Download “Proving” or “Disproving” Theories

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts

Frankfurt School wikipedia , lookup

Differentiation (sociology) wikipedia , lookup

Third culture kid wikipedia , lookup

Structural functionalism wikipedia , lookup

Sociological theory wikipedia , lookup

Postdevelopment theory wikipedia , lookup

Sociology of culture wikipedia , lookup

Working poor wikipedia , lookup

Development theory wikipedia , lookup

Transcript
Theory and research II
1.
2.
3.
4.
(3/26)
The input of theory to research
Knowing how: the e.g. of COP
Macro: regressions in States
The conceptual scheme of One World
5 Inputs of Theory to the
Research process
1.
2.
3.
4.
5.


Conceptualization – operationalization
Model Specification
Domain Specification (When?)
Generalization
Explanation (How? Why?)
In a research course you are mainly
interested in whether a particular
association exists.
In a theory course, you are interested in
what it means and when, how and why?
Knowing How v. Knowing That

These issues are relevant to the kinds of
disagreement that people have analyzing the
data on the effects of poverty (e.g. INCOME
@16) and culture of poverty (e.g. FAMILY
@16) on opportunity (e.g. $ RANK).
 That is, there are issues of conceptualization
and measurement.
 And, there are issues of interpretation of the
coefficients and partial coefficients.
Levels Re Culture of Poverty
We have discussed the relation of poverty and culture of poverty
several times. It is useful to think how these “levels” might relate
to the data that we analyzed last Monday. (See Below)
Kind of Assumption
Example Re Culture of poverty
General ideas about causality
In analyzing data on culture of poverty, are there
general issues about causality creating disagreement?
Causal Imagery
Do different theorists have different ideas about what
kinds of things are affecting what other kinds of things?
Classification of causes
What are the main structures and dynamics relevant to
the analysis of the culture of poverty?
Kinds of causes affecting other
kinds
What are the main theories about the negative effects
on opportunity of poverty and of culture of poverty?
Thesis that a variable explains
variation in another variable
What do the main theories about the effects of poverty
and culture of poverty imply, empirically?
Empirical consequences in
particular conditions
Empirically, what are the effects of INCOME @16
controlling FAMILY @16 and vice versa?
View that some particular data
are examples of a concept
What does this particular data suggest of show about
the effects and dynamics of culture of poverty?
I Theory and Operationalization

In order to see whether X
Y , you want to
get an index or measure of X and one of Y and
see whether and when they are associated
such at when there is more X there is more Y.
 An index or measure is an observable in some
data set that you have reason to suppose is
associated with the underlying variable of
interest.
 E.g. is FAMILY @16 and index of “Culture of
Poverty”? Aren’t some rich families not intact?
 Note: you often only operationalize a part of a
theory .
What is the effect of
measurement error?

Suppose that your index has a lot of noise in
it so that it is not a very good measure of the
underlying variable.
 This will usually just “attenuate” the
relationship, making it appear weaker than it
really is,
 So that the data is a conservative test of the
hypothesis.
 Noise is different from bias.
The effect of INCOME @16

INCOME @16


BELOW AVER


AVERAGE


ABOVE AVER


Missing

TOTAL8315

by
$ RANK
BELOW AVG
3653
39.3%
3699
23.8%
963
20.0%
2988
15417
8.1%
AVERAGE
4309
46.4%
9154
59.0%
1954
40.6%
4920
5877
52.1%
ABOVE AVG
1324
14.3%
2658
17.1%
1895
39.4%
2288
TOTAL
9286
100.0%
15511
100.0%
4812
100.0%
10375
29609
19.8%
•Gamma = .305
•What is the size of the effect of growing up poor on opportunities?
•What does this prove, what does it imply, and what does it suggest
•about the complex of cumulative poverty?
The effect of FAMILY @16

FAMILY @16

by
BELOW AVG
 YES 7638

26.2%
 NO 3662

34.5%
$ RANK
AVERAGE
15072
51.7%
5256
49.5%
ABOVE AVG
6469
22.2%
1694
16.0%
TOTAL
29179
100.0%
10612
100.0%
TOTAL11300
20328
8163
39791
28.4%
51.1%
20.5%



