determined by the quantitative
relationships in the control model I proposed, not by memory of the
previous bid. So one of us is wrong. What’s your justification for saying
I’m the one?

To ask someone to ignore the
previous bids is likely to work about as well as a judge asking a jury to
ignore some prejudicial information the prosecution
“inadvertently” let slip about the defendant.

Perhaps so. But you’ll have to spell out the model you’re using which
explains how memory of previous bids affects future bids, and show how it
predicts the data in the experiment I propose. If your model is the
better one, it will predict behavior that is different from the behavior
my model would predict and yours will be quantitatively more like the
observed behavior.

I think we’re at the point here where only quantitative models applied to
real data can resolve any issues.

Best,

Bill P.

[From Bill Powers (2-009.11.19.0745 MDT)]

Rick Marken (2009.11.18.1330) --

Why not do it as a test for the controlled variable? I suppose that's what you would implicitly be doing after fitting the model to the data. If the model fit nearly perfectly for one individual but not another, then what?

I agree that every model fit is a test for the controlled variable. Some models fit better than others even with the best adjustment of parameters.

As to differences between individuals, we expect different people to have different parameters. The tracking demo is an example. What would surprise me (though not too much) would be for one person to have a different architecture from another. In the Arielly example, people would probably have different reference levels for the price of a given item, and different gains, and a given person might have different reference levels and gains for different items. But the same model would probably work for most people. Just changing the parameters doesn't change the model.

Best,

Bill P.

[Martin Taylor 2009.11.19.09.52]

[From Bill Powers (2009.11.18.2335 MDT)]

Martin Taylor 2009.11.18.23.54 --

The problem is that after the first time a person has bid on the object, he has established a mental price for that object. We are asking the question of whether a prior suggestion of a price will affect the price the person sets.

I'm not assuming that people work that way. In my model, the (composite) person establishes a reference price that remains the same throughout. What the person changes is the bid price, with those changes being like an attempt to bring the price down to the reference price. The higher the asked price is relative to the reference level, the harder the person tries to lower the price, but that increased effort is caused by an increased error signal when the asked price goes up. The bid price increases, but not as much as it would if the buyer didn't resist what he sees as the seller's excessive asked price.

That's good. It's a testable model that can be compared with the other suggestions. The more suggestions we have that are compared with one another, the more likely it is that the best-fitting one has some relation to what is actually going on.

Incidentally, I'm not assuming that people work the way my proposal says, either. I'm suggesting that it's a possibility to be considered. Nor am I assuming they don't work the way you suggest above, or the way Rick has suggested. If I think of yet another possibility, I'll suggest that as well, without assuming it to be correct.

Martin

[From Richard Pfau, 2009.11.19.1015 EDST]

From Dick Roberson,2009.11.18/1056CST]

Richard P.: You seem to be in the best position to contact Ariely, given that you are farthest along in his book, and can cite the particular original studies. ??

Sorry, I’m just a layman when it comes to modeling and possibly stimulating Ariely to think about PCT. In his book he mentions that he receives hundreds of e-mail messages a day. I think one good way to connect with him, if it made sense to do so, would be for a PCT researcher to request copies of the individual data he collected on his social security number and product bid experiment at MIT, since preliminary PCT modeling and related theory seem to provide a good explanation for the results he obtained, but that the individual-level data he collected, if available, would help to further refine the modeling and explanatory effort being made. Such a request. from one professional to another, is reasonable and could be a first step in linking him to PCT.

If the results of such modeling, based upon his data, were published, we’d have a nice link between the worlds of behavioral economics and PCT.

But, alas, I don’t think that I am the right person to connect with Ariely in a way that will stimulate his interest.

With Regards,

Rich Pfau

The problem is that after the first time a person has bid on the object, he has established a mental price for that object. We are asking the question of whether a prior suggestion of a price will affect the price the person sets.
We want that prior suggestion to be just the random number, but if the person already has a price in mind, it’s unlikely that the prior suggestion will be only the random number. It’s quite likely that the price bid the last time will have a much stronger influence than the random number, perhaps to the extent of acting as a reference value for the price to be bid.

