Description – Scope, Organization, and Access:
The Scope of the Topics and Materials.
The class will be organized around issues of research
development and design. We will use a good, standard textbook
by Singleton & Straits that provides a reasonable overview of
the knowledge commonly expected in a basic graduate, social science
methodology class. We will augment this with a
highly-regarded book, The Craft of Research, that aims at
the conceptual level of the processes by which we go from
disconnected ideas, to research questions, to transforming our
research into arguments, and to making those arguments into
effective papers. Along the way, we will also read a range of
articles and other materials that provide further illustrative
examples and allow us to examine some important topics that are too
often neglected. These other materials will be
available on line, through the class web site.
The class
organization and goals. We will run all class meetings as
discussions. Every student must come prepared to every class and
participate. Each week, a written task analysis will be
due. A term paper in the form of a full research proposal
will be due at the end of the semester.
Written assignments will include short papers every week. On
occasion these will be group efforts of several students, otherwise
individual efforts. In the weekly writing tasks, students
will attempt to solve research design issues related to the week's
readings. In the second half of the semester, these weekly
efforts will also be aimed at the research proposal that is the
culminating work for the class. All assignments must be
handed in on time. I will not give incompletes.
Generally, all students' weekly writing assignments will be
circulated among all class members before our class meeting, so
that we can discuss them.
The research proposal produced as the final project of this class
is intended to be the MA thesis proposal for AQR students.
This is a daunting goal for all of us.
Academic integrity. In this class, the AQR
Program, and at NYU more broadly, we expect and require students to
adhere to the highest standards of academic conduct. Students
who engage in academic dishonesty will be subject to review and the
possible imposition of penalties in accordance with the standards,
practices, and procedures of NYU and its college and schools.
Violations may result in a failure on a particular assignment, a
lower course grade, failure in a course, suspension or expulsion
from the University, or other penalties.
Students are often encouraged to seek outside assistance from
tutors, writing coaches, and online resources. Such behavior is
permitted when the intellectual contribution of completed work is
that of the student, and when outside assistance is appropriately
acknowledged. Outside assistance must never contribute to
intellectual content. Assignments, research papers, and other
materials submitted for evaluation or review must be a student’s
own in their entirety.
Further detail on NYU policies and procedures related to academic
integrity are
available online.
Books Recommended for Purchase:
Author
|
Title
|
Publisher |
Royce Singleton & Bruce Straits |
Approaches to Social Research (6th ed)
|
Oxford |
Wayne Booth, et. al. |
The Craft of Research (6th ed) |
Chicago |
Jane Miller |
Writing About Multivariate
Analysis |
Chicago |
The Topics
1. Introduction.
Let us start by thinking about
the task ahead of us.
- Analytical
Task
- Prepare to discuss the strengths and the weaknesses of
the research designs behind the two New York Times articles
in our readings.
- Common Readings
- Approaches to Social Research: Ch 1
- Craft of Research: Ch 1
- Examples of quality newspaper social research:
- Hartmann, Stephan, and Jan Sprenger. "
Mathematics and Statistics in the Social Sciences." In
The Sage Handbook of the Philosophy of Social Sciences,
edited by Ian Jarvie and Jes Zamora-Bonilla. London: SAGE
Publications Ltd, 2011. [doi:10.4135/9781473913868]
- Coser, Lewis A. "Two
Methods in Search of a Substance." American
Sociological Review 40, no. 6 (1975): 691-700.
[doi:10.2307/2094174]
- Recommended
Readings
- Related
Readings
2. Elements of research design.
Research design, not statistical
expertise, largely decides if a research project
succeeds. What does it take to go from a question about
how things work in the world to an answer that can be defended with
evidence and sound logic? It takes knowledge about the
specific issue, about the relevant social processes, and about how
to do good research; it requires a thoughtful plan that balances
scientific aspirations with practical possibilities; it takes a lot
of work over time; and it takes a nimble responsiveness to the
unexpected. To do it really well also requires a disciplined
willingness to recognize and respond to the limits of the research
and to that which very few researchers can abide: evidence that we
are wrong.
- Analytical
Task
- The general
analytical problem. Academic departments in
universities vary considerably in size. Why?
What explains which are large and which are small?
- Transform this general problem into a research
project. Consider the elements of a research project
that are discussed in the readings. Among others,
these include: clear formulation of the question,
alternative explanations, measurement, appropriate data,
sampling design, and analytic strategy.
- This week's task will be done in assigned
groups.
- Common Readings:
- Approaches to Social Research: Chs
2, 4
- Craft of Research: Chs 2-4 (pp.
27-64); as needed, also Chs 5-6
- The regular generation of survey and census data:
- Anderson, Margo. "Censuses:
History and Methods." International
Encyclopedia of the Social and Behavioral Sciences
(2015): 302-07. [doi:10.1016/b978-0-08-097086-8.41007-x]
- Valente, Paolo. "Censuses:
Current Approaches and Methods." International
Encyclopedia of the Social and Behavioral Sciences
(2015): 296-301. [doi:10.1016/b978-0-08-097086-8.41006-8]
- Kukutai, Tahu, and Victor R. Thompson. "Censuses,
Population: Comparative International Aspects." International
Encyclopedia of the Social and Behavioral Sciences
(2015): 290-95. [doi:10.1016/b978-0-08-097086-8.41004-4]
- Hoffmeyer-Zlotnik, Jürgen H. P., and Uwe Warner. "Data
Bases and Statistical Systems: Sociology." International
Encyclopedia of the Social and Behavioral Sciences
(2015): 844-50. [doi:10.1016/b978-0-08-097086-8.41021-4]
- Juster, F. Thomas. "Data
Bases and Statistical Systems: Economics (General)."
International Encyclopedia of the Social and
Behavioral Sciences (2015): 742-50.
[doi:10.1016/b978-0-08-097086-8.41032-9]
- Kaase, Max. "Data
Bases and Statistical Systems: Political Science
(General)." International Encyclopedia of the
Social and Behavioral Sciences (2015): 830-35.
[doi:10.1016/b978-0-08-097086-8.41019-6]
- Kreyenfeld, Michaela, and Frans Willekens. "Data
Bases and Statistical Systems: Demography." International
Encyclopedia of the Social and Behavioral Sciences
(2015): 735-41. [doi:10.1016/b978-0-08-097086-8.41013-5]
- Otte, Gunnar, and David Binder. "Data
Bases and Statistical Systems: Culture." International
Encyclopedia of the Social and Behavioral Sciences
(2015): 727-34. [doi:10.1016/b978-0-08-097086-8.41069-x]
- Schmitt-Beck, Rüdiger. "Data
Bases and Statistical Systems: Political Behavior and
Elections." International Encyclopedia of the
Social and Behavioral Sciences (2015): 824-29.
[doi:10.1016/b978-0-08-097086-8.41075-5]
- Smith, Tom W. "International
Social Survey Programme." International
Encyclopedia of the Social and Behavioral Sciences
(2015): 550-56. [doi:10.1016/b978-0-08-097086-8.44080-8]
- Quandt, Markus, and Ruud Luijkx. "Data
Bases and Statistical Systems: International
Comparative Research." International
Encyclopedia of the Social and Behavioral Sciences
(2015): 789-97. [doi:10.1016/b978-0-08-097086-8.41017-2]
- Straf, M. "Government
Statistics." International Encyclopedia of the
Social and Behavioral Sciences (2015): 301-07.
[doi:10.1016/b978-0-08-097086-8.42130-6]
- Schnell, Rainer. "Record
Linkage." In The Sage Handbook of Survey
Methodology, 662-69, 2016.
