CLASS 1

FOUNDATION OF SOCIAL DATA ANALYSIS

SOWK 2144 Introduction to Social Data Analysis
Yu-Chih Chen, PhD
CONTENTS
Part 0: Getting to know me and you!
Part 1: Course overview
Part 2: Basic concepts of data analysis
Part 3: For this course
Part 4: Homework
PART 1
Course overview
THE CAUSE OF CHOLERA?
Does cholera airborne or what?
THE CAUSE OF DEATH IN CRIMEAN WAR?
How to reduce mortality rates
for soldiers in the battlefield?
 Asking
questions
ABILITY AND SKILLS Data extraction and
management
Able to construct questions for social inquiry
Able to identify and appraise data sets and Conduct data/
data sources statistical analyses
Able to extract meaningful variables for
analyses
Perform data/statistical analyses using a Coding skills
correct and scientific approach
Presenting results from data and offering Data presentation
interpretation
Disseminating findings and interpretation
COURSE OBJECTIVES
To obtain an understanding of using quantitative methods to analyze social
data and interpret the results appropriately
Use a statistical approach to test simple hypotheses to help answer social-
related questions
Ethically use, report, and deliver the results effectively and clearly
Construct meaningful and readable charts, tables, and figures of appropriate
data
Prepare written, oral, and visual reports for different audiences using
appropriate and straightforward (statistical) language
COURSE SCHEDULE
Class Date Topic
1 Sep. 4 Foundation of social data analysis: the statistical approach
2 Sep. 11 Measurement, scale, and graphical presentation of data
Sep. 18 [Public holiday, no class]
3 Sep. 25 Presenting data using descriptive analyses
4 Oct. 2 Continuous measure by categorical groups (two groups)
5 Oct. 9 Continuous measure by categorical groups (three or more groups) (Quiz 1)
6 Oct. 16 Correlation between two continuous measures (class continues during the reading week)
7 Oct. 23 (In-class midterm quiz)
8 Oct. 30 Explanation and prediction models using regression approach (continuous data)
9 Nov. 6 Categorical data analyses with non-parametric tests (Quiz 2)
10 Nov. 13 Project consultation via Zoom*
Nov. 20 No class; for project completion (compensation for the reading week)
11 Nov. 27 Project presentation
CLASS ACTIVITIES
Lecture
Participations
Readings
Lab exercise (software: R)
Formal assessments
 In-class quiz
 In-class midterm quiz
 Final project (presentation and paper)
CLASS ACTIVITIES: IN-CLASS QUIZ (30%)
Two in-class quizzes are administered in two sessions; each quiz may take approximately 20-
30 minutes.
The in-class questions aim to encourage class participation and deepen understanding of key
concepts. These two quiz tests will be tested using a paper-based format. Questions may
include different forms, such as (a) multiple-choice; (b) true/false question; (c) cloze test/fill-in
test.
Quiz 1 (15%) will include contents covered from class 1 to class 5. This test is scheduled on
Oct. 9.
Quiz 2 (15%) will include contents covered from class 1 to class 9. This test is scheduled on
Nov. 6.
Make-up quizzes (20% score reduction) will be offered for those who can’t attend the quiz,
except those with legit reasons (see late submission policy)
CLASS ACTIVITIES: IN-CLASS MIDTERM QUIZ (30%)
The mid-term quiz will be held on Oct. 23. Students will be asked to answer questions
on key concepts covered in the first 6 lectures (Class 1-6). The questions include R
codes related to:
 Data management (e.g., merge, select, and others)
 Data analyses (e.g., statistical models)

