Sociology Notes PDF

Lecture 1 Objectives of the Course - Look at the course outlines: 1. Basic logics of doing social research → a better consumer of research results 2. Steps in conducting a social investigation → to conduct a small research paper How to Achieve the Objectives a) Course M...

Lecture 1 Objectives of the Course - Look at the course outlines: 1. Basic logics of doing social research → a better consumer of research results 2. Steps in conducting a social investigation → to conduct a small research paper How to Achieve the Objectives a) Course Materials - Recommend Text Book (Neumann & Robson, 2014) - Recommended Text (Wagner, 2019) - Class/Lab Participation (10%) - Assignment 1 (25%) - Assignment 2 (25%) - Tests: 1. Test 1 (30%) 2. Test 2 (10%) b) Proff’s responsibilities - Class and lab lectures - Make sure you are following the topics: 1. Shall ask you questions 2. Class discussion - Guide you in your assignments c) Your Responsibilities: - Regular, active, prepared class and & lab participation - Accomplishment of the course requirements WEEKLY - Seek help from me or TA 1. Do WEEKLY the steps in your assignment - I need your comments on my lectures and teaching Chapter 1 (Chapter One) Introduction Introduction - We live in a complex world, with various social problems or issues such as: - Unemployment, school dropouts, political movements, sexual harassment, health inequality, out-of-wedlock births, low fertility - Personal or social problems? - People usually ask questions about the social problems. E.g., “school dropouts” in Canada Research Problem & Question (e.g.) - Research Problem: “Every year we have a significant portion of students leaving school forever.” - Descriptive research questions - “What” is the proportion of school dropouts in Canada? - “Who” are the school dropouts? - Explanatory research questions? - “Why” does it happen in Canada? - What are the causes? - Predictive (influencing) research questions - “How” can the school dropouts be prevented? - How do we find the answers to these questions? Sources of Knowledge - Non-scientific sources of knowledge” 1. Authority: - Person in a position of authority (e.g. Prof X believes that poor parenting → school dropouts in Canada) - Authoritative publication (e.g., An article published in the American Sociological Review says that family poverty → school dropouts in Canada) 2. Tradition: “It is the way things have always been.” 3. Common Sense: “It makes sense that lazy students usually dropout of school 4. Media: TV shows, movies, newspaper, etc. - Goals: entertainment and giving limited information based on their interests. - E.g., ignoring shortage of budget and recognizing teachers’ performance as a cause of dropouts in State-Funded TV program 5. Personal Experience: “seeing is believing” - E.g. “dropouts are aggressive students like 4 of my school friends” this has four limitations - Overgeneralization (not all aggressive students are dropouts) - Selective observation (supporting my belief by selected cases) - Premature closure (jump to conclusions) - Halo effect: rubs off overgeneralized reputation and prestige onto other areas. E.g.: accepting argument of an article about school dropouts because its author is from Toronto University, regardless of its scientific merit. Scientific Source of Knowledge - Scientific Knowledge:  - Research Findings generated through a scientific social research method, with much less errors & biases and high accuracy - What is a social research? - Finding answer(s) to research question(s) about a social problem, or an issue of interest, using a scientific approach. - What is science & a scientific approach? Science - “Science is a way to produce knowledge” by using specialized techniques to gather data (i.e., empirical evidence) - Data are collected carefully and systematically through scientists’ sens (touch, sight, hearing, smell, and taste) - Free of personal judgments to support or reject theories - Subjective concepts (e.g., attitudes, power, intelligence, alienation) - Sense hardly work for collecting data about subjective concepts - So, indirect techniques of data collection were developed Steps in the Social Research - A social research proceeds in 10 steps - You practice most of these steps in the assignments - The linear ten-steps research process is oversimplified for educational purposes - Need to go back and forth several times to conduct a good research 1) Select a Research Problem (a Topic) 2) Develop Research Question (Focus the Topic) 3) Review Literature & Theories (Past Research and Theories) 4) Develop Hypothesis (possible answers to RQ) 5) Conceptualize & Operationalize Concepts (measurement) 6) Select Method of Data Collection (Research Method) 7) Select Cases or Respondents (Sampling Method) 8) Gather Data (Based on the selected Research Method) 9) Analyze the Data (use of statistical methods) 10) Interpret the Data (Write Research Report) Types of Social Research - By purpose: 1. Explorative Research (“What is it?”) 2. Descriptive Research (“Who” it is?) 3. Explanatory Research (“How” or “Why are things the way they are?” - By Time Dimension 1. Cross-sectional Research  - Examines a single point in time or one time snapshot - Less costly and fast 2. Longitudinal Research: examines change over time points a. Time Series study: collects the same info over time b. Panel Study: observes the same cases over time c. Cohort study: observes the same cases sharing a similar life experience over time, e,g., marriage cohort 3. Case Study: Observes or interviews with one or a few cases deeply over time - By usage 1. Basic Research: advances knowledge (method & theory). E.g., The Protestant Ethic and Spirit of Capitalism (Weber, 1904) 2. Applied Research: usually commissioned by private or public agencies to find a solution