[doi:10.4135/9781473957893.n41]
- Ultee, Wout. "Problem
Selection in the Social Sciences: Methodology." International
Encyclopedia of the Social and Behavioral Sciences
(2015): 49-55. [doi:10.1016/b978-0-08-097086-8.44043-2]
- Recommended
Readings
- Related
Readings
3. Causality - What are causes, mechanisms, and the like?
We casually refer to causes and
effects in normal interactions all the time. We all conduct
our lives – choosing actions, making decisions, trying to influence
others – based on theories about why and how things happen in the
world. From the early stages of childhood we attribute
causes, building a vision of the social (and physical) world that
makes it understandable. Every action, every choice about
what to do, is based on our anticipation of its effects, our
understandings of consequences. Analytical and scientific
reasoning has a similar form, but requires that we approach
causation more systematically and self-consciously.
- Analytical
Task
- Special individual task ...
- Look over this information
on data sets that are recommended for
AQR MA theses. Look at the
descriptions online and investigate the
data that could support research projects
you might want to pursue for your thesis
(and your research proposal for this
class).
- Note that next week each student must
provide a brief prospectus for three
distinct possible projects that would serve
as the focus of this class's
research proposal and the MA
thesis. Look at these data sets with
that goal in mind.
- Task summary: Develop a simple
comparison of two (or more) competing possible
causal arguments and show how you might resolve
which causal argument better fits the real world
with each of two data sets.
- Restrict yourselves to the most basic social
variables, including (among others): sex, age,
race/ethnicity, education, income, marital
status, political leaning, and religion.
You can add one less basic social variable if it
is the focus of your argument, for example
attitudes regarding abortion.
- Decide on a reasonably simple, but hopefully
still interesting causal question. For
example, in the U.S, is a high proportion of the
difference in incomes by race due to differences
in the economic status of people's parents
(independent of race)?
- On the substantive, theoretical level, try to
specify causal models or arguments that show the
distinct mechanisms that would make the two
alternate possibilities "work".
- If we are considering race, class origins,
and income, for example, we might write
about one scenario where only class origins
systematically affect people's adult incomes
but the historical relationship between class
and race means that this intergenerational
income effect tends to reproduce the race
difference.
- The alternative scenario might include
arguments about the ways that racial
discrimination (potentially both current and
past discrimination that had institutional or
structural consequences like geographical
segregation) has an effect on racial or
ethnic groups independent of their class
origin.
- Write a brief, but reasonable,
description of these causal alternatives.
- On the research design level, pick two of the data sets on our
list that let you assess which of the
alternative arguments are more
accurate.
- Identify the variables that are
relevant, including any controls that you
feel need to be considered.
- Indicate how you can use these variables to
distinguish between the alternatives.
Do not worry about the details of which
statistical procedures would be
appropriate. Focus on the logic of the
design.
- Suggest how the presence of different
data (e.g., the same variables measured
differently, the presence of some variables
absent from the data set, a different sample
of the population) could improve the research
design.
- In short, each group must agree on a causal
question that mainly involves the basic
characteristics recorded in surveys, then it must
write a report that develops the causal
alternatives and shows how the question might be
resolved using two of the listed data sets.
- Clarifications. You are
trying to devise a straightforward
project to test the relative validity or
explanatory power of two competing
explanations.
- For two explanations to compete, the two
causal models must imply different outcomes
under some conditions (although they may
agree on the expected outcomes for many
conditions).
- This task does not have any restrictions
about the number of "variables" (independent,
intervening, or other) in your causal
models. Nor does it say how much or in
what way the competing explanations must
differ, only that they are significantly
different in some way that permits
testing. For example, you could be
trying to explain why some people become
liberal and some conservative using a model
with a few independent variables and some
intervening variables where the difference
between your competing causal models concerns
whether education does or does not moderate
the influence of class background and
parents' political perspective.
- With respect to the data sets, the goal is
to show how you could test the alternative
explanations using the data available in two
different data sets, assessing for each of
the two data sets what their strengths or
weaknesses might mean for testing the causal
question. One of the important,
practical, methodological challenges is to
decide which of the many existing data sets
can better allow us to test a causal
hypothesis, or if no existing data is
sufficient for our purposes. So, for
each of the two data sets you choose as
plausible for your study, your goal is to say
how well they will work for solving your
problem. For example, you might find
the sample is too small for you to look at
subgroups separately as you feel you should,
or you might find that the education variable
is too crude because you feel you need the
quality of colleges attended, or you might
decide that the questions asked for one of
the data sets do not allow you to identify
respondents' political leanings with enough
precision or confidence. We have to
look closely and critically at data before we
use it to understand its virtues and
shortcomings.
- This week, therefore, we are doing a trial
run on how we formulate competing
explanations of some problem we believe we
can test using existing data and, then, how
we assess how well the data will serve to
resolve our explanatory competition once we
have specified it.
- Common Readings:
- Recommended
Readings
- Wikipedia. "Causality"
- Barringer, Sondra N., Scott R. Eliason, and Erin Leahey.
"A
History of Causal Analysis in the Social Sciences."
In Handbook of Causal Analysis for Social Research,
edited by Stephen L. Morgan, 9-26. Dordrecht: Springer
Netherlands, 2013. [doi:10.1007/978-94-007-6094-3_2]
- Abbott, Andrew. "The
Causal Devolution." Sociological Methods &
Research 27, no. 2 (1998): 148-81.
[doi:10.1177/0049124198027002002]
- Marini, Margaret Mooney, and Burton Singer. "Causality
in the Social Sciences." Sociological Methodology
18 (1988): 347-409. [doi:10.2307/271053]
- Nagel, Ernest. "Determinism
in History." Philosophy and Phenomenological
Research 20, no. 3 (1960): 291-317.
[doi:10.2307/2105051]
- Dray, W. H. "Determinism
in History." In Encyclopedia of Philosophy,
edited by Donald M. Borchert, 35-41. Detroit: Macmillan
Reference USA, 2006.
- Related
Readings
- Little, Daniel. 1991. Varieties of Social Explanation: An
Introduction to the Philosophy of Social Science. Boulder,
CO: Westview Press. Pp.
1-87
- John Gerring. "Causation:
A Unified Framework for the Social Sciences." Journal
of Theoretical Politics, (2005) 17(2), 163-198.
(doi:10.1177/0951629805050859)
- Epstein, Joshua M. "Why
Model?." Journal of Artificial Societies and
Social Simulation 11, no. 4 (2008).
-
-
4. Causality 2 - Actors, Conditions, Processes, Structures
& the role of "counterfactuals"?
Without becoming philosophers of
science, effective social science researchers must have a
reasonable grip on thinking about causality. Even most social
research not aimed at questions about social causation usually
relies on critical assumptions about causation and can only be used
in arguments or policies that overrun with causal thinking.
Unfortunately, causal thinking is difficult and fads guide causal
argument choices as much as (and often more than) rigorous
logic. For some time the framework of counterfactuals has
held considerable influence in social science, and scholars steeped
in its rhetoric often hold themselves superior to those who do not
adhere to its argument style. Truthfully, one would be hard
pressed to identify research or theoretical development in social
science that have gone further using a counterfactual language than
they would have without it, but lacking that rhetorical skill puts
one at a disadvantage. So here we want to get the basic ideas
clear. At the same time, for a contrast, it is worth looking
carefully at the practical strategy for assessing causality put
forth by the epidemiologist and statistician, Sir Austin
Bradford Hill a half-century ago that had profound and lasting
influence on real-world assessments of causality and public health.
This should remind us that in science and in life, causality is
ultimately a practical problem, not a philosophical one.
- Analytical
Task
Major Task: Present three
alternative possible research plans that could
serve as your semester research proposal
project. These will necessarily be highly
provisional and undeveloped. Still, for
each try to indicate:
- What is the research problem or research
question?
- What is the principal literature relevant to
the question.
- What, initially, are two or more likely causal
interpretations? These should simply be
clearly stated.
- What data might you use? Indicate the
data source and, if possible, the principal
variables.
- If possible, state how the data could be used
to answer the research question. Do not
attempt details about a statistical analysis, but
rather just a substantive statement about what
comparisons might be used. It is fine to
use a graphical representation if preferred.
- Why is this research worth doing?