You will be given R data set(s) and a paper-based answer sheet for completing R-
related questions in 3 hours (9:30 am – 12:30 pm). Students must put their answers
(generated from R programs) on the answer sheet and submit them to the instructor.
Please note that no time extension will be given for students who attend the quiz late
Required equipment: your laptop and related devices (mouse, extension cord, etc.)
and a pen/pencil. Ensure that your laptop lasts three or more hours if no electronic
outlets are available.
CLASS ACTIVITIES: IN-CLASS MIDTERM QUIZ (30%)
Students are required to attend this mid-term quiz. If the students cannot attend the
quiz, a make-up quiz with an 80% grade discount will be arranged.
However, those with reasons related to family and health emergencies or with
University approved leave of absence (see examples in the late submission policy)
will receive a make-up quiz with no discount in grades.
Please provide proof of evidence by Oct. 25 (5:00 pm) to receive a make-up quiz.
The 3-hour make-up quiz will be arranged on Oct. 28 (Monday) anytime from 1:30-
5:30 pm.
This is an open-book quiz. Students can use course materials (e.g., course PPTs and R
lab) when completing the mid-term quiz. Using materials other than course materials
is considered academic misconduct; students will be disqualified and receive 0 marks
on the mid-term quiz.
CLASS ACTIVITIES: FINAL PROJECT (40%)
Each group may house up to 5* students (# of students will be finalized after the
add-drop period when having clear enrollment statistics)
The final individual project has two parts: (1) presentation (15%) and (2) report
(25%). Each project will choose a specific topic of interest and develop a high-
quality peer-review journal-style report, and at least one statistical analysis is
multiple regression
You will need to identify a data set (either cross-sectional [preferred] or longitudinal
data) that fits into your interests.
 A cross-sectional data set means the data are collected at a one-time point
 A longitudinal data set indicates respondents would be visited/collected at many time points.
PROJECT CONSULTATION (ZOOM; ON DEMAND)
Each group can discuss their project on Nov. 13 (project consultation, on demand) via
Zoom. To facilitate discussion, students are encouraged to present the following things:
 The proposed project title or topic of interest
 Study question(s)
 A brief introduction of the selected data set, including data collection methods, measures, sample size
 Present basic descriptive statistics for the key measures
 Present bivariate or multivariate tests

Each group is reserved with 15-20 minutes for project consultation. An online
schedule system will be released for groups to register on November 6-7.
Please note that project consultation is not mandatory and does not require all
group members to attend the consultation. However, those who attend the project
consultation should communicate effectively with those absent regarding contents
discussed during the consultation
CLASS ACTIVITIES: FINAL PROJECT
In the presentation and report it should cover the following areas:
Background
Research questions or objectives/aims
Methods (data, measures, variables, statistical analysis)
Results/findings
Discussion/implications
CLASS ACTIVITIES: FINAL PROJECT
Presentation (15%)
Each project should be presented in class. The presentation is limited to 15 minutes,
followed by Q & A for up to 5 minutes. A discussant will be assigned for each
project; the presenter is responsible for answering questions raised by the discussant
and students.
A PowerPoint presentation is required. It can be a poster-format (1-page) or slide-
format. PowerPoint should be submitted to Moodle before 11:59 pm on Nov. 26 (i.e.,
the day before the presentation).
Upon finishing your presentation slide, please upload it on Moodle (beware of due
time) and use the following format for your filename: GroupX_Presentation. For
example, Group1_Presentation.
CLASS ACTIVITIES: FINAL PROJECT
Final report (25%)
Students are required to write a final report on their selected topics. The page limit is 15-
pages (main text), excluding references. The report must be written in an academic style and
tone in accordance with American Psychological Association (APA) 7th ed. guidelines.
https://owl.english.purdue.edu/owl/.
The report should follow this format: A4 size; Times New Roman 12 Font; 1-inch (2.54 cm)
margin; in double spaced. This report will be submitted through the Turnitin system on
Moodle.
The final report is due at 5:00 pm on Dec. 4. The late-submission policy applies in this final
report.
Upon finishing your presentation slide, please upload it on Moodle (beware of due time) and
use the following format for your filename: GroupX_Report. For example, Group1_Report
LATE SUBMISSION POLICY
Late for % of marks to be deducted
Past due time but within the same day
5%
of the due date
1 day 10%
2 days 20%
3 days 30%
4 days 40%
5 days 50%
6 days 60%
7 days 70%
More than 7 days No mark will be given
LATE SUBMISSION POLICY
The instructor will accept late assignments for an extreme emergency, such as
death in the family, University approved leave of absence, or health-related
issues.
Health-related issues must be documented with a doctor's certificate attesting
to the student’s ill health. Such evidence mentioned above (family and health
emergency and University approved leave of absence) should be submitted
within seven (7) days.
Except under extremely unusual circumstances, papers must be turned in on
time. Inability to finish or submit assignment(s) due to busy schedules, personal
negligence or carelessness, or other reasons does not constitute sufficient
justification as mentioned above, and a late submission policy will be applied.
USE OF GENERATIVE AI
According to the assessment guidelines on Generative AI from the Teaching and Learning
Innovation Centre (pp. 16), while citing AI generated language models like ChatGPT, detailed
information about the model’s specific version, its training data, and the responsible
organization should be provided.
For in-text citations, it is recommended to include the name of the organization, such as
OpenAI, and the year of publication. The language model’s title (for instance, ChatGPT), its
version number, the responsible organization (like Open AI), and the URL to access the model
should all be included in the reference list entry.
In order to maintain clients’ confidential data and in line with the ethical practice of the
profession and the requirements of the personal data (Privacy) Ordinance, 1995, which are
reiterated in the social work fieldwork placement handbook (pp.16), it is prohibited to use the
AI generated language models for fieldwork courses.
USE OF GENERATIVE AI
HKU has an AI detector for all essay submissions. We will contact you directly for
quality and content assurance if the AI index of your work is significantly higher than
others.
For example, if another student’s work has an AI index of 20% and yours is 40%, you
will be contacted.
 GRADE DESCRIPTION
Grade Grade standard Description
A+, A, A- Excellent Demonstrate evidence of original thought, strong analytical and critical abilities as well
as a thorough grasp of the topic from background reading and analysis; should
demonstrate excellent organizational, rhetorical and presentational skills.
B+, B, B- Good Demonstrate evidence of critical and analytical thinking but not necessarily original in
their thinking; show adequate grasp of the topic from background reading and
analysis; should demonstrate strong organizational, rhetorical and presentational skills.
C+, C, C- Satisfactory Demonstrate evidence of a reasonable grasp of their subject but most of their
information is derivative, with rather little evidence of critical thinking; should
demonstrate fair organizational, rhetorical and presentational skills.
D+, D Pass Demonstrate evidence of being able to assemble the bare minimum of information,
poorly digested and not very well organized in presentation. There is no evidence of
critical thinking
F Fail Demonstrate evidence of poor knowledge and understanding of the subject, a lack of
coherence and organization, and answers are largely irrelevant. Work fails to reach
degree level.
TEXTBOOK
Salkind, N. J. & Frey, B. B. (2020).
Statistics for People Who (Think
They) Hate Statistics. Thousand Oaks,
CA: Sage.

