1
Social Science Research Quantitative Methods I
SYA 6315-U01 / Fall 2013
Dr. María Aysa-Lastra
Department of Global and Sociocultural Studies
T 2:00 p.m. – 5:00 p.m. / SIPA 200 (Lab)
Getting in touch
Tel: 305/348-2258
Office hours: T 12:00 -1:00 pm (or by appointment)
Location: SIPA 311
E-mail: [email protected]
Objective
This course provides a graduate-level introduction to applied statistics within the
framework of social research and analysis. The course complements the graduate
seminars in social theory and social research methods. Its objective is to present basic
conceptual and practical tools in social statistics so that—whether or not you intend to
pursue a career of doing quantitative studies—you’ll be better equipped, first, to critically
assess social (and policy) research carried out from a wide array of methodological
perspectives; and second, to make sound methodological decisions and wise
interpretations in carrying out your own research projects.
If you complete your assignments, work with your team, understand the materials
provided, read the assigned material before class and attend all lectures, by the end of the
semester, you should be able to:
Design a research project
Design a sound and viable sample
Critique a research design
Analyze and interpret data
Perform a hypothesis test
Perform and interpret multivariate analysis
Use statistical software
Books, software & supplements
The required books are:
Moore, McCabe & Craig, Introduction to the Practice of Social Statistics, 7th
ed.
Utts, Seeing Through Statistics, 3rd
ed.
Acock, A Gentle Introduction to Stata, 3rd
ed.
2
Strongly recommended as supplementary reading are:
Ragin and Amoroso, Constructing Social Research, 2nd
ed.
Hamilton, Statistics with Stata Version 10
Allison, Multiple Regression. A Primer
Fox, Regression Diagnostics
Achen, Interpreting and Using Regression
Pampel, Logistic Regression. A Primer
Associated with Moore, McCabe & Craig there are a couple of web sites (open access:
http://www.whfreeman.com/ips7e; and access code required:
http://www.yourstatsportal.com ) and a CD-ROM that contain helpful materials, not least
of which are conceptually-oriented quizzes on each chapter and “applets” that permit
hands-on, interactive learning. In addition, there is an excellent PBS-Annenberg video
series, “Against All Odds: Inside Statistics.” This series is available on the shelves in the
audio-visual section of the Green Library (5th
floor).
Software—Stata Version 12: www.stata.com. Discounted purchase via FIU/Stata
GradPlan. Stata is installed on computers in the SIPA Lab and it is available through FIU
elabs: http://elabs.fiu.edu. Read the Quick Start Guide and install Microsoft Remote
Desktop Connection Client 7.0 in your computer.
UCLA’s Academic Technology Services has created a website that makes available an
impressive set of free, downloadable materials for learning statistics in tandem with Stata
or SPSS, SAS, or specialized statistical software. The subsite
http://www.ats.ucla.edu/stat/seminars/ contains introductions to these and other statistical
software programs, including movies. The Stata subsite is located at:
http://www.ats.ucla.edu/stat/stata/.
Software, though, is just a tool. The focus of this course is learning statistics as one way
to describe and understand significant aspects of social relations.
Classroom policy, projects, exams & grades
It is assumed that students will attend all class sessions & arrive on time.
Cell phones must be turned off during class sessions, and using your lab computer
for emailing or internet browsing during that time is prohibited.
Questions, comments & discussion are enthusiastically encouraged.
Graded assignments:
o Students are responsible for all materials covered in the assigned readings
& problems, as well as all materials covered in class sessions.
3
o All graded assignments must be completed on time in order to earn a
passing grade in the course. Late assignments will not be graded.
o Moore, McCabe and Craig homework problems. These include practice
not only in solving statistical problems but in using and interpreting
statistics wisely. Hands-on practice, done virtually every day, is the only
way to learn statistics (and software) as well and as fast as we need to do
in this course. The Moore, McCabe and Craig problems are meant to
foster active learning. In that spirit, odd-numbered exercises have
answers, but not step-by-step solutions, in the back of the book. Some of
them, moreover, are pegged to the web site’s and CD-ROM’s interactive
applets, which in general should be used frequently for active learning of
concepts. Grading: pass/fail and worth 10% of the final grade.
Assignments are due on the dates indicated in the class schedule.
Present only the Stata commands, properly formatted tables & your
interpretation of the output; do not present the statistical output itself.
o One research project based on a data set to be chosen in consultation with
the instructor. Students will use Stata to conceptualize and apply the
statistical methods we will have covered, and will interpret the results as
well as explore the pro’s and con’s of statistical social research.