Each of
the above points and others are covered in
greater detail in the analytical task description
for Section 8 of the course below. Please
look there if any of the above is unclear.
Try to limit the
description of each of the possible
projects to no more than one single-spaced page.
These are each meant to be précis; they should be
clear but they should suggest what each
project might look like, not describe
it.
- Common Readings
- Freese, Jeremy, and J. Alex Kevern. "Types
of Causes." In Handbook of Causal Analysis for
Social Research, edited by Stephen L. Morgan, 27-41.
Dordrecht: Springer Netherlands, 2013.
[doi:10.1007/978-94-007-6094-3_3]
- Angrist, Joshua D., and Jörn-Steffen Pischke. "The
Experimental Ideal." Chap. 2 In Mostly Harmless
Econometrics: An Empiricist's Companion, 11-24:
Princeton University Press, 2008.
- Lebow, Richard Ned. "What's
So Different About a Counterfactual?" World
Politics 52, no. 4 (2011): 550-85.
[doi:10.1017/S0043887100020104]
- Freedman, David. "From
Association to Causation: Some Remarks on the
History of Statistics." Statist. Sci.
14, no. 3 (1999/08 1999): 243-58.
[doi:10.1214/ss/1009212409]
- Winship, Christopher, and Stephen L. Morgan. "Causality
and Empirical Research in the Social Sciences"
and "Counterfactuals
and the Potential Outcome Model." Chap. 1
& 2 In Counterfactuals and Causal
Inference: Methods and Principles for Social
Research, 3-76. Cambridge: Cambridge
University Press, 2014. [doi:
10.1017/CBO9781107587991.002,
10.1017/CBO9781107587991.003] - (Give these,
particularly Ch 2, a first reading. For
most, trying to absorb this in depth the first
time round will not be productive.)
- Hill, Sir Austin Bradford. "The
Environment and Disease: Association or Causation?." Journal
of the Royal Society of Medicine 108, no. 1 (2015 [orig 1965]):
32-37. [doi:10.1177/0141076814562718]
- Supplementary Class Notes:
- Recommended
Readings
- Hedström, Peter, and Petri Ylikoski. "Causal
Mechanisms in the Social Sciences." Annual
Review of Sociology 36, no. 1 (2010):
49-67. [doi:10.1146/annurev.soc.012809.102632]
- More on counterfactual analyses:
- King, Gary, and Langche Zeng. "When
Can History Be Our Guide? The Pitfalls of
Counterfactual Inference." International
Studies Quarterly 51, no. 1 (2007):
183-210.
[doi:10.1111/j.1468-2478.2007.00445.x]
- Mahoney, James, Gary Goertz, and Charles C.
Ragin. "Causal
Models and Counterfactuals." In Handbook
of Causal Analysis for Social Research,
edited by Stephen L. Morgan, 75-90.
Dordrecht: Springer Netherlands, 2013.
[doi:10.1007/978-94-007-6094-3_5]
- Fearon, James D. "Counterfactuals
and Hypothesis Testing in Political Science."
World Politics 43, no. 2 (1991):
169-95. [doi:10.2307/2010470]
- Menzies, Peter. "Counterfactual
Theories of Causation." In The
Stanford Encyclopedia of Philosophy,
edited by Edward N. Zalta: Metaphysics
Research Lab, Stanford University, 2014.
- Winship, Christopher, and Stephen L.
Morgan. Counterfactuals
and Causal Inference: Methods and
Principles for Social Research.
(2014). [doi:10.1017/CBO9781107587991] - (This
book is a sophisticated, sustained analysis
by true believers.)
- The issue of path dependence:
- Liebowitz, Stan J., and Stephen E.
Margolis. "Path
Dependence." In Encyclopedia of Law
& Economics, edited by Boudewijn
Bouckaert and Gerrit De Geest: Edward Elgar,
2000.
- Hardin, Garrett. "The
Tragedy of the Commons." Science
162, no. 3859 (1968): 1243-48.
[doi:10.1126/science.162.3859.1243]
- Platt, John. "Social
Traps." American Psychologist
28, no. 8 (1973): 641-51.
[doi:10.1037/h0035723]
- Goldstone, Jack A. "Initial
Conditions, General Laws, Path Dependence,
and Explanation in Historical Sociology."
American Journal of Sociology 104, no.
3 (1998): 829-45. [doi:10.1086/210088]
- David, Paul A. "Path
Dependence: A Foundational Concept for
Historical Social Science." Cliometrica
1, no. 2 (July 01 2007): 91-114.
[doi:10.1007/s11698-006-0005-x]
- Mahoney, James. "Path
Dependence in Historical Sociology." Theory
and Society 29, no. 4 (August 01 2000):
507-48. [doi:10.1023/a:1007113830879]
- Liebowitz, S. J., and Stephen E. Margolis.
"Path
Dependence, Lock-in, and History." The
Journal of Law, Economics, and Organization
11, no. 1 (1995): 205-26.
[doi:10.1093/oxfordjournals.jleo.a036867]
- Related
Readings
5. Surveys and experiments: Archetypes for Research
Design
The experiment is the classic
research design by which all others are measured. Physicists,
medical researchers, biologists and others all use experiments when
possible. Experiments are exercises in control that directly
match our understanding of causation. The researcher divides
subjects (whether they be people, rocks, events, or something else)
into two or more groups by random assignment or some analytically
controlled selection rule, then the subjects receive differential
"treatments," and finally the subjects are observed to discern if
any consistent differences between the experimental groups suggest
that exposure to the treatment has a causal impact. In
sociology and related disciplines, we are unable to use experiments
to discover the impact of most causal influences. We must
rely on observation of "natural" variations. From participant
observation of ongoing activities to counting artifacts (such as
wills) to sorting through government records (such as death
certificates), social scientists mine the social world for
data. Here the sample survey substitutes for the experiment
as the archetypal research design. If we fully
understand the issues of survey research design, we are generally
able to handle any research design in the social sciences.
- Analytical
Task
- The
general analytical problem. Choose
one of your possible research proposal
topics. Select two different sources of
social survey data that might be used to study
this topic. Systematically compare their
strengths and weaknesses.
- Among things to consider:
- What is the population that was
surveyed? How was that population
sampled? What response rate did they
achieve?
- When did the survey take place? If it
is a longitudinal study, what period does it
cover, and is it a trend study with repeated
cross-sectional surveys or a panel study that
follows the same respondents over time?
- How were the data gathered from respondents?
- What kinds of sample weights exist?
- What principal variables exist in the data
that would serve your project? How were
the data collected for those variables.
For variables that directly reflect responses
to a question, examine how the question was
worded, asked, and coded. For variables
such as scales or indices that were assembled
by combining and recoding, how were they
produced?
- Suggest which data set would be preferable to
use for the project, and defend that choice.
- Common Readings
- Peer Reviewing the Initial Project Ideas of Section
4
- Prepare notes based for our class meeting on the
project ideas submitted last week by
other members of your new work group as
described below. The papers
containing the project ideas from Section 4 can be
found in the shared space on Google Drive where
they were submitted.
- Click here
for an outline of issues to consider while
reviewing the initial project ideas (both of others
in your group and your own). You should
prepare brief notes on each of the three project
ideas submitted by each member of your work group.
- Click here
to see the work group membership for this
exercise.
- Recommended
Readings
- Related
Readings
6. Sampling
Selecting a subset of the
target population for research is fundamental
to all social science research. Even research that
seems to observe every member of a population generally
depends on sampling ideas. A population census, for
example, has to be sampled to produce the public use
samples made available to scholars for research.
Experimental designs often neglect sampling concerns,
relying on random assignment to comparison groups; yet,
such randomization occurs within the sample of subjects
(people or otherwise) available for research.