We will use this book throughout the

semester.
SUGGESTED RESOURCES
[Statistics]
 Bors, D. (2018). Data analysis for the social sciences: Integrating theory and practice. Thousand Oaks,
CA: Sage.

[R textbook]
 Dalgaard, P. (2008). Introductory statistics with R (2nd ed.). New York, NY: Springer.
 Schumacker, R. & Tomek, S. (2013). Understanding statistics using R. New York, NY: Springer.

[R examples]
 Quick-R: https://www.statmethods.net/
 UCLA offers many hands-on tutorials on R program and many other popular statistical packages, e.g.,
STATA, SAS, and SPSS). Interested students are encouraged to explore those resources using link
below: https://stats.idre.ucla.edu/r/
EXAMPLES OF PUBLIC AVAILABLE DATA SETS
[Data archive]
 ICPSR (US): https://www.icpsr.umich.edu/web/pages/ICPSR/index.html
 SRDA (Taiwan): https://srda.sinica.edu.tw/index_en.php

[Education] Programme for International Student Assessment (PISA):

https://www.oecd.org/pisa/data/

[Family] Panel Study of Family Dynamics (PSFD): https://psfd.sinica.edu.tw/web/

[Aging] Gateway to Global Aging Data: https://g2aging.org/
[Health] WHO Study on Global Aging and Adult Health (SAGE)
https://www.who.int/healthinfo/sage/en/
OFFICE HOUR AND EMAIL POLICY
Instructor: [email protected]
 Office: Rm. 505, The Jockey Club Tower, Centennial Campus, The University of Hong Kong
 Office hour: Tuesday 11 am – 12:00 pm. Please email first to make an appointment