Grading: The project is worth 30% of the course grade. The project is due
at the start of the class session on November 26.
o Two take-home exams, which will combine statistical problems with
essays focusing on the development of sober judgment in selecting,
applying, interpreting, and critiquing statistics.
Grading: each take-home exam is worth 30% of your course grade (for a
total of 60%). Exam one is due on Oct 15 and Exam two is due on Dec 3.
Preparing for class sessions
Each class session will cover the minimum technical information that’s necessary
to learn statistics and Stata, and the maximum possible to put social statistics within the
frameworks of social research methodologies.
Regarding the basics of statistics, we’ll stick closely to Moore, McCabe and
Craig’s textbook presentation, emphasizing the broadest conceptual issues. At the start of
each session we’ll review some of the homework problems and/or the take-home exam
you will have completed. As much as possible we’ll use the problems and exams to raise
the big issues about doing social research.
4
Here’s how to prepare for each session:
Complete the assigned Moore, McCabe and Craig problems and/or the assigned
take-home exam or project.
Review the “Social Research Study Questions” and the “Statistical Methods:
Some Pro’s & Con’s” (both of which are attached to the syllabus) before the first
class session and throughout the semester. We’ll refer to them frequently.
The assigned readings from Moore, McCabe & Craig should be read before each
session, then should be thoroughly covered when doing the homework problems
(including use of the instructor-supplied, Stata-formatted text data sets).
Do the web or CD-ROM quiz corresponding to each Moore, McCabe & Craig
chapter assignment to make sure you’ve mastered the course material. The
quizzes, however, will not be graded.
Everything else: Do whatever works best for you. In my experience working in
study groups is the best way to learn statistics.
Policy on make-up examinations, assignments or performance measures
In case the student is experiencing an unexpected circumstance, he/she should
communicate to the instructor before the due date of the weekly assignment or exam and
asked for a reasonable extension. All students must take the two take-home exams.
FIU’s Policies and Codes
Florida International University is a community dedicated to generating and imparting
knowledge through excellent teaching and research, the rigorous and respectful exchange
of ideas, and community service. All students should respect the right of others to have
an equitable opportunity to learn and honestly to demonstrate the quality of their learning.
Therefore, all students are expected to adhere to a standard of academic conduct, which
demonstrates respect for themselves, their fellow students, and the educational mission of
the University. All students are deemed by the University to understand that if they are
found responsible for academic misconduct, they will be subject to the Academic
Misconduct procedures and sanctions, as outlined in the Student Handbook.
Please for additional information see: Policies on Academic Misconduct
Other Policies and Codes
Academic Calendar, please note that November 4 is the last date to drop the course with
a DR grade.
5
Schedule
Aug. 27
Week 1 Data Distributions
Required Moore, McCabe & Craig, Introduction and chapter 1
See “Social Research Study Questions” & “Statistical Methods: Some
Pro’s & Cons” attached to syllabus
Utts, chapters 1 and 2
Acock, chapters 1 and 5
Recommended Ragin & Amoroso, chapters 1 & 2
Assignments
(due Sep 3)
1) Go to the website for “ASR Manuscript Submission Information
for Authors” or the main journal in your discipline and navigate
to the “Preparation Checklist for Manuscripts”. Provide a
summary of the main items and bring an example using each
type of text citation, and bibliographic reference.
2) Provide an example of a table published in any academic
quantitative research paper of your interest.
3) Moore, McCabe & Craig problems 6th
ed: 1.12, 1.17, 1.21, 1.37,
1.46
Moore, McCabe & Craig problems 7th
ed: 1.14, 1.17, 1.21, 1.41,
1.46
4) Submit a topic and a question for your quantitative research
project.
Sep 3
Week 2 Data Distributions (continued)
Required Utts, chaps. 7, 8 and 9
Moore, McCabe & Craig, chapter 1
Acock, chapters 1 and 5
Recommended Ragin & Amoroso, chapter 3
Assignments
(due Sep 10)
1) Answer the following questions: a) Which are the characteristics
of a well-designed graph? b) Which are the most common
problems in plots, graphs and pictures? c) Which are the
characteristics of a well-designed table?
2) Moore, McCabe & Craig problems 6th
ed: 1.57, 1.58, 1.62, 1.65,
1.82, 1.83, 1.84, 1.86, 1.87, 1.88; 1.111, 1.120,1.122, 1.140,
1.141, 1.142, 1.146
Moore, McCabe & Craig problems 7th
ed: 1.69, 1.70, 1.73, 1.74,
1.80, 1.90, 1.91, 1.92; 1.119, 1.127, 1.128, 1.146, 1.147, 1.148,
1.151
3) Study questions: Why do univariate distributions matter—in
terms of substantive issues in social science and in terms of
statistics? Why should we graph a univariate distribution before
we numerically summarize it? What are the advantages of the 5-
number distribution?