Generalization to the full population or beyond depends
on the implicit sampling that has occurred. Research
using organizations, nations, or other collective
entities faces significant issues establishing the
population being studied, often reversing the logical
sequence by trying to define the population based on the
available sample. In short, all social research
must contend with sampling as a crucial facet of research
design and defense of inferences. Since acquiring
data involves time, effort, and expense, and may be
curtailed by many obstacles, sampling is both a practical
and a conceptual problem for research.
- Analytical
Task
- The general
analytical problem.
Select one of your possible research projects as
the focus for this task. Assuming you were
free to design a new research survey, consider
some of the standard sampling strategies that
might be appropriate for this project. Assess
how the research aims would be better or worse
served by these alternate strategies.
- As shown in the readings, alternative
sampling strategies involve trade-offs in
costs, richness of the data collected,
ability to study sub-samples, applicable
statistical procedures, how far we can
generalize results, and the like. These
are the concerns you should bear in mind
while assessing the strengths and weaknesses
of a sampling strategy for your project.
- Consider a fair range of sampling
possibilities, such as simple random,
stratified random, some form of
nonprobability sampling with constrained
sample size, and longitudinal sampling.
Try to show how and why your research
objective would be better or worse served by
the various sampling strategies.
- Include consideration of varying scope
possibilities for the research project, such
as that you have to collect the data personally,
that you are asked to include a comparison
across countries, or that you are asked to
focus your research on cities with over one
million inhabitants.
- After completing the general assessment of
fit between your research agenda and possible
sampling strategies, examine what this
implies about the possible fit between your
proposed project and existing data sets that
you could consider using.
- In short, for your selected research
project, you will assess the trade-offs of
competing sampling strategies, under varying
conditions about the intended scope of the
research project, and decide what this
assessment of sampling implications suggests
about the fit of existing data sets with your
project.
- Common Readings
- Approaches to Social Research:
Ch 6 (Sampling)
- Statistics Canada, Social Survey Methods Division. Survey
Methods and Practices. 2003. Read Ch. 6
"Sample Designs" and Ch 8 "Sample Size".
- Duncan, Greg J. "Panel
Surveys: Uses and Applications." In International
Encyclopedia of the Social & Behavioral Sciences
(Second Edition), edited by James D Wright, 462-67.
Oxford: Elsevier, 2015.
- Fienberg, Stephen E., and Judith M. Tanur.
"
Sample Survey Methodology, History Of."
International Encyclopedia of the Social and
Behavioral Sciences (2015): 875-80.
[doi:10.1016/b978-0-08-097086-8.03229-3]
- Bautista, René. "An
Overlooked Approach in Survey Research: Total Survey Error."
In Handbook of Survey Methodology for the Social Sciences,
edited by Lior Gideon, 37-49. New York, NY: Springer New
York, 2012. [doi:10.1007/978-1-4614-3876-2_4]
- Berk, Richard A. "An
Introduction to Sample Selection Bias in Sociological Data."
American Sociological Review 48, no. 3 (1983): 386-98.
Read only pages 386-390. [doi:10.2307/2095230]
- Johnson, David R. "Using
Weights in the Analysis of Survey Data."
(2008). (This is a PDF of PowerPoint slides
from a talk on weighting.)
- Read the sampling documentations for these two surveys
to get a feel for the sampling processes used for
professional, real world surveys:
- "Appendix
A: Sampling Design & Weighting ." In
General Social Surveys, 1972-2016:
Cumulative Codebook, edited by Tom W.
Smith, Michael Davern, Jeremy Freese and
Michael Hout, 3110-29. Chicago: NORC, 2017. (Note
the history of changes in the sampling design
for this important general survey that has
been around much longer than you have.)
- "Appendix
1: Survey Methodology." In Survey of
LGBT Americans, Attitudes, Experiences and
Values in Changing Times, 113-19.
Washington, D.C.: Pew Research Center, 2013. (This
example shows an effort to reach a special
population. The strategy only works if
one has considerable resources.)
- Comments
on Three Possible Research Proposals⇒ Click Here
- Recommended
Readings
- Statistics Canada, Social Survey Methods Division. Survey
Methods and Practices. 2003. Pp. 119-130 on weighting.
"Sample Designs" and Ch 8 "Sample Size".
- Groves, Robert M., Floyd J. Fowler, Jr., Mick P. Couper,
James M. Lepkowski, Eleanor Singer, and Roger Tourangeau.
"Weighting."
In Survey Methodology, 347-54. Hoboken: John Wiley
& Sons, 2009.
- United Nations, Statistical Division. Designing
Household Survey Samples: Practical Guidelines.
Studies in Methods Series F. New York: United
Nations, 2008. (another general resource, with
an international orientation)
- Elliott, Caroline, and Dan Ellingworth. "Assessing
the Representativeness of the 1992 British Crime Survey:
The Impact of Sampling Error and Response Biases." Sociological
Research Online 2, no. 4 (1997).
- Till, Yves, and Alina Matei. "Basics
of Sampling for Survey Research." In The Sage
Handbook of Survey Methodology, 311-28, 2016.
[doi:10.4135/9781473957893.n21]
- Related
Readings
7. Observation and gathering data
Data have to be collected and
processed. Even if we use existing data from past censuses,
surveys, or government records, our research depends crucially on
the processes that produced our data. If we treat data as
simple, easily interpreted, unambiguous, valid indicators of what
we want to know, our plan is sunk before we leave the dock.
To have a good idea what we need to look for and worry about, we
need to understand how data comes into being, what good practices
are, and what are the many reasons that our data might not
represent what we want.
- Analytical
Task
- The general analytical
problem. Revise a set of
proposed survey questions to avoid common issues.
- Scenario: You are employed in a
research institute that conducts surveys. A
group seeking to understand how college students use
drugs and alcohol submits a sequence of
questions. These will become part of a questionnaire
that will be answered by a sample of
undergraduates from NYC colleges. Your job
is to revise these questions as well as possible
so that they conform to good practices for
surveys. You do not need to address
the range of information being gathered, only the
quality of the questions being used.
- This task will be done in these
work groups (click here).
- The original questions are
available here. Note that this
question web page was produced by MS Word, so you
can open it directly in MS Word and save it as a
Word *.docx file (the formatting will be
preserved).
- Common Readings
- Approaches to Social Research:
Ch 10-12 (Survey Instrumentation, Field Research,
Research Using Available Data)
- Smyth, Jolene D. "Designing
Questions and Questionnaires." In The Sage Handbook
of Survey Methodology, edited by Christof Wolf,
Dominique Joye, Tom Smith and Yang-chih Fu. London: SAGE
Publications Ltd, 2016. [doi:10.4135/9781473957893]
- Statistics Canada, Social Survey Methods Division. Survey
Methods and Practices. 2003. Read Ch. 5
"Questionnaire Design" and Ch 10 "Processing".
- Gideon, Lior. "The
Art of Question Phrasing." Chap. 7 In Handbook of
Survey Methodology for the Social Sciences, edited by
Lior Gideon, 91-107. New York, NY: Springer New York, 2012.
[doi:10.1007/978-1-4614-3876-2_7]
- ⇒Compiled
responses to the analytical task, organized by
survey question.⇐
- Recommended
Readings
- Related
Readings
- Lyberg, Lars E., and Herbert F. Weisberg. "Total
Survey Error: A Paradigm for Survey Methodology." In The
Sage Handbook of Survey Methodology, edited by Christof
Wolf, Dominique Joye, Tom Smith and Yang-chih Fu. London:
SAGE Publications Ltd, 2016. [doi:10.4135 9781473957893]
8. Assembling the pieces of a research project. Good
aspirations and causal concerns.
Midway through our journey, let us
pause, stand back, and consider the research project as a whole,
given our understandings up to this point.
- Analytical
Task
Major Task:
Present a thesis topic proposal. This will
be your initial sketch of the proposal. It will be
followed by a full rough draft in three weeks. In
this preliminary version you should include all the
sections below. For each, provide a brief statement of
what you now think will be the substance of this part and
indicate what you believe you need to do to complete the
section effectively. For example, for the literature
review section, you might here give a brief assessment of
what literature is relevant and what you expect it will
show, then indicate what you need to do to produce a full
literature review.