Email policy
 I will not read emails during weekend.
 During the working days (i.e., Monday to Friday), I will reply to your email in two working
days. If not, please send a follow-up email.
PART 2
Basic concepts of data
analysis
A ONE-MINUTE HISTORY
OF STATISTICS
Dating back to the 17th century, scientists and
mathematicians needed to develop tools to
describe data sets. This is when the use of
descriptive statistics began.
Most of the basic statistical procedures were
developed and used in the fields of agriculture,
astronomy, and politics. Application to human
behavior came much later.
Although data are collected in many different
ways across a variety of fields, all statisticians
use the same basic techniques to answer
different questions.
STATISTICS: WHAT IT IS?
A set of tools and techniques used for describing, organizing, and interpreting
information or data.
Two major branches of statistics: descriptive and inferential
What Are Descriptive Statistics?
 Used to organize and describe the characteristics of a collection of data.
 The collection of data is sometimes called a data set or just data.
 Makes describing more than a few people or things much easier.

What Are Inferential Statistics?

 Inferential statistics are often the next step after collecting and summarizing data.
 Used to make inferences from a smaller group of data.
 Works best when the sample is representative of the population.

Both types of statistics need data for analyses

EXAMPLE
Questions:
 Descriptive or inferential?
 How will you describe the
data?
 What are the patterns that
can be found?
HOW WE COLLECT DATA
1. Survey
A survey typically (but not always) collects data by asking
questions (in person or other means) or providing questionnaires
to study participants (as a printout or online).
 Opinion poll
 Social survey (e.g., American Community Survey; Panel Study of Income
Dynamic, etc.)

Ideally, a survey is conducted in a way which makes the chosen

sample representative of the population of interest
 Representativeness by randomness
 If not random, we may run into an issue called “self-selection”
 Other biases, such as recall bias, social desirability, may also exist
HOW WE COLLECT DATA
2. Experiment
Experimental data is obtained in “controlled” settings. This can mean many things, but
essentially it is data which is generated by the researcher with full control over one or
many variables of interest.
We do experiment because we want to draw a clearer causal effect, it involves
 Randomise assignment
 Single-blinded or double blinded

Very costly and expensive in terms of time and money

HOW WE COLLECT DATA
3. Observation
Researchers use a “check list” to document their
observation
 Simple observation: collects simple numbers. However, it doesn’t
account for why those numbers might be
 Behavioural observation: it observes and interprets people’s
behavior
Observation is ideal for situations in which nonverbal
communication is important for complete research.
Observation provides a more reliable measurement of
actual behavior than self-reported metrics.
“Hawthorne effect” Individuals modify an aspect of their
behavior in response to their awareness of being
observed
HOW WE COLLECT DATA
4. Using existing records
This is also called secondary data, meaning we use data that have been collected by
someone. This approach is the focus of this course
The advantages of using secondary data include: (1) highly available; (2) less cost
and (3) time saving
The disadvantages include: (1) some data may be outdated; (2) may not have the
measures you would like to test
DISCUSSION
Consider the research questions of describing parents’ attitudes towards Covid
vaccination for their kids, and whether the attitude differs by parent’s gender and
age.
(a) Which data collection method is the most suitable one to answer the above
questions: survey, experiment, observation, or secondary data?
(b) How would you capture the attitudes towards vaccination?
(c) Which variables are needed to answer all the above questions? Describe the scale
of each of them.
(d) Reflect on what an appropriate data set would look like. Now, given this data set,
try to write down the above research questions as precisely as possible.
THE NATURE OF SOCIAL RESEARCH
Social scientists attempt to explain and predict human behavior using social data to
answer questions like:
1. Do older people going out more often compared to their younger counterpart?
2. Do men play more TV games than women?
3. Do people live in HK island have better health care access than those who live in remote
islands?
4. Do people get fat more if they have friends who are obese?
5. Do people live in urban areas have higher life expectancy than those in rural area?
 THE NATURE OF SOCIAL RESEARCH
These characteristics that describe human behavior (and are measurable) are called
“variables”
Questions Variables
Do older people going out more often compared to Age (younger vs. older in age)
their younger counterpart? Going out (frequency)
Do men play more TV games than women? Gender (males vs. females)
Play TV games (time allocation or duration)
Do people live in HK island have better health care Location (HK island vs. remote islands)
access than those who live in remote islands? Proximity to hospital (travel time)
Do people get fat more if they have more friends who Obese friends (#)
are obese? Weight (changes)
Do people live in urban areas have higher life Area (urban vs. rural)
expectancy than those in rural area? Life expectancy (years)
THE NATURE OF SOCIAL RESEARCH
Variables mean something can vary, but not all human characteristics vary. Let’s do
some exercise together
 Socioeconomic status (SES)
 Gender
 Race / ethnicity
 Income
 Age
 Depressive symptoms
 Satisfaction
 Trust to government
THE NATURE OF SOCIAL RESEARCH
Unit of observation: individual vs. aggregate
1. Do older people going out more often compared to their younger counterpart?
2. Do males play more TV games than females?
3. Do people live in HK island have better health care access than those who live in remote
islands?
4. Do people get fat more if they have friends who are obese?
5. Do people live in urban areas have higher life expectancy than those in rural area?