•Gamma = -.179
•What is the size of the effect of growing up in a non-intact family
•on opportunities?
•What does this prove, what does it imply, and what does it suggest
•about the complex of cumulative poverty?
What can we conclude from the
data?
 What
does it prove that gamma is .305?
 What does it imply?
 What does it suggest?
 What does it prove that gamma is -.179
 What does it imply?
 What does it suggest?
 What does it prove that .305 > .179
 What does it imply?
 What does it suggest?
II Theory and “model
specification”

Whenever one looks at any causal relation
empirically, there are always an indefinitely
large number of ‘other forces’ going on.
 The overall assumptions about the forces that
are operating are established and justified by
theory.
 A crucial element of model specification is
causal order. Does INCOME @16 $RANK
or does $RANK INCOME @16 ?
Why do we care which way?

The observable data often do not prove which
way the causal arrow goes.
 Some people in the 1950’s then said, “Let’s
stick with what can be proved from the data
(associations) rather than causal inferences.”
 Almost no one believes that; what is
important and interesting is the underlying
causal forces that brought about the
associations.
Why Systems and Feedbacks
are Inconvenient
 Often
there are a lot of specific causal
influences that have been demonstrated.
 But it is not clear how they fit together;
what is their dynamic; under what
conditions the effects obtain, etc.
 Whenever there are feedbacks, the
problems become intricate.
 E.g. Myrdal.
Systems and feedbacks about
the culture of poverty

Virtually all sociologists would agree the poverty and
the culture of poverty are mutually reinforcing.
Poverty
+
+

Culture of
Poverty
Most would also agree that INCOME @16 is a
reasonable measure of the effect of poverty and that
broken families (e.g. FAMILY @16) are a reasonable
measure of culture of poverty.
Why Systems and Feedbacks
are Inconvenient
 Often
there are a lot of specific causal
influences that have been demonstrated.
 But it is not clear how they fit together;
what is their dynamic; under what
conditions the effects obtain, etc.
 Whenever there are feedbacks, the
problems become intricate.
 E.g. Myrdal.
Clues about Causal order and
systems dynamics
 The
size and the relative size of the
empirical associations and partial
associations gives one indications of the
system dynamics.
 But one will always have to make model
specification assumptions.
 These must be theoretically motivated.
Controls
Some people believe that giving poor children’s
parents money (e.g. AFDC) will largely or entirely fix
the problems of those poor children who also have
broken homes (which is many of them.)
 Partly they believe that this will cause fewer homes to
break up.
 Some people believe that fixing children’s broken
homes (e.g. faith based programs) will largely or
entirely fix the problems of poor children.
 Partly they believe that this will pull most of the homes
out of poverty.
 The size and the relative size of INCOME@16 effects
and FAMILY @16 effects can be suggestive.
 The effect of one, controlling the other is even more
sugestive.

The effect of INCOME @16
controlling FAMILY @16

INCOME @16
by
$ RANK
 Controls:
FAMILY @16: NO


BELOW AVG
BELOW AVER 1446
42.6%
AVERAGE
907
8.9%
ABOVE AVER 208
23.7%
TOTAL
2561

34.6%






AVERAGE
1522
44.8%
1803
57.4%
392
44.7%
3717
ABOVE AVG
427
12.6%
430
13.7%
277
31.6%
134
50.1%
15.3%
TOTAL
3395
100.0%
3140
100.0%
877
100.0%
7412
•Partial Gamma = .301 (conditional gamma .260)
•What is the size of the effect of growing up poor on opportunities
• controlling culture of poverty?
•What does this prove, what does it imply, and what does it suggest
•about the complex of cumulative poverty?
Effect of FAMILY @16 controlling
INCOME @16 (showing only 1 conditional table.)
st

FAMILY @16
by
$ RANK
 Controls:
INCOME @16: BELOW AVER







BELOW AVG
YES 2207
37.5%
NO 1446
42.6%
TOTAL3653
39.3%
AVERAGE
2786
47.3%
1522
44.8%
4308
ABOVE AVG
896
15.2%
427
12.6%
1323
46.4%
14.3%
TOTAL
5889
100.0%
3395
100.0%
9284
•Partial Gamma = -.133 (conditional gamma -.098)
•What is the size of the effect of culture of poverty on opportunities
• controlling growing up poor?
•What does this prove, what does it imply, and what does it suggest
•about the complex of cumulative poverty?
Controls as an answer to
“because”