6
Sep 10
Week 3 Data Relationships: Bivariate Distributions
Required Utts, chapters 10 and 11
Moore, McCabe & Craig, chapter 2
Acock, chapter 6
Recommended Ragin & Amoroso, chapters 6 and 7
Assignments
(due Sep 17)
1) Moore, McCabe & Craig problems 6th
ed: 2.6, 2.7; 2.30, 2.31,
2.48; 2.58, 2.59, 2.60, 2.61
Moore, McCabe & Craig problems 7th
ed: 2.24, 2.25; 2.47, 2.48,
2.58; 2.73, 2.74, 2.75, 2.76
2) Study questions: What can be misleading about measures of
correlation and regression, and why? When might there be a
strong bivariate relationship but a low correlation or regression
coefficient? What are the advantages of regression versus
correlation
Sep 17, Sep 24
Week 4 Data Relationships: Bivariate Distributions
Required Moore, McCabe & Craig, chapter 2
Acock, chapter 6
Recommended Utts, chapter 12
Assignments
(due Oct 1)
1) Moore, McCabe & Craig problems 6th
ed.: 2.91, 2.104; 2.111,
2.112, 2.113; 2.123, 2.126; 2.159
Moore, McCabe & Craig problems 7th
ed.: 2.105, 2.114; 2.125,
2.126, 2.127; 2.136, 2.137; 2.173
2) Study questions: What are the basic ways of establishing and
explaining causation? What critical questions must we ask
concerning the possible relationship between two variables?
Oct 1
Week 5 Producing Data
Required Moore, McCabe & Craig, chapter 3
Utts, chapter 3, 4 and 5
Acock, chapters 2 and 3
Recommended Ragin & Amoroso, chapters 4 and 5
Assignments
(due Oct 8)
1) Moore, McCabe and Craig problems 6th
ed: 3.17, 3.18; 3.52,
3.53, 3.56, 3.61, 3.65; 3.82, 3.83, 3.84; 3.120, 3.127.
Moore, McCabe and Craig problems 7th
ed: 3.17, 3.18; 3.52,
3.53, 3.58, 3.63, 3.67; 3.82, 3.83, 3.84; 3.126, 3.121.
2) Study questions: What is “statistical” significance? Give a
social-science example of how it can be different from
“practical” or “theoretical” significance? What is bias? What is
variability? Why is randomization important? Why is
comparative design important? What are the advantages of
experimental research and the reasons for these advantages?
7
What are the disadvantages of experimental research? What are
the main sampling methods? What are the advantages and
disadvantages of each sampling method? What are possible
“sampling” and “non-sampling” sources of error in surveys?
3) EXAM 1 assigned: The exam will cover the materials in
chapters 1, 2 and 3 of the Moore, McCabe and Craig textbook
and the assigned chapters from Utts.
Oct 8
Week 6 Probability & inference
Required Moore, McCabe & Craig, chapter 4
Recommended Utts, chapter 15, 16 and 17
Assignments
(due Oct 15)
1) READ CHAPTER 4
2) Study questions (OPTIONAL, for extra credit): Why is
randomization important? What is a random variable? What is
a probability distribution? Why is the reason for using the term
“expected value of random variable” instead of “mean of a
random variable”? What is the Law of Large Numbers? How
many trials are need to guarantee a mean outcome close to the
population mean? What are “independent observations” (or
events), and when can this requirement be relaxed? What is
“conditional probability,” and what would be a social-science
example?
3) EXAM 1 due: bring a hard copy of your answers at the start of
class session.
Oct 15
Week 7 Sampling distributions
Required Moore, McCabe & Craig, chapter 5
Recommended Utts, chapter 18
Assignments
(due Oct 21)
1) Moore, McCabe and Craig problems 6th
ed: 5.9, 5.10, 5.11, 5.12,
5.24, 5.26; 5.41, 5.61; 5.64.
Moore, McCabe and Craig problems 7th
ed: 5.41, 5.42, 5.43,
5.44, 5.56, 5.60, 5.80
2) Study questions: What is a sampling distribution? What is a
population distribution? What would be a social-science
example of both? What is a count and a sample proportion, and
a social-science example of each? What is the “binomial
setting,” and what is a social-science example? What is a
binomial distribution and a sampling distribution of a count?