Which one is individual, and which one is aggregate?

 THE NATURE OF SOCIAL RESEARCH

Hypothesis: statement of the relationships between two or more variables (discussed later)
 Independent variable (cause), or called “predictor variable”
 Dependent variable (consequence), or called “outcome variable”
Questions Independent variable Dependent variable
Do older people going out more often compared to their younger Age Frequency of going
counterpart? out
Do men play more TV games than women? Gender TV games
Do people live in HK island have better health care access than Location Health access
those who live in remote islands?
Do people get fat more if they have more friends who are obese? Obese friends Weight
Do people live in urban areas have higher life expectancy than Area Life expectancy
those in rural area?
THE NATURE OF SOCIAL RESEARCH
When using social data to answer questions we need to consider:
 Variables (predictor and outcome variables)
 Unit of observation
 Hypothesis

Social scientists use social data to answer a wide range of investigations, and the
most frequent way to do so is to use statistical method
THE STAGE OF SOCIAL DATA ANALYSIS
Initial observation
Secure data
(Study question)

Generate/apply
theory(s)

Test questions &

Identify variables
hypotheses

Collect data to test

Measure variables
theory(s)

• Describe data Present findings

Analyze data
• Fit a model (Dissemination)
THE STAGE OF SOCIAL RESEARCH
STAGE 1: INITIAL OBSERVATION
Identify the topics of interest and question(s) that need to be answered, based on
observation
 Does Tai Chi improve physical and mental health of older adults?
 Do younger people involve more in social movement than the middle aged in the Taiwan?
 Why disparity in education achievement by gender does not manifest in early stage (of schooling) but
in later stage?
 Does children care cash benefit increase the fertility rates in Hong Kong?

Action: Identify databases or datasets to answer these questions

 ICPSR or SRDA (Survey Research Data Archive)
 Taiwan Social Change Survey (via SRDA)
 Programme for International Student Assessment (PISA): https://www.oecd.org/pisa/data/
 Administrative data websites (e.g., Social Welfare Department, HK Census Bureau)
THE STAGE OF SOCIAL RESEARCH
STAGE 2: GENERATE/APPLY THEORY
Theory is a statement that offers to explain phenomena (we observed), indicating the
relationships that connect key constructs of interest
We often seek certain theoretical or empirical evidence to construct the theory (i.e.,
application of theories)
Question: Why disparity in education achievement by gender does not manifest in
early stage (of schooling) but in later stage?
Let’s discuss:
What are the key constructs?
What (factors) makes the differences? Who is the predictor, and who is the outcome?
What are the connections across the factors?
THE STAGE OF SOCIAL RESEARCH
STAGE 3: TEST QUESTIONS & HYPOTHESES
Generate “testable” research questions and Hypotheses offer the relationships between
hypotheses two or more variables, with “proper
direction”
How to make them testable? We have to
identify variables
Variables are testable form of vague and Hypothesis A: In high school, males have higher
abstract constructs math score than females
Question: Why disparity in education
achievement by gender does not manifest in
early stage (of schooling) but in later stage? Hypothesis B: In high school, there is a
difference in math score by gender
Variables
 Gender: males/females
 Education achievement: math score, GPA, other tests
Which one is a “good” hypothesis? A or B
 Stages of schooling: primary, secondary, and higher-
education
 Other variables…

Action: Identify variables of interest from the selected data set

THE STAGE OF SOCIAL RESEARCH
STAGE 4: COLLECT DATA
Understand your variables
 Independent / predictor variable
 Dependent / outcome variable

The measurement or scale of the variables

 Categorical or continuous data (will talk this later)
 Why we need to know my data types? The nature of data will guide the analyses, and if we don’t pay
attention, the results from the analyses may be wrong or biased (i.e., GIGO)

Do our data valid and reliable?