for a specific problem - Evaluation Research: evaluates effectiveness of an implemented program (e.g., Family Planning Program) - Social Impact Assessment Research: estimates the social impact of a planned program on an issue of interest to the agency - Examples: 1) Physician Assistant → patients’ waiting time. 2) Condom distribution in school → students’ pregnancies - By Data Collection Methods: 1. Quantitative Methods - Experiments: examines causal effect of a treatment in an experimental condition - Survey: studies large population using a sample, questionnaires, and statistical methods - Content Analysis: examines information, symbols, and content of documents, movies, and so on. - Existing Statistics: analyze government reports and collected, published data 2. Qualitative Methods - Field Research: Conducts a case study over a length of time - Historical-Comparative Research: examines historical documents, compares two or more cultures or civilizations. Readings & Works for This Week - Read carefully Chapter 1 in the Text - Assignment 1: Now you can work on the following two sections in Assignment 1: 1) Research Problem (RP): after choosing a topic, write a RP 2) Research Question (RQ): write a scientific explanatory RQ, based on your review of the literature. I will talk about this further in the next class Week 2 (Chapter two) Step 3: From Research Question to Theory & Hypothesis Research Question and Theory - You saw in the last class how to turn a - General research topic into one researchable question - Now, find a logical answer for the question - Scientific knowledge is cumulative - Build your research on the previous research findings and theories → advancement of scientific knowledge - Develop a hypothetical answer (hypothesis) to your research question by - Thinking, reviewing past research and relevant theories - Collecting data without reviewing past research and theories is just like sailing without a compass and a map, made by previous sailors - Today, we learn what is a social theory & a conceptual model, used particularly in health research? - What do you understand about theory? What is a social theory? - Theory = a generalized “story of operation” of a social issue or problem, based on systematic observations of researchers and the collected data using scientific methods. So, - You can also have a theory, since you observe the problem in everyday life, and have one answer to the question - E.g.,: “what is the cause of school dropouts in Canada?” - But, your observations are NOT systematic: - Limited to a specific time, place and persons(s) - Mixed with your own value-judgments - NOT abstract enough to generalize to every time, place, or cases Components of a Theory - A theory described the relationship between two concepts (variables) based on some assumptions - E.g. : “Lack of sufficient intimate social contact among members of different racial groups” → “racial prejudice” - Three components of a theory: 1. Concepts: Intimate social contact (cause) + racial prejudice (effect) 2. Assumption: for concept of “racial prejudice”: - “People make distinctions among individuals based on their race, color, and racial characteristics”  - Itf this is not the case → concept of racial prejudice should not exist 3. Relationship: relates two concepts (variables). E.g. : “intimate social contact” with “racial prejudice” - The relationship is tested by a hypothesis, taken from the theory and stated in a probabilistic (or comparative) language. “The higher intimate social contact, the lower racial prejudice” - After developing the hypothesis. 1. Review past research findings about their relationship 2. Define the concepts and figure out how to turn into measurable variables - These two steps will be described in the next class A Research Example 1 - In a co-author paper, entitled “attitudes towards childbearing outside marriage in Canada”, I have used different social, psychological, and economic theories to generate some hypotheses. The paper is a very good example for how to use “theories” and “previous research findings” to develop some “hypotheses”. A first draft of this paper is available in - And a later draft, published in Journal of Comparative Family Studies (Erfani & Beaujot, 2009) - Research Problem (RP): - Non-marital childbearing is more likely to create female-headed families, and children born to unmarried parents are estimated to live longer years with a single mother → affecting child development (due to inadequate resources) - Research Question (RQ): - What are Canadians' attitudes towards childbearing outside marriage? (Descriptive RQ) - Why are Canadians different in their attitudes? (Explanatory RQ) - A specific RQ: Is experience of cohabitation related to attitudes toward childbearing outside of marriage? - Theoretical Framework - “A person’s attitude towards any given object is determined largely by his/her social learning, past behavior, and socio-demographic status” - Social learning theory: “Most human behaviors are learned observationally through modeling: from observing others one forms an idea of how new behaviors are performed, and on later occasions this coded information serves as a guide for action.” (Bandura 1977:22) - Hypothesis 1: “individuals whose parents cohabited or divorced vs. married are expected to have a more positive attitudes toward having children outside of marriage.” - Cognitive Consistency Theory: - “Persons who have children in cohabiting unions may experience some degree of “post-decisional dissonance” wherein their evaluation of alternatives is  inconsistent with their behavior. One way of reducing the dissonance is to magnify the acceptability of having children in a cohabiting union.” - Hypothesis 2: “Individuals who begin their first relationship as a cohabitating vs marital union are