SOC 220 - Sept 22

Overview

Jeff Brassard explains core tools and standards in quantitative sociology, from Likert-scale surveys and coding open-text responses to face, concurrent, construct, and convergent validity. He connects reliable measurement to causal inference, representative sampling, and replication, then has students practice operationalizing social concepts before outlining the two parts of Mini Assignment 2.

Key takeaways

Open-ended survey answers can be made quantifiable through coding, but codes need clear rules and mutually exclusive, exhaustive categories; consistent rules help multiple coders achieve intercoder reliability.
Reliability means a measure produces consistent results, while validity means it measures the intended concept; reliability alone cannot prove validity.
Face validity is the surface-level test students must justify for their assignment measures: the proposed indicator should plausibly match the concept.
A representative sample supports generalization only to the population it was drawn from, and random selection improves—but does not guarantee—representativeness.
Replication can overturn a result’s apparent generality: a more diverse replication of the marshmallow test found a weaker link between delaying gratification and later outcomes.
Operational definitions should state concretely how a study will measure a variable, and the indicator must avoid capturing unrelated activity, such as counting Netflix time as social media use.

Chapters

0:00 Midterm Location and St. Joseph’s College Campus History

The midterm is scheduled for October 15 in Education South, room 129, rather than the usual classroom.
Brassard identifies St. Joseph’s College’s September 22, 1926 cornerstone as a campus-history landmark.
The lecture opens with course logistics before returning to quantitative research methods.

4:10 Quantitative Research Tools: Surveys and Existing Data

Questionnaires and surveys are presented as sociology’s most common quantitative tools.
Secondary data analysis uses datasets collected by other researchers and can speed up graduate research.
Brassard describes researchers who finish PhDs in roughly four and a half years by analyzing existing data.

7:30 Likert Scales and Open-Ended Survey Responses

Likert items commonly ask respondents to select from strongly agree to strongly disagree.
Frequency scales such as always, sometimes, seldom, and never are another closed-response format.
Open-ended survey questions let respondents answer in their own words, often through a text box.

9:30 Coding Open Text into Quantifiable Categories

Researchers code unstructured answers by assigning labels and grouping similar statements into categories.
In the breakfast example, comments about serving Froot Loops could be coded for convenience, ease, or child preference.
Codes should be mutually exclusive and exhaustive, with clear application rules to support intercoder reliability.

14:30 Face Validity: Does the Measure Make Sense on Its Face?

Face validity asks whether a measure appears, at first glance, to capture the concept it claims to measure.
Counting how often students go drinking is a poor face-valid measure of commitment to studies.
Library hours or class attendance are more plausible surface-level measures of study commitment, though neither is perfect.

19:30 Concurrent Validity: Comparing Measures Taken at the Same Time

Concurrent validity concerns whether a measure agrees with a relevant, already-established assessment taken at the same time.
Student evaluations of teaching could be compared with a department chair’s professional assessment.
For student commitment, study time and GPA could be compared; a lack of expected association would raise validity concerns.

24:20 Construct and Convergent Validity: Theory and Established Scales

Construct validity is supported when measured relationships match what a researcher’s theory predicts.
Convergent validity tests a new measure against an established measure of the same concept using a different technique.
Brassard’s hypothetical loneliness scale could be compared with the 20-question UCLA Loneliness Index; a strong correlation would support the new scale.

31:00 How Validity Types Differ—and When a Measure Misses Its Construct

Face validity is a basic surface-level check; concurrent, construct, and convergent validity each assess different kinds of evidence.
A measure can consistently capture the wrong concept, such as treating Sunday 9 a.m. departures from home as a measure of religiosity when many people are going to brunch.
Brassard says students must justify face validity for their assignment measures, while other validity types may appear as definitions or examples on the midterm.

35:10 Reliability Is Necessary for Validity, but Not Sufficient

A measure producing inconsistent results is not valid; inconsistent data collection can also make results unusable.
A measure can be reliable but invalid if it consistently captures something other than the intended concept.
The research goal is both reliability and validity: measuring the intended construct consistently.

38:22 Quantitative Research Seeks Measurement and Causal Explanations

Quantitative research turns social phenomena into numbers so researchers can compare concepts and identify correlations.
Correlation does not equal causation: researchers must establish which variable causes the other and consider whether the direction could be reversed.
Brassard notes that quantitative research can struggle to establish causal direction and previews mixed methods as a later course topic.

42:57 Generalization Depends on Population and Representative Sampling

Quantitative research often aims to generalize from a sample to the full population of interest.
A representative sample can support generalization only to the population from which it was drawn; an Alberta poll does not represent all Canadians.
Random probability sampling improves the chance of representativeness but does not eliminate sampling bias; declining mail and phone survey response rates complicate recruitment.

46:27 Replication Tests Findings: The Marshmallow Study Revisited

Replication lets researchers check whether results recur and can expose bias or routine methodological errors.
The original marshmallow test linked children’s ability to delay gratification with later life outcomes.
A later replication using a more diverse sample found a weaker relationship, prompting researchers to reconsider how family background, social class, and race shaped the original result.

49:41 Testing Operationalizations: Social Class, Motivation, Health, and Intelligence

“Do you feel rich or poor?” is too subjective to measure social class well; household income better accounts for shared resources, such as a non-earning partner in a $250,000 household.
Counting daily Canvas opens does not establish academic motivation; the count records access, not what a student does on the platform.
A yes-or-no question about being healthy leaves “healthy” undefined; possible indicators include BMI, blood work, exercise, heart rate, medical conditions, or sleep.
GPA is a weak proxy for intelligence because it measures academic performance and grading standards differ across fields such as fine arts and engineering.

1:01:20 Writing Operational Definitions for Study Effort and Social Media Use

Possible indicators of study effort include weekly study hours, number of study sessions, completed reading pages, or a self-rated Likert scale.
An operational definition should use an active statement describing what the study will do, such as asking participants how much time they spent studying.
Social media use could be measured through minutes on social platforms, daily app opens, regularly used platforms, or posts in the past 24 hours.
Total screen time alone may include Netflix or other non-social-media activity, so the indicator must match the intended concept.

1:07:30 Operationalizing Campus Belonging and Financial Stress

Campus belonging could be indicated by a self-rating, campus clubs, close friends at the university, or monthly event attendance.
A concise operational definition might measure belonging by asking survey questions about students’ sense of connection to campus.
Financial stress indicators include self-rated stress, money worries during the past week, funds remaining after monthly expenses, and weekly paid-work hours.
Relationship satisfaction offers another example: possible indicators include a satisfaction Likert scale, weekly conflicts, daily communication, and relationship length in months.

1:12:09 Mini Assignment 2: Consent Letter, Variables, and Face Validity

Part one requires using the provided template to write an information and consent letter for an imagined research project, addressing its purpose, risks, benefits, and method.
Students should complete every template section; a hypothetical supervisor name can be changed or left in place, since the content matters more.
For part two, students revise or replace their Mini Assignment 1 quantitative research question and identify its variables.
Each variable needs a nominal definition, an active operational definition, and an explanation of why the measure has face validity; the assignment is due Sunday night.

Summary, takeaways, and chapters were generated by AI from the video's transcript and may contain errors. The video belongs to its creator, Jeff Brassard.