- What is the
research problem or research question?
- Be succinct.
- If needed, review Craft of Research,
pp. 3-48; Approaches to Social Research,
pp. 81-83, 105-108.
- Who are the
expected audiences?
- For our purposes, the audiences will usually be
scholars, other social scientists. Yet, we want
to consider which social scientists we
anticipate will find our research of interests.
Is the group broad or narrow? Is it confined to
one discipline or does it go across discipline
boundaries? Will public officials or
practitioners of some kind want to know the answers
from our research?
- The motive here is not to justify the research, but
to identify whose existing knowledge and expectations
concern us.
- This part should be brief. Just state who are
the audiences and why. The goal is to keep them
in mind while completing everything else.
- What are the
relevant literatures that need to be explored?
- You are not expected to gain familiarity with the
relevant literature for these project sketches,
beyond what you need to know to complete the
sketch. Still, you should show you know where
to start, what are some of the principal works that
inform the kind of work you are proposing. Try
to explain what strategy you will use to pursue a
full literature search for this project.
- To put it differently, try to give a sense of the
scope and direction of the literature review that is
broad, yet explicit enough that someone else who read
it carefully could begin to carry out the literature
for you.
- If needed, review Craft of Research,
pp. 65-83; Approaches to Social Research,
pp. 558-64.
- What, initially,
are believed to be the competing causal
interpretations?
- Your research design should state Cn
competing causal explanations (where n is 2
or higher)
- These competing explanations should have at least
surface plausibility, in the sense that it should not
be easy to show an explanation is wrong based on
common knowledge or ten minutes research on the
internet.
- The competing explanations should also permit
representation with the kinds of "variables" in
existing social data sets. (It is possible for
people to generate new data sets for their MA theses,
for example by linking data from two or more existing
data sets. However, only those with appropriate
experience and skill should consider such a strategy,
and even for those, this is probably too early to be
considering such a strategy.)
- As needed, review Approaches to Social
Research, pp. 92-105.
- Be wary of the pitfalls involved in causal
reasoning, for example
- Sometimes two conditions have the capacity to
influence each other so that we cannot easily
talk about one being the cause. For
example, harsh parenting my produce rebellious
children but rebellious children may produce
harsh parenting. Without further
data, we cannot call one side of this
relationship the leading cause for it escalating
over time.
- Sometimes the composition of the groups being
compared is influenced by the condition we would
like to assess as a cause. For example, a
new charter school serving an area of a city may
appear a success given its relatively high
graduation rate, but that could reflect its
selective admission of the best students rather
than it providing better education. Without
further information, we cannot distinguish the
causal effects of the selection process and the
educational process.
- What if the central goal of your research project
is descriptive, not causal?
- Research that aims to discover telling
empirical conditions can be valuable without
seeking to test any causal arguments. For
example, no one really knows just how much police
violence occurs against different ethnic groups
under varying conditions in the U.S. or how often
police have to make spontaneous decisions under
dangerous conditions - any research that could
provide an accurate assessment would be very
important. How are economically successful
adults who come from poor families different from
those who come from affluent families? How
do young adults who are American Vietnamese,
American Korean, American Japanese, and American
Chinese and others cope with their common
characterization as Asians?
- If the project goals are primarily descriptive,
causal facets will be secondary, but they are
rarely absent. Our expectations about
police violence, the effects of economic
mobility, or the experience of a frustrated
ethnic identity all hinge on ideas about causal
processes, even if probing those causes is not
our research agenda.
- Therefore, if the project is primarily
descriptive, aiming to discover new knowledge
about the way things look rather than why they
look that way, we should make that clear, but
spend some effort exploring the related causal
processes. By showing knowledge of the
possible causal processes, we demonstrate our
command of the problem and validate the goal of
establishing the empirical conditions.
- What comparisons
do you anticipate are possible starting points for a
research design?
- The simplest research design usually implies a two
by two table: we have two possible causal conditions
and two possible outcomes. For example, some
subjects get the trial drug while the others get a
placebo, some subjects improve over time while some
do not. Or, some people are identified as
ethnically Caucasian and others ethnically Asian, and
some people get post graduate degrees and some do
not.
- Social research questions commonly involve more
complex comparisons, but they always focus on
comparisons. Selecting the right comparisons is
critical for research success.
- Hypotheses are statements about the expected
outcomes of these comparisons, based, hopefully, on
sound theoretical premises.
- As needed, review Approaches to Social
Research, pp. 105-110, and read 511-521.
- What data might
you use?
- The basic requirement here is that the project
proposal identifies some appropriate data that we can
expect will work to solve the research
question. Usually, this will mean identifying
one or more existing data sets, indicating the key
variables needed from those data sets, and stating
why the samples from those data sets are appropriate
for the research objectives.
- Clarify the relevant units of analysis for the
research. What are the "cases" the study looks
at? Are they individuals, households, firms,
cities, or what? Note that the unit of analysis
for a project may differ from the original cases in a
data set. For example, we could take households
as our unit of analysis while using data that has
cases for all individuals in all surveyed households;
or we could have city neighborhoods as our unit of
analysis but use individual level data from the
public use samples of census data that we aggregate
to the neighborhood level.
- What kind of sampling applies to using these data
for your research? If we use existing data
sets, the initial sample concerns the sampling
procedure used for that data set. If we plan to
use only some of the cases (e.g., only married
couples with both spouses employed) or to aggregate
cases, than we have to consider the sampling
implications of our procedure on top of the original
sampling for the data set.
- Even though this is the initial proposal for the
project, a sketch, try to be as clear as you can
about the factors (represented by variables) that are
the elements of your anticipated research and how
they will be used.
- What plausible
timetable could the project be designed around?
- Every stage of a research project takes time, most
take longer than we expect. To protect
ourselves and to appear moderately reliable to
others, we need to have a practical plan.
The goal for these projects is to function as the
basis for an MA thesis. So, consider how much
time you can reasonably have available for this
activity for perhaps the months January through
August this coming year, after the time required for
your studies, for any jobs you will hold, for your
other obligations, and the time you will have
available each week to work on a thesis
project. Then, consider how to budget time for
each part of the project.
- As this is just the initial proposal for the thesis
topic, this should be brief. Still, try to
paint a rough picture of a schedule. You and we
need a starting point for considering the viability
of the project in terms of its size (i.e., the time
and effort required).
- Why is this
research worth doing?
- In our circumstances, the honest answer is that it
is a means toward getting a degree. We should
not lose sight of this criterion. An MA thesis
needs to demonstrate the capacity to carry out a
basic research project from beginning to end.
Realistically, it does not need to make an important
contribution to scholarly literature, and rarely will
do so.
- Nonetheless, to produce an appropriate research
proposal, we want to claim that our project has a
purpose that is relevant to our audience.
Generally this means that it should resolve a
question that is apparently not answered in the
existing literature, if the proposal is being judged
by scholars. Alternatively, it should claim to
supply an answer to a practical question or evidence
to support a policy position if the proposal is being
judged by a commercial employer, a policy
organization, or the like.
- Always think through and stress discovery.
The aim of social research is to find out something
we did not know before we did the research.
This would be true even of research that seeks to
replicate previous findings, as the goal would be to
discover how confident we can be in those prior
findings.
- In more general terms, research is potentially
important only if the outcome has bearing on
theoretical claims that matter to scholars or
the outcome has bearing on practical policies in the
world that some people care about. Essentially,
the research results must have the potential to alter
some people's future behavior.
- For research to have this potential, two things
must be true.
- First, either some people must feel
they do not know what the outcome will be or
some must believe something about the world that
we could plausibly show was wrong by the
research. This means the answer to the
research question must be open to disagreement
and the research must have the potential to
provide evidence toward resolving that
disagreement. Doing research to show that
desperately poor people given an opportunity are
more likely to steal small amounts of money and
valuables than are the very affluent is probably
a waste of time. Absent some considerable
class difference in cultural prohibitions against
theft or casual indifference to antisocial
behavior, many more poor people will experience
circumstances making small thefts valuable.
