Suzi Mayhem Outdoes John Cleese (page 7)

JAVA

Wendy,

Don't let Zazu fool you with that comment about being "old," she's one foxy woman! However, I really don't know what she sees in larc.

--JAVA

...counting time at the Coffee Shop

larc

Hey JAVA,

I feel compelled to defend myself. Zazu appreciates the gifts I give her. Why just last year I gave her a new air conditioner, concrete driveway, and a new lawn mower.

Larc (of the tongue in cheek class)

BugEye

Hey AlanF

Just wondering, what are the current penalties for disobeying a
law of physics. Last time I checked, it was a $1000 fine and a
good behaviour bond.

Dave

AlanF

To BugEye:

You can't disobey a "Law of Physics". Therefore there are no penalties.

In fact there is no such thing as a "law of physics" in the sense of "law" being a behest from some authority that you obey the authority. The universe is built in such a way that physical objects behave in certain ways, and our overall observations of how objects behave are distilled into succint statements that we mis-term "physical laws". This mis-terminology is left over from earlier centuries where "philosophers of science" thought that things behaved the way they did because they followed certain laws that were behests from God. People who think that there are actually such physical behests don't know what science is all about.

AlanF

Tallyman

.
.
Well,
all I know is
"Physics is Phun!"
.
.

JAVA

Brother larc,

I gave her a new air conditioner, concrete driveway, and a new lawn mower.

I didn't know they called the hand-held fans passed out at funeral homes with a picture of Jesus "air conditioners." Yeah, I'll bet Zazu appreciates the lawn mower you picked up at the gargage sale; she didn't know they made them without motors--what a gem. BTW, I saw your "concrete dirveway" the other day, and was glad you could use the broken pieces from your neighbors old sidewalk--it looks very new age. See what happens when you back out of beer night at the pub?

AlanF,

Guess I better stop using the term "law of gravity." Old habits or old sayings are hard to break. Now when I slip and say the "law of gravity" I'll need to stop and correct myself and give a little dissertation. So many words, so little time. . .

--JAVA

...counting time at the Coffee Shop

AlanF

To larc:

Thanks for the book suggestion. What's the book?

As for the questions that you and Focus have been discussing, I would need a lot more information about the subject you're dealing with, to comment effectively. I'd need concise definitions of the terms you're using, along with examples, since I'm quite unfamiliar with math as applied to psychology.

However, I can comment in a general sense. From your example of rating the 51 college students, it seems to me that things are not very clearly defined. Let me give you an example to illustrate my misgivings. Suppose you were asked to evaluate the responses of 1000 people who were asked to rate the color value of 100 color samples of carpet on a scale of zero to nine. From the responses you could form a 1000 X 1000 reliability matrix, from which you might try to say something about each person's reliability. But given the above, would you have any idea what I was talking about? I would hope not. Similarly, need a lot more information to understand the problem you're posing enough to comment.

One problem I see is to establish quantitatvely what is meant by a "rating scale", so that both the investigator and the subjects know exactly what they're being asked to do. Another is to precisely define what is meant by "correlation coefficient", since this means different things in different contexts. For example, if you're trying to fit data to an equation, you can define a correlation coefficient that is a measure of how well the equation fits the data. If you're trying to correlate some sort of semi-periodic, noisy data stream with another stream or with itself, to get an idea of periodicity or whatever, the definition of "correlation coefficient" would be somewhat different.

My impression is that in a "soft science" such as psychology, practioners don't know enough to be nearly as precise about definitions and such as in the "hard" sciences, so to be on the same wavelength practioners have to be particularly careful about definitions.

AlanF

Howdy JAVA

: Guess I better stop using the term "law of gravity." Old habits or old sayings are hard to break. Now when I slip and say the "law of gravity" I'll need to stop and correct myself and give a little dissertation.

I should say so! Otherwise the thought police will get you!

Seriously, these kinds of terms are perfectly fine as long as people know their limitations. It's like talking about "sunrise" -- perfectly fine as long as a person doesn't try to use the term itself as proof that the sun goes around the earth. Since I can't tell if BugEye is being serious or facetious, I figured I'd give a response that can equally be taken either way.

AlanF

larc

Alan,

I am not sure how to address your example about color. In my example, people all have some idea of the nutritional value of foods. People may vary in their knowledge and therefore in their accuracy in making such judgements, and that is fine for the purposes of my study. The scale is anchored a zero - no nutritional value and nine - highly nutritious, with five being of average nutritional value. I know that this lacks precision, but it is "good enough" for the purposes of the study. The judgements are analogous to those made in job performance appraisals, where much of the human judgement studies have been conducted. The studies clearly show that the reliability of such judgements do not increase when very accurate definiions are used and/or very detailed behavioral verbal ratings scale values are provided. Brief definitions and general verbal anchors work just as well. Although, the science is "soft", the implications are important. For example, the average reliability of managers judging your kind of work is only about .65. For peers, it is only about .40, and for self versus manager it is only about .25. If you are in management, the numbers are even worse. Upper level management's reliability is only about .55, and self versus your manager's judgement drops to .17. For practical reasons, it would be a good thing if we could harden up this soft science. We are working on it. By the way, there are some known methods to improve these numbers, but that story is for another day.

The correlation is between an equation, y= the square of x versus data, where x is a person's reliabilitity and the data, is found by correlationing a person's ratings with the sum of 50 other ratings. It is the square root of the correlation between two people as one variable versus the correlation between one person and the sum of 50 others as the second variable.

This is getting confusing. Let's try it this way. The square root of r1-2 correlated with r1-sum50 = .982.

larc

Regarding rating scales,

The rating is printed and instuctions given, so the subjects know what is expected. As far as scientist - pracitioners go, they have spent huge amounts of time on both the difinition of work dimensions being measured and the verbal anchors on the scale values. As I said above it doesn't seem to make much difference in improving human reliability. The only method that seems to work is toget multiple imputs. Something called 360 appraisal can significantly improve the resulting accuracy of appraisals. Given that r is the reliability of one person, then Kr/(k-1)r+1, where k is the number of people giving independent judgements, predicts how much more reliable the judgements will becme with increasing numbers of judgements.

Suzi Mayhem Outdoes John Cleese

--JAVA

--JAVA

Share this