 Validity: whether an instrument really measures what it sets out to measure
 Reliability: whether an instrument can be interpreted consistently across different occasions
THE STAGE OF SOCIAL RESEARCH
STAGE 5: ANALYZE DATA
After we have carefully constructed our research questions and hypotheses (based on
theoretical framework), identify proper variables and understand their measurement,
we can analyse our data
There are different ways to analyse data
 Univariate analyses using descriptive and graphical approach (for only one variable) Descriptive statistics
 Bivariate analyses (for two variables)
 Multivariable analyses (for three or more variables) Inferential statistics

We will use our whole semester on talking social data analyses!

 Three classes for univariate analysis (e.g., central tendency/dispersion/graphical presentation)
 Three classes for bivariate analysis (e.g., t-test, ANOVA, correlation, non-parametric tests)
 Three classes for multivariate analysis (e.g., OLS regression)
PART 3
For this course
THE FUNCTION OF STATISTICS
We often use “number” to “quantify” nominal, ordinal, and interval/ratio data, and
we can use statistics as a tool to analyze these numbers
This course we will use statistical approach to analyze the data, it should be noted
this course will be taught using “English,” not “Greek.” This means:
This course is applied in nature, not theoretical
This course focuses more on the concepts of statistics, particularly on how to use them
to analyze different types of data
This course is not focused on the mathematical parts of statistics—theoretical statistics
THE FUNCTION OF STATISTICS
In this course we will use the statistical approach to learn:
 How to evaluate the data in terms of scale/measurement (will be taught next week)
 Conceptualize study or research questions (RQs) based on the data you have
 Test the hypotheses based on the RQs
 Select appropriate statistical methods to analyze the data and answer RQs
 Present your data and results in a professional way (e.g., graphics, tables, charts, etc.)
 Able to discuss and illustrate your findings
WHAT SOFTWARE IS USED FOR DATA ANALYSIS?
Common statistical software
Licensed software
 SPSS
 SAS
 Stata
 Excel

Open-sourced software
R
 MY DATA ANALYSES EXPERIENCES
My experience
 SPSS: since 2005
 SAS and Stata: since 2013. These two are my major
tools for data analysis
 R: since 2014 and I used it for very specific models
(e.g., latent, longitudinal, etc.)

I prefer to use licensed software as they are

much easier to use. However, I decided to use
R for this course because:
 R is free software, and everyone can afford it (if you
have a laptop/desktop)
SOME NOTES BEFORE USING R PROGRAM
The learning curve for R may be steep for some. Practice can always make perfect,
but you will get more familiar with it
I use the Windows OS system, but I know some of you use Mac. My codes are
constructed using the Windows system. There may be slight variations in code
between Windows and Mac, and the best way is to Google it to resolve the
differences—it is no difficult at all
HOW TO SUCCEED IN THIS COURSE?
Remember—we are not dumb.
Don’t skip lessons!
Form a study group.
Ask questions when feeling confused or stuck.
Read readings.
Do the exercises after each class is finished.
Practice, practice, and practice (especially for the R program)!!
PART 4
Homework
HOMEWORK FOR YOU
Browse websites for R and R studio (https://www.r-project.org/)
Visit the following websites
 ICPSR** https://www.icpsr.umich.edu/web/pages/ICPSR/index.html
 SRDA** https://srda.sinica.edu.tw/index_en.php
 Quick-R: https://www.statmethods.net/
 UCLA offers many hands-on tutorials on R program and many other popular statistical packages, e.g.,
STATA, SAS, and SPSS). Interested students are encouraged to explore those resources using link
below: https://stats.idre.ucla.edu/r/

Read Salkind & Frey (2020), Chapter 2 and 6