[From Dick Robertson,2009.11.19.1025CST]

[From Bill Powers (2009.11.18.2335 MDT)]
Martin Taylor 2009.11.18.23.54 --> The first time you do it, the random number will very probably have a much bigger influence than on subsequent occasions.
I’m not assuming that people work that way. In my model, the (composite) person establishes a reference price that remains the same throughout. What the person changes is the bid price, with those changes being like an attempt to bring the price down to the reference price. The higher the asked price is relative to the reference level, the harder the person tries to lower the price, but that increased effort is caused by an increased error signal when the asked price goes up. The bid price increases, but not as much as it would if the buyer didn’t resist what he sees as the seller’s excessive asked price.
DR: Naturally enough you’re looking at the S’s action as control over a variable like “acquiring an object of value,” or the like.
My understanding of Ariely’s own view of his experiment was that he wants his audience (us) to be amazed that – just having an irrelevant number in mind – when thinking of purchasing something, will influence the amount the S would bid for it. The hidden mechanism, as I interpret Ariely – since he doesn’t actually offer an opinion of what it is – is the old “association of ideas.”
When I first raised that speculation, you BP came out with your blockbuster hypothesis of “association of ideas” in terms of neural spillover, etc., that comes closer to a real explanation of the “phenomenon” of AoI than i have ever seen previously, and should be published in its own right. All references to the term in the whole literature of learning-“theory” (IMO) seem to regard it as a given, that either can’t be explained, or doesn’t need to be explained–like “all living organisms breath in some way.”
This would be easier to understand if we included more than just this one control system. If you ask why the person doesn’t just pay the asking price whatever it is, I think the reason is simply that the person has a limited budget and needs to spend money on other things, too. So spending the money on the item disturbs other control processes like buying food, keeping warm, running a car, and so on. If it weren’t for the need to spend the money on other things, the person would just pay the asking price, the way we do when we shop in stores.
DR: This approach comes way closer to trying to explain real behavior than does Ariely’s, where you have to accept that the lab results somehow are truly modeling something about real life.
It often happens that we want or need something, but find that the price is more than we wanted to spend on it. This is like one round of bargaining. The asked price for milk is $3.50 per gallon instead of the $2.85 we anticipated paying, so we have to decide whether to buy it anyway – that is, to offer more than our reference price for it. Maybe our gain for keeping the price down is so high that the price we’re willing to pay is only $3.00 when the asked price is $3.50. In that case we don’t buy the milk, but perhaps go shopping at a different store, or decide to buy skimmed milk which is closer to $2.85 per gallon, though we don’t want it as much.
DR: ditto
I’m challenging the assumption that subtle effects of perceptions on other perceptions are enough to cause significant changes in behavior. I’m saying there is another explanation for the observed apparent effects in the Arielly experiment. The control-system model fits the data well enough so my challenge isn’t thrown out automatically, which leaves us to ask how well another theory can plausibly explain the same observations. I don’t know if we could come up with a plausible control-system model in all such cases, but it seems to work quite well in this case.

DR: This is the real challenge of PCT vs S/R. I first thought it would be interesting to see if Ariely’s experiment would be reproducible (dammit, now I can’t think of the conventional term for that), and then, if they were, go on expanding the study to look for the real CV of the average subject. You have gone far beyond that now, but I gather that in the discussion between BP, MT and RM, you are working to agree on a procedure for modeling the control system. In trying to follow this discussion I lost track of how close you all are to a common-denominator of a model. Could that be summarized at this point?

Of course we could just fit y = ax + b to the data, but that doesn’t suggest any explanation for why that should be observed. In fact, that is the observation that needs to be explained. Aside from the control-system model, what other explanation is there, beside “That’s just the way it happens”?

DR: Yes, that is a different study, and potentially more interesting than just testing whether AoI is a potent influence in thinking about a “next” thought-in-line decision.

Best,

Dick R