Date: Wed, 25 Jun 1997 20:52:20 -0400
From: Thomas F Brown <tombrown@jhunix.hcf.jhu.edu>
Subject: Re: tracking again
To: socgrad@csf.colorado.edu
Message-id: <97Jun25.205221edt.762-4@jhunix.hcf.jhu.edu>
Content-transfer-encoding: 7BIT
References: <97Jun25.145411edt.2555-8@jhunix.hcf.jhu.edu>


>i'm being picky here, but this is interesting.
>so people's attitudes regarding the regime structure are considered a
>component of tracking until they (the respondent/the reported attitude)
>mentions unfairness -- and then the attitudes (the data) are
>conceptually thrown into the category of "outcome?"

Well, you're conflating the conceptual structure with the measurement
protocol. I haven't addressed how we're going to measure this stuff.
It is very tricky because of the overlap you describe. Another confounding
factor would be social influences that affect both an individual's choice
of discipline and also affect how people think disciplines should be
organized according to demographic characteristics. It could get messy.
It might be impractical.

But the only attitudes we need to know for this bivariate conceptual design
are what people think about who should go into which specialty.

>consider this:
>in order for an individual to have an attitude at all regarding the
>regime structure, one must have experienced it in some way -- which
>would really make all attitudes about the regime stucture "outcomes" to
>some extent, would it not?  

Yeah, that is exactly what I said--there is likely a reciprocal influence
between distributive outcomes and people's attitudes about how outcomes
should be distributed.

>why is it that you would value one's subjective attitudes when it comes
>to describing the regime structure, but would not equally value one's
>subjective attitudes regarding the fairness or unfairness of the regime
>structure?
 
In the first one, we would be surveying causal elements. We have to
assume the hegemonic discourse is in their collective head, and if we want
to get at it, we have to ask them. I doubt if they'd tell the truth.
This is a fantasy design. I think it's impractical. My intent was to
conceptualize the relationship between a hegemonic tracking discourse
and the distribution of specialties by demographic characteristics.

What I wanted to demonstrate was that if we infer tracking attitudes 
from the outcomes, we are defining the outcome in terms of the input--
a tautology. 

As long as we are defining the hegemonic discourse as a factor,
then I don't see any way around measuring subjective tracking attitudes and 
assuming that those attitudes determine how people are tracked. Whether or
not we can actually measure those attitudes reliably is another question 
entirely.

In the second one, it's irrelevant who thinks they've been discriminated 
against. We're looking at the relationship between the hegemonic discourse
and the actual distributive outcome.

>of course, you can conceptualize any way you want to (and i know you
>recognized the overlap in your post), but i'm not convinced that these
>two things are so different, even with your conceptual differentiation. 

They are conceptually different. One is a "should"--people should go into 
specialties according to their demographic characteristics. The other one 
is a "shouldn't"--I should not have been subjected to tracking pressure.
Separating them out in the measurement model would be tricky. Let's not
confuse measurement problems with conceptual problems.

>if you are allowing the measurement of attitudes in this analysis, why
>disallow attitudes of unfairness?  to me, this borders on conceptually
>screening the data.

Of course it's conceptually screening the data! That's what theory does--
it tells us which phenomena are important to the story and which are not.

I'm discriminating between endogenous and exogenous factors. The aggregate 
pattern of attitudes that describe a hegemonic discourse constitute one 
factor. The distribution of people across specialties is the other. Causal
direction is unknown, probably reciprocal.

Attitudes of unfairness are not included in my bivariate design.
How do you think they should fit in? Do you think they cause tracking?
I think they represent either a consequence of tracking or else they are 
some exogenous frustration that is being sublimated into a subjective
experience of being tracked. I do not think they cause either the hegemonic
discourse or tracking, therefore they are not included in the conceptual
model. They might constitute resistance to the hegemonic discourse.