In contrast, research on whether affluent or poor
people are more likely to steal amounts that are
significant relative to their current
circumstances, given equivalent opportunities,
asks a question for which we cannot predict the
answer.
- Research also can have significance only if
relevant people are open to being swayed by the
research results. Will those who would
make wrong predictions about the outcomes, or
those who believe no one knows the answer, be
convinced by the research? Are there people
defending a position whose expectations would be
confirmed by the research, and whose resolve or
sense of purpose might be bolstered even if the
research results are neglected by their
opponents?
- Thus, the research plan seeks to show that it
can provide evidence that has bearing on a point
of contention and the evidence should be
convincing to (rational) people concerned about
the questions.
- For some common flaws in presenting thesis topic
proposals, ⇛go
here⇚.
- Common Readings
- Hamming, Richard. "You
and Your Research." Bell Communications Research
Colloquium Seminar (1986). (Read this with
care. Few have ever said more clearly what
distinguishes those who do important work from those who do
not. Ignore the discussion after the formal talk.)
- Mahoney, James, and Gary Goertz. "A
Tale of Two Cultures: Contrasting Quantitative and
Qualitative Research." Political Analysis 14,
no. 3 (2017): 227-49. [doi:10.1093/pan/mpj017] (Read
this not for insights on differences between quantitative
and qualitative research strategies, but for the way it
makes you think about research opportunities and choices.)
- Lieberson, Stanley. "Modeling
Social Processes: Some Lessons from Sports." Sociological
Forum 12, no. 1 (1997): 11-35. (Read this for
insights on how you can broaden your ways of thinking
through analytical problems and strategies. Lieberson
uses sports, but the lessons are about expansive and
imaginative thinking.)
-
Comments on the Thesis Topics Proposals (Individuals) ⇒
Click Here ⇐
Comments on the Thesis Topics Proposals (Common) ⇒
Click Here ⇐
- Recommended
Readings
- Erren, Thomas C., Paul Cullen, Michael Erren, and
Philip E. Bourne. "Ten
Simple Rules for Doing Your Best Research, According to
Hamming." PLOS Computational Biology 3, no.
10 (2007): e213. [doi:10.1371/journal.pcbi.0030213] (For
an effort to translate Hamming into a pithy list)
- Hartmann, Stephan, and Jan Sprenger. "Mathematics
and Statistics in the Social Sciences." In The
Sage Handbook of the Philosophy of Social Sciences,
edited by Ian Jarvie and Jes Zamora-Bonilla. London: SAGE
Publications Ltd, 2011. [doi:10.4135/9781473913868]
- Related
Readings
9. Measurement
All research
depends on the assumption that we can assess
the characteristics of actors, relationships among actors,
and events accurately and realistically enough to allow
us to describe them and analyze how the influence each
other. Our capacity to do effective analyses depend
crucially on the processes by which we categorize and
count social phenomena. These processes are prone
to errors at many levels.
- Analytical
Task
- Analytical
Task: Develop a generalized indicator of
relative standing in American society.
- Starting point: We have an idea that every adult
in the U.S. has a definable general or aggregate standing
compared to any other adult. We know, of course,
that we have varied inequalities that order people, such
as wealth, income, race/ethnicity, gender, education,
neighborhood of residence, occupation, parent's social
status, and so forth. Our hypothesis is that these
all contribute to a person's generalized social standing,
and that we should be able to merge the indicators in
some fashion that will produce a fairly consistent and
effective indicator of that generalized standing in
society.
- Our goal is to create a provisional generalized
indicator using the General Social Survey from recent
years, and to offer a plan to revise and augment GSS
indicators in future years to produce an improved
generalized social standing index.
- To this end, we want to do the following:
- Identify what
existing indicators in the GSS are relevant to our
goal.
- Explain and
justify how we would produce a generalized social
standing index from these indicators. Include
an answer to each of the following:
- How might you
decide what each has to contribute?
- Would you
reduce the number of indicators you use, and if so,
how?
- How would you
combine the selected indicators to produce the
aggregate index of social standing.
- How would you
try to verify to worth of your combined
index? How would you assess its validity,
reliability, and accuracy?
- Offer suggestions
of how to revise or augment the existing indicators
to produce a better generalized index in the future.
- In short, you will attack the conceptual and technical
problem of creating an index that tries to measure
inequality in American society as an across-the-board
condition.
- This task will be done in
groups. (The
resulting papers.)
- Common Readings
- Approaches to Social Research: Ch
5 (Measurement); Ch 13 pp. 426-34 only ("Multiple
Measures ... ")
- Streiner, David L. "Being
Inconsistent About Consistency: When Coefficient Alpha Does
and Doesn't Matter." Journal of Personality
Assessment 80, no. 3 (2003/06/01 2003): 217-22.
[doi:10.1207/S15327752JPA8003_01]
- Schaeffer, Nora Cate, and Stanley Presser. "The
Science of Asking Questions." Annual Review of
Sociology 29, no. 1 (2003): 65-88.
[doi:10.1146/annurev.soc.29.110702.110112]
- Alwin, Duane F. "Survey
Data Quality and Measurement Precision." In The
Sage Handbook of Survey Methodology, 527-57, 2016.
[doi:10.4135/9781473957893.n34]
- Billiet, Jaak. "What
Does Measurement Mean in a Survey Context?." In The
Sage Handbook of Survey Methodology, 193-209, 2016.
[doi:10.4135/9781473957893.n14]
- Schwarz, Norbert. "Attitude
Measurement." International Encyclopedia of the
Social and Behavioral Sciences (2015): 178-82.
[doi:10.1016/b978-0-08-097086-8.24006-3]
- European Commission, Joint Research Centre.
"10
Step Guide" to creating composite indicators.
Read through all 10 steps (see the Handbook on
Constructing Composite Indicators in
Recommended Readings for a fuller treatment).
- Recommended
Readings
- Related
Readings
10. How to Prepare a Good Literature Review.
Preparing a good literature review
is a task easily overlooked. Yet, a weak literature review
in the early stages can jeopardize one's chances of developing a
good research design. At later stages, a poor literature
review can damage a paper that otherwise reflects good
research. A good literature review depends on two critical
skills: (1) knowing how to find the existing research and
theoretical work that is relevant to one's project and (2)
knowing how to select and present the important ideas and
findings in that literature.
- Analytical
Task
- Prepare a basic literature review, aimed at the
topic of the thesis proposal.
- This literature review should conform to
recommendations about good standards for a literature
review from the readings.
- Among other goals, the literature review should:
- Give a full account of existing
research (and theoretical work) on the same question
- or very similar questions - as the research
question of this proposal. Depending on what is
available, this should include the most important
publications of the past and the most recent
publications (for recent publications, we include all
that seem plausible).
- Provide a selective account of the research that
informs the previous work and your own proposal, but
is not directly on the research topic. For
example, if we were doing a research project asking
how having a baby during the first two years of
marriage influences the likelihood of divorce in the
subsequent three years, we would want to include a
selective review of the literature on the determinants
of reproduction in marriage and of divorce.
- Pay close attention to what has been discovered and what
we do not know about the issue that is the aim of your
research proposal. Also pay attention to what are
sources of disagreement in the literature.
- Common Readings
- Craft of Research:
review Chs 5 & 6 (From Problems to Sources, Engaging
Sources - for literature reviewing - Read these
two chapters to help preparing your literature review);
Part IV, pp. 173-267 (Writing Your Argument -
for general writing strategies - Look over these
chapters to find material that will help you more
generally with writing the thesis proposal )
- The following offer advice about literature
reviews. Use these as practical guides.
- Knopf, Jeffrey W. "Doing
a Literature Review." PS: Political Science
& Politics 39, no. 1 (2006): 127-32.