When should we use a binomial sampling distribution? Why are
sample means used in statistical inference? What is the sampling
distribution of a sample mean, and what is a social-science
example? How do we compute the mean and standard deviation
of a sample mean? What is the Central Limit Theorem, and why
is it important? Why is it that the sample mean of an SRS from
8
a normal population has a normal distribution, and what
principle does this illustrate?
Oct 21
Week 8 Introduction to inference
Required Moore, McCabe & Craig, chapter 6
Acock, chapter 7
Recommended Utts, chapter 20, 21, 22 and 23
Assignments
(due Oct 29)
1) Moore, McCabe and Craig problems 6th
ed: 6.116, 6.120, 6.125,
6.126, 6.132.
Moore, McCabe and Craig problems 7th
ed: 6.116, 6.120, 6.125,
61.26, 6.132
2) Study questions: What is “statistical inference”? For what kinds
of data sets is it valid or invalid? What is the difference between
statistical significance and “practical” or “theoretical”
significance? What is a confidence interval, and how is it
computed? What are the data requirements for a valid
confidence interval, and how do we check these requirements?
How can we reduce a confidence interval? How do we test a
hypothesis? What is wrong with accepting a null hypothesis?
What is wrong with accepting an alternative hypothesis? What
is a P-value? What is wrong with “searching for statistical
significance”? When should statistically insignificant results be
reported and explained? What are Type I and Type II errors, and
what is an example of each?
Oct 29
Week 9 Inference for distributions
Required Moore, McCabe & Craig, chapter 7
Acock, chapter 7
Recommended Utts, chapter 20, 21, 22 and 23
Assignments
(due Nov 5)
1) Moore, McCabe and Craig problems 6th
ed: 7.113, 7.115, 7.130,
7.131, 7.132, 7.133.
2) Moore, McCabe and Craig problems 7th
ed: 7.113, 7.115, 7.134,
7.135, 7.136, 7.137.
3) Study questions: What is a t distribution, when is it used, and
how is it computed? How do t distributions differ from z
distributions, and at one point do they become more or less
identical? When do we use one-sample and two-sample t tests,
and what is a social-science example for each? What are the
data requirements for such tests? What in general are test
alternatives if the data requirements are not met?
9
Nov 5
Week 10 Inference for proportions
Required Moore, McCabe & Craig, chapter 8
Acock, chapter 7
Recommended Utts, chapter 19
Assignments
(due Nov 12)
1) Moore, McCabe and Craig problems 6th
ed: 8.57, 8.64, 8.77,
8.78, 8.79, 8.83
Moore, McCabe and Craig problems 7th
ed: 8.61, 8.74, 8.95,
8.96, 8.97, 8.101
2) Study questions: What is a social-science example for single
proportion and for two proportion inference? What are the data
requirements?
Nov 12
Week 11 Inference for two-way tables
Required Moore, McCabe & Craig, chapter 9
Acock, chapter 6
Recommended Utts, chapter 12
Assignments
(due Nov 9)
1) Moore, McCabe and Craig problems 6th
ed: 9.6, 9.7, 9.15, 9.16,
9.26, 9.36, 9.42, 9.44
Moore, McCabe and Craig problems 7th
ed: 9.26, 9.27, 9.33,
9.34, 9.42, 9.50, 9.52
2) Study questions: What are the data requirements for two-way
tables? What is a social-science example of a two-way table?
What is Simpson’s Paradox, what is the reason for it, and how
can we guard against it? What is the chi-square statistic? How
do we test a hypothesis for a two-way table?
Nov 19
Week 12 Inference for regression
Required Moore, McCabe & Craig, chapter 10
Acock, chapter 8
Recommended Utts, chapter 10
Allison book
Assignments
(Optional)
1) Moore, McCabe and Craig problems 6th
ed: 10.6, 10.7, 10.8,
10.9, 10.10, 10.39, 10.40, 10.41, 10.47, 10.48, 10.49, 10.51
Moore, McCabe and Craig problems 7th
ed: 10.6, 10.7, 10.8,
10.9, 10.10, 10.37, 10.38, 10.39, 10.45, 10.46, 10.47, 10.49
2) Study questions: What is simple linear regression? How does it
differ from correlation? When might there be a strong bivariate
relationship but a weak correlation or regression coefficient?
What are the data requirements for regression? What do the
following mean: DATA=FIT + RESIDUAL? How do we test a
hypothesis for a regression coefficient? What does it mean if the
plotted residuals of a regression model are not randomly
distributed?
10
Nov 26
Week 13 Inference for regression (continued)
Required Moore, McCabe & Craig, chapter 11
Acock, chapter 8
Recommended Utts, chapter 10
Allison book
Assignments
(Optional)
1) Study questions (OPTIONAL, for extra credit): What is the
advantage of multiple regression over simple regression?