Ordinarily if there is a relations between X and Y and
you control T, and the original relationship goes away,
that means that the original relationship is “due to” or
“because of” the controlled variable.
And if there is a relations between X and Y and you
control T, and the original relationship does not go
away, that means that the original relationship is not
“due to” or “because of” the controlled variable.
And if there is a relations between X and Y and you
control T, and 1/3 the original relationship goes away,
that means that 1/3 the original relationship is “due
to” or “because of” the controlled variable.
What can we conclude from the
data?

What does it prove that the partial gamma of
INCOME @16 controlling FAMILY @16 is
about the same as the bivariate?
 What does it imply?
 What does it suggest?
 What does it prove that the partial gamma of
FAMILY @16 controlling INCOME @16 is a
little smaller than the bivariate?
 What does it prove that gamma is -.179
 What does it imply?
 What does it suggest?
Two different cases of because
(why causal order makes a diff.)
urbanism
storks
Birth rate
spuriousness
bleeding
Shot in heart
Intervening
variable
death
III) Domain Specification







A theory is a claim.
Usually it applies to some set of cases more limited
than all social structures in all of recorded history,
but much more general than the cases on which the
claim is based.
Theory involves establishing the domain of the
theory.
Statistical interactions are the main clues about
domains.
A mechanism (a “WHY?”) establishes a general
domain.
If the coins in my pocket are quarters because of the
hole in the bottom of my pocket, then the coins will be
quarters whenever such a hole exists.
IV) Generalization
 Particular
findings, empirical
generalizations, and hypotheses (e.g.
Protestants have higher suicide rates)
need to be related to more general
processes.
 Conceptualization (e.g. “deviance”
rather than “crime” or “suicide”) is partly
a matter of generalizing.
Theory simplifies to the
“essential”

It is trivially true that both:


and that:



Culture influences social structure,
and social structure influences culture.
and also that:


functional and conflict processes operate
individuals create social structures and social structures
shape individuals.
However, it is also trivially obvious that any theory
must simplify, and that models that include everything
are usually too complex to use or test.
V) Explanation
 The
conceptualization, and the
establishment of the conditions and size
of the effects is basic to establishing
what is the mechanism that brings it
about.
 The main paradigms propose
mechanisms.
Summary:Theory and research

Research establishes that there is an
association.
 Theoretical questions involve Why? How?
and When?
 I.e. what direction does the causal arrow run
in, under what circumstances, why and how?
 Often it is only the cumulative result of the
scientific process over generations
The main paradigms in
sociology


1.
2.
P. 267-276 of OW shows that the different
maps of the main theoretical positions in
sociology can be translated into each other.
They boil down to two dimensions: functional
v conflict and micro v macro. However:
The 20-odd different sections of sociology
such as medical sociology contain
importantly different theoretical positions.
Any way of dividing the 20,000 or so
practicing sociologists into a small number of
“schools” is bound to simplify
The main map
Functional
macro-theory
e.g. Durkheim,
Parsons
Organizational- Conflict macromacro-theory
theory
e.g. Weber’s
e.g. Marx
rationalization
Functional
microtheory
e.g. some
aspects of
Merton
Interactionist
micro-theory
e.g. Mead
Conflict microtheory
e.g. Mills
Organization theories as a Mix
 The
interactionist/organization theories
stemming from Weber, Mead, and
others, can be viewed as an ambivalent
synthesis of elements of conflict and
functional theory.
 Often the elements that distinguish
them from functional or conflict theory
appear at the micro level.
Micro-theory v Macro-theory

Micro-theories mainly treat social structure as
the outcome of individual choices and actions.
 Parsons took Weber’s action theory as the main
model.
 Other American sociologists took George
Herbert Mead’s interactionism as a model.
 The main difference between rational-action
theories stemming from Weber and symbolic
interaction theories, stemming from Mead is the
nature of the tinker-toy, but they are both tinkertoy models.
Macro-theory
 Macro-theories
focus on the fact that
humans and human behavior is shaped
by the social structure.
 This leads to concentrating on how
social structures influence their
members and other social structures.