[doi:10.1017/S1049096506060264]
- Graham, Charles R. "Reviewing
the Literature When There Is So Much of It."
Chap. 3 In Conducting Research in Online and
Blended Learning Environments, edited by Anthony
G. Picciano, Charles D. Dziuban, Charles R. Graham and
Patsy D. Moskal, 28-42: Routledge, 2015.
[doi:10.4324/9781315814605]
- Boote, David N., and Penny Beile. "Scholars
before Researchers: On the Centrality of the
Dissertation Literature Review in Research
Preparation." Educational Researcher 34,
no. 6 (2005): 3-15. [doi:10.3102/0013189x034006003]
- Jackson, Robert Max. "How
to Develop a Literature Review – Some Suggestions."
- The following offer further advice about literature
reviews. Take a quick look at them, read further if
their approach or specific information seems to better fit
your needs than those above.
- Recommended
Readings
- Denney, Andrew S., and Richard Tewksbury. "How
to Write a Literature Review." Journal of
Criminal Justice Education 24, no. 2 (2013):
218-34. [doi:10.1080/10511253.2012.730617]
- Hart, Chris. Doing a Literature Review: Releasing
the Social Science Research Imagination. London:
Sage Publications, 1998.
- Porter, Alan L., Alisa Kongthon, and Jye-Chyi Lu. "Research
Profiling: Improving the Literature Review." Scientometrics
53, no. 3 (March 01 2002): 351.
- Related
Readings
11. Writing, diagramming, and Modeling
Multivariate Relationships.
Good research design commonly
involves a precarious balance between too much and too
little. Most social phenomena of interest involve a complex
interplay between actors, processes, structures, and expectations
over time. Good theory depends on a selective abstraction
from the complexities of life as experienced to create a
simplified but telling model of decisive facets. Good
research must similarly pursue a selective view of reality,
focused on a limited set of causal relations that are sufficient
to make sense of some phenomena we wish to understand, without
losing us in irrelevant details or misleading us by leaving
out things that are important. Being able to think clearly
about the relationships among multiple facets of a theoretical
model is the mirror of thinking clearly about the possible
relationships among multiple variables in a research design and a
data analysis. Hypothetical models of the possible
relationships between social phenomena give us the goals for our
research design while analyzing the patterns found in our data
give us the insights to adjust our models.
- Analytical
Task
- Full rough draft of thesis research proposal.
- All parts of the future proposal should exist in this
draft and be analytically developed.
- All comments received for the thesis topic proposal
should be addressed in this draft.
- For a research proposal outline, ⇒
Click
Here ⇐. This is a generic proposal
outline. You can use it as a suggestive template,
but you do not need to follow it in detail.
- Common Readings
- Recommended
Readings
- Judd, Charles M., and David A. Kenny. "Data
Analysis in Social Psychology: Recent and Recurring
Issues." In Handbook of Social Psychology:
John Wiley & Sons, Inc., 2010.
[doi:10.1002/9780470561119.socpsy001004]
- Keeble, Claire, Graham Richard Law, Stuart Barber, and
Paul D. Baxter. "
Choosing a Method to Reduce Selection Bias: A Tool for
Researchers." Open Journal of Epidemiology
05, no. 03 (2015): 155-62. [doi:10.4236/ojepi.2015.53020]
- Austin, Peter C. "
An Introduction to Propensity Score Methods for
Reducing the Effects of Confounding in Observational
Studies." Multivariate Behavioral Research
46, no. 3 (May 2011): 399-424.
[doi:10.1080/00273171.2011.568786]
- For anyone particularly interested in propensity
scoring and its use in matching or weighting strategies,
to go beyond the article above, try the following. These
selections cut across disciplines and include general
guidelines about why and how to use the methods and some
practical guides to the steps involved, plus one
interesting strong critique by King and Neilsen.
- Caliendo, Marco, and Sabine Kopeinig. "Some
Practical Guidance for the Implementation of
Propensity Score Matching." Journal of
Economic Surveys 22, no. 1 (2008): 31-72.
[doi:10.1111/j.1467-6419.2007.00527.x]
- D'Agostino, Ralph B. "Propensity
Score Methods for Bias Reduction in the Comparison
of a Treatment to a Non-Randomized Control Group."
Statistics in Medicine 17, no. 19 (1998):
2265-81.
[doi:10.1002/(SICI)1097-0258(19981015)17:19<2265::AID-SIM918>3.0.CO;2-B]
- Dehejia, Rajeev H., and Sadek Wahba. "Propensity
Score-Matching Methods for Nonexperimental Causal
Studies." The Review of Economics and
Statistics 84, no. 1 (2002): 151-61.
[doi:10.1162/003465302317331982]
- Austin, Peter C, Nathaniel Jembere, and Maria Chiu.
"Propensity
Score Matching and Complex Surveys." Statistical
Methods in Medical Research (2016).
[doi:10.1177/0962280216658920]
- Garrido, Melissa M., Amy S. Kelley, Julia Paris,
Katherine Roza, Diane E. Meier, R. Sean Morrison, and
Melissa D. Aldridge. "Methods
for Constructing and Assessing Propensity Scores."
Health Services Research 49, no. 5 (2014):
1701-20. [doi:10.1111/1475-6773.12182]; also see the
slides from a related conference presentation:
Starks, Helene, and Melissa M. Garrido. "The
Nuts and Bolts of Propensity Score Analysis." 8th
Annual Kathleen Foley Palliative Care Retreat
(2014).
- Lee, Jaehoon, and Todd D. Little. "A
Practical Guide to Propensity Score Analysis for
Applied Clinical Research." Behaviour
Research and Therapy 98, no. Supplement C
(2017/11/01/ 2017): 76-90.
[doi:10.1016/j.brat.2017.01.005]
- King, Gary, and Richard Nielsen. "Why
Propensity Scores Should Not Be Used for Matching."
(2016).
- Sovey, Allison J., and Donald P. Green. "Instrumental
Variables Estimation in Political Science: A Readers’
Guide." American Journal of Political Science
55, no. 1 (2011): 188-200.
[doi:10.1111/j.1540-5907.2010.00477.x]
- Angrist, Joshua D., and Alan B. Krueger. "Instrumental
Variables and the Search for Identification: From
Supply and Demand to Natural Experiments." Journal
of Economic Perspectives 15, no. 4 (2001): 69-85.
[doi:10.1257/jep.15.4.69]
- For anyone particularly interested in instrumental
variable techniques, to go beyond the articles above, try
the following articles that offer introductions or
overviews. Note that these tend to get technical in
places that might be inaccessible until you have
sufficient statistical skills, but they all offer less
technical insights. They show the use and
perception of instrumental variable strategies for
several disciplines.
- Linden, Ariel, and John L. Adams. "Evaluating
Disease Management Programme Effectiveness: An
Introduction to Instrumental Variables." Journal
of Evaluation in Clinical Practice 12, no. 2
(2006): 148-54.
[doi:10.1111/j.1365-2753.2006.00615.x]
- Martens, Edwin P., Wiebe R. Pestman, Anthonius de
Boer, Svetlana V. Belitser, and Olaf H. Klungel. "Instrumental
Variables: Application and Limitations." Epidemiology
17, no. 3 (2006): 260-67.
[doi:10.1097/01.ede.0000215160.88317.cb]
- Bascle, Guilhem. "Controlling
for Endogeneity with Instrumental Variables in
Strategic Management Research." Strategic
Organization 6, no. 3 (2008): 285-327.
[doi:10.1177/1476127008094339]
- Bollen, Kenneth A. "Instrumental
Variables in Sociology and the Social Sciences."
Annual Review of Sociology 38, no. 1
(2012/08/11 2012): 37-72.
[doi:10.1146/annurev-soc-081309-150141]
- Imbens, Guido W. "Instrumental
Variables: An Econometrician's Perspective."