Explain why or why not a strong/weak bivariate relationship
may not result in a strong/weak multivariate relationship,
including what this has to do with our previous reading on causal
relations?
2) Bring a hard copy of your assigned research project
3) EXAM 2 assigned
Dec 3 (Final exam week)
Exam two 1) EXAM 2 due: print a hard copy of your answers and drop it in
my mail box (SIPA 3rd
floor) by noon.
11
Social Research Study Questions
1. What is social research? What are the principal differences between social
research and other ways of representing social life?
2. What is the scientific method? What steps does the scientific method apply in
conducting social research?
3. What is a research strategy? What are the differences between research strategies
that paticularize and those that generalize? What are the potential similarities
between such strategies?
4. What is the social construction of reality? How does it pertain to the scientific
method, social research/research strategies in general, and to other ways of
representing social life—including the promises and risks of the various
approaches?
5. What are data? What are interplays between data and the social construction of
reality? Is everything worthwhile measurable?
6. What is statistics? What is the difference between descriptive statistics and
inferential statistics? How do descriptive statistics and inferential statistics
pertain to the principal kinds of research strategy?
7. What are advantages and disadvantages of using statistics in social research?
8. What are the intersections between the uses of statistics in social research and the
social construction of reality? Conversely, what are the intersections between the
“non-uses” of statistics in social research and the social construction of reality?
12
Statistical Methods: Some Pro’s & Con’s
Some advantages of using statistics
Summarizes complex data
Makes assumptions explicit
Imposes explicit standards of evidence and comparison
Raises the possibility of chance associations
Emphasizes skepticism about hypotheses and findings
Facilitates the testing of competing hypotheses and the building of theories
Permits the examination of certain questions that couldn’t otherwise be examined
Some pitfalls of using statistics
The use of statistics represents a strategic tool in the social construction of reality.
Thus its use in general must be situated in the historical/geographic context of
bureaucratization, state formation and geopolitical competition, industrial/
technological revolution, commodification, and urbanization; and the biases of
statistical premises and the tendency of the statistically inclined research
establishment to claim intellectual/policy hegemony on the basis of a “scientific
approach” must be critically examined.
The use of statistics impedes the examination of certain questions that otherwise
would be examined, and obfuscates crucial kinds of social, cultural, and political
analysis.
Theory and substantive importance must guide the use of statistics (although the
data must inform the theory as well [e.g., making sense of unanticipated
nonlinearities or outliers]).
Statistical research needs to emphasize theoretical/substantive significance and
the magnitude of relationships between variables, rather than mere “statistical
significance” as narrowly defined by mainstream statistical methodology. The
research needs to recognize the arbitrariness of institutionalized significance-test
standards and to consider alternative criteria for statistical significance.
Statistical research needs to test a study’s findings, not just against its null
hypothesis but also against competing theories/hypotheses with the objective of
long-term theory building.
We need to use statistics wisely as one of many tools in social research.
13
Reminders
How were the numbers produced—in the sense of culture and power, and
according to the (cultural) rules of scientific method?
Is the sample random and representative of the population? Insofar as this is not
true, then the use of inferential statistics is invalid.
What is the shape, center, and spread of the distribution? Are there outliers?
Do the numbers make sense? (adapted from Moore, The Basic Practice of
Statistics):
o What’s the explicit or implicit agenda behind them?
o Is any essential information left out?
o Are the numbers consistent?
o Are the numbers plausible, including are they too good to be true?
o Is the math correct?
o What do the numbers signifiy about the social relations being studied?
Always take “outlying” observations, “non-significant” findings, and otherwise
“contrary” findings seriously: What insights do they potentially convey about the
social relations that you’re studying, and possibly about social relations more
generally?
You’ve estimated something’s magnitude or likelihood. Don’t lose sight of
uncertainty: What’s the thing’s estimated range of magnitude or likelihood? What
does this range imply about the social relations being examined?
Are all worthwhile things measurable? What do your conclusions imply about the
social relations and public policies that you’re studying, and perhaps about social
relations and public policies in other spheres?
Benchmarks for assessing the usefulness of
any application of social statistics
Does the use of statistical methods in any given instance notably improve our
intellectual understanding of social relations and public policies?
In any given instance, what insights does the use of statistical methods provide (or
not) in comparison with insights provided by other methods of social research,
and in comparison with insights provided by other ways of interpreting the world?
A strategic point in interpreting & summarizing your study’s results
What are the wider, comparative ramifications of your study for the
understanding of social relations and social policy/political practice?