[In en]. Statistical Science 29, no. 3
(2014/08 2014): 323-58. [doi:10.1214/14-STS480]
- Winship, Christopher, and Stephen L. Morgan. "Instrumental
Variable Estimators of Causal Effects." In Counterfactuals
and Causal Inference: Methods and Principles for
Social Research. Analytical Methods for Social
Research, 291-324. Cambridge: Cambridge University
Press, 2014. [doi:10.1017/CBO9781107587991.010]
- Baron, Reuben M., and David A. Kenny. "The
Moderator–Mediator Variable Distinction in Social
Psychological Research: Conceptual, Strategic, and
Statistical Considerations." Journal of
Personality and Social Psychology 51, no. 6 (1986):
1173-82. [doi:10.1037/0022-3514.51.6.1173]
-
- Related
Readings
12. The Art of a Critical Manuscript Review.
Every professional involved in
social research must be prepared to give and receive critical
reviews. The standards for critical reviews are high.
A good review needs to identify the important limitations of the
work under review as well as recognize its potential
contribution. The measure of weaknesses and strengths are
the standards and existing knowledge in the field. A review
that overlooks significant shortcomings or fails to notice
possible contributions is seriously flawed.
- Analytical
Task
- Analytical task: peer reviews of the thesis
proposal rough drafts.
- Each student will review three proposals by other members of the
class. These reviews should follow the standards for peer
reviewing discussed in the readings. Please do go through the
readings before writing reviews. One to two pages
is a reasonable length, although the specifics of a review may call
for a shorter or longer effort.
- The readings offer advice for reviews of manuscripts submitted to
scholarly journals. Most of the advice applies to reviews of
research proposals, adjusting for the differences in content and
circumstances. (No one appears to offer advice about
reviewing proposals, so these are as close as we can get.)
- When preparing for writing the reviews, try to be
systematic. Look at one of our outlines for the parts of a
proposal. Consider how well the proposal manages each
part. Is it clear, is it complete, does it have logical
flaws, does it seem convincing, does it leave you confused?
Identify the main shortcomings for each part, describe why it falls
short, and give your best though about how the author might go
about improving it. Once you have done this for each part or
facet of the proposal, step back, think about the proposal as a
whole, and do a similar assessment and response for the overall
impact.
- These reviews will be due one week after the submission of the
rough drafts.
- The reviewer assignments are available ⇒
(Click)
Here ⇐
- Common Readings
- Recommended
Readings
- 10
Tips From An Editor On Undertaking Academic Peer Review For
Journals
- Step By Step Guide
To Reviewing A Manuscript
- Hancock, Gregory R., and Ralph O. Mueller, eds. The
Reviewer's Guide to Quantitative Methods in the Social Sciences:
Routledge, 2010. [doi:10.4324/9780203861554] (This book, available
for download through the library, is an unusual collection of
chapters that seek to show what to look for in reviewing
quantitative articles, with each chapter stressing a different
quantitative method. I may be helpful to read the chapters
relevant to the methods you use to see what others will expect.)
- Related
Readings
13. Sources of Error. Where do we go wrong? How do we
minimize, reveal, and overcome errors?
All research efforts produce
mistakes. Research is difficult and complex. We are
flawed and error prone. A strategy that relies strictly on
avoiding mistakes will fail, because it impedes recognition and
correction of errors more than it prevents mistakes.
- Analytical
Task
- Common Readings
- Allchin, Douglas. "Error
Types." Perspectives on Science 9, no. 1
(2001): 38-58.
- Ioannidis, J. P. "Why
Most Published Research Findings Are False." PLoS
Med 2, no. 8 (Aug 2005): e124.
[doi:10.1371/journal.pmed.0020124]
- Gelman, Andrew, and Eric Loken. "The
Statistical Crisis in Science." American
Scientist 102, no. 6 (2014): 460-65.
- Simmons, Joseph P., Leif D. Nelson, and Uri Simonsohn. "False-Positive
Psychology." Psychological Science 22, no. 11
(2011): 1359-66.
- Recommended
Readings
- Ioannidis, John, and Chris Doucouliagos. "What's
to Know About the Credibility of Empirical Economics?."
Journal of Economic Surveys 27, no. 5 (2013):
997-1004. [doi:10.1111/joes.12032]
- Collaboration, Open Science. "Estimating
the Reproducibility of Psychological Science." Science
349, no. 6251 (2015). [doi:10.1126/science.aac4716]
- Gilbert, Daniel T., Gary King, Stephen Pettigrew, and
Timothy D. Wilson. "Comment
on “Estimating the Reproducibility of Psychological
Science”." Science 351, no. 6277 (2016):
1037-37. [doi:10.1126/science.aad7243]
- Anderson, Christopher J., Štěpán Bahník, Michael
Barnett-Cowan, Frank A. Bosco, Jesse Chandler,
Christopher R. Chartier, Felix Cheung, et al. "Response
to Comment on “Estimating the Reproducibility of
Psychological Science”." Science 351, no.
6277 (2016): 1037-37. [doi:10.1126/science.aad9163]
- Diekmann, Andreas. "Are
Most Published Research Findings False?." Jahrbücher
für Nationalökonomie und Statistik 231, no. 5-6
(2011): 628. [doi:10.1515/jbnst-2011-5-606]
- Stroebe, Wolfgang. "Are
Most Published Social Psychological Findings False?."
Journal of Experimental Social Psychology 66, no.
Supplement C (2016/09/01/ 2016): 134-44.
[doi:10.1016/j.jesp.2015.09.017]
-
- Related
Readings
14 & 15. Research is an ethical issue. (&
Class Presentations of Research Proposals.)
The professional who conducts
social research is a scientist, and we expect scientists to
conform to ethical codes. In academic environments (and many
non-profit research organizations), this is made salient to
social scientists by the requirement that research must be
approved by institutional review boards (IRBs), that try to
ensure we do not mistreat people. Other ethical concerns, such as
selectively presenting only research findings that support the
researcher's argument, are all too often neglected. Only when
someone is exposed as flagrantly violating rules, such as
inventing the data, do our ethics get much recognition (and even
then, social scientists commonly cope with the issue as one of
public perception). The misuse of social science in public
controversies has led many to believe that we can always
manipulate the numbers to match our argument. Every time social
scientists bend the rules - and this happens with embarrassing
regularity - we contribute to the erosion of trust in social
scientists ... by the public, by decision makers, and by other
scientists. Without integrity, we become known as modern
alchemists, practicing pseudo-science.
- Analytical
Task
- This section will take the last two class
meetings. We have two "tasks". One half the
class, chosen randomly with do the first task for the
first week and the other half will do the second
task. The following week, the tasks will be
reversed.
- Task 1: One half the class will give a brief
presentation of their research proposals.
Each presenting student will have a maximum of seven
minutes for this presentation. Three students
will be assigned to offer critical questions after the
presentation and the presenter will have five minutes
maximum to respond to those questions.
- Task 2: The other half of the class will
complete the Social and Behavioral Basic Course for human
subjects training required at NYU. This training is
contracted out to a service called CITI Program
(Collaborative Institutional Training Initiative).
You will find instructions on how to register at this
site at: http://www.nyu.edu/research/resources-and-support-offices/getting-started-withyourresearch/human-subjects-research/tutorial.html
.
- Next week, the two halves of the class will do the
reverse tasks.
- Common Readings
- De Vaus, David A. "What
Is Research Design?" In Research Design in
Social Research, 1-16. Thousand Oaks, CA: Sage, 2001.
- for Task 1 (Presentations)
- for Task 2 (Ethics)
- For
oral presentation dates and responder assignments
⇒ Click
Here ⇐
- Comments
on the Final Thesis Proposals ⇒ Click
Here ⇐ (Note that this
includes both the comments on the final draft on the
right and the earlier comments on the rough draft on the left.
For aspects that were largely unchanged in the final draft, the
comments on the left still apply.)
- Recommended
Readings
- Related
Readings
To search this page, first click here [§] to reveal all the hidden text.
(Refresh page to rehide)