expected to hold more positive attitudes to non-marital childbearing.” - Economic Theory: - “Out-of-wedlock childbearing will be most prevalent when females are in excess supply, when females are in excess supply, when they have sufficient income to support a family on their own, and when the gains to marriage are small because male incomes are low” (Willis, 1999). - Hypothesis 3: “Women who are engaged in a full-time vs part-time or no job are expected to hold more positive attitudes towards non-marital childbearing” The Research Example 2 - RP: prevalence of induced abortion in the context where abortion is illegal and standard abortion and post-abortion health services are limited. - RQ: what factors are related to abortion use? - Theory: a person’s access to health services is determined by individual-level and community-level predisposing, enabling and need factors - Behavioral Model of Health Service Use (Andersen, 2014)  - Hypothesis: Women having unmet need for effective contraceptive (e.g., contraceptive failures) are more likely to use abortion (Erfani, 2021) Aspects of Social Theory - Many social theories, but different by direction, range, and forms of explanation - Linking a research question to a given social theory, be aware about these aspects: 1. Research approach (Direction): - Deductive approach: moves from abstract ideas and theory to empirical evidence → used for testing a theory - Inductive approach: moves from empirical evidence to generalized & abstract ideas, and theoretical statements → used for building a theory 2. Range of Theory: - Empirical Generalization: theories with the least abstract concepts and very narrow scope or range. E.g. : “Older grad students finish their program faster than younger ones” - Middle-Range Theories: Theories with more abstract concepts and broader scope. E.g.: “Health status of people is directly determined by their socioeconomic status.” 3. Explanation vs. Interpretation of social reality - Causal Explanation (Positivism): studying a causal relationship (social reality can be measured by objective facts (like natural science) → “cause” and “effect” - E.g. : “intimate social contact → racial prejudice” - E.g. : “Blackout → baby boom” - Three prerequisites for causality (each one is a necessary but insufficient condition for inferring a causality): 1. Association: cause and effect occur together (are correlated) 2. Temporal order: cause before effect 3. Elimination of plausible alternatives: X→ Y - No spuriousness → statistical control - Usefulness of longitudinal data - Interpretive Explanation: understanding “meaning” of an action, placing it within a specific social context. - Getting inside the worldview of the actor, using Verstehen or empathetic understanding → field research - E.g.: Street Corner Society (William Foote Whyte, 1955) lived in an Italian slum with people at the North end of Boston (Cornerville) as a friend → to understand their behavior and organization through a case study and as a participatory observer Readings & Works for Today - Read Chapter 2 & the research article - Read Chapter 5 for next week. We will cover chapter 3, and 4 later, it time permits. Otherwise, they will be omitted. - Assignment 1: - Research question (RQ): Continue working on RQ by revising your scientific explanatory RQ, based on your review of the literature - Research Hypothesis: you can begin working on this section, by proposing an initial statement of Research Hypothesis, and continue working on your hypothesis next week. Lecture 3 (Chapter 5) Step 4: Hypothesis and its Logical Issues Quantitative vs. Qualitative Recall from last class: - Quantitative Research (deductive approach): - Positive approach to social research: - Looking for causality, where research question tells us to explain a causal relation - Testing a theory, through a hypothesis - Collecting qualitative data (in numbers) - Qualitative Research (inductive approach): - Interpreting the “meaning” of a social reality within its context and time, where research question tells us to interpret the meaning of a social problem. - Developing a theory by observing cases through time - In your assignments, you will do a quantitative positive, social research A Good Hypothesis - Today, we learn - How to develop a good hypothesis in response to your Research Question (RQ) - What are the logical issues related to a hypothesis? - Hypothesis: a tentative, testable, value-neutral statement of relationship between two variables - What is a variable? Variable - Variable: a concept or a trait varying from case to case (e.g., gender, race, religiosity) - With different values or attributes (e.g., Gender: male/female) - Do not mix value (attributes) with variables in a hypothesis - Practice: Which one is a variable or value (attribute)? - Income, high socioeconomic status, poor, crime rate, negative attitude, married, satisfied, health status, old, inflation, fertility - Types of Variables by function: 1. Independent Variable (cause, determinant) shown by X. 2. Dependent Variable (effet, outcome) shown by Y. E.g. : The higher level of income, the better the health status. 3. Intervening variable: examines the mechanism or X → Y Examples: 1) Income → access to health services (X → Z → Y) 2) Socio-economic development → contraceptive use → fertility Hypothesis (Con’t) - Characteristics of a causal hypothesis: 1. Has at least two variables 2. Contains a causal relationship 3. Implies a clear prediction (i.e., an expected future outcome) → specifies direction of relationship 4. To be falsifiable (or testable): Practice: are there anything wrong with these hypotheses? - “Opinions on foreign aid are related to political affiliation” - Salvation of human depends on his purity - Identify the dependent and independent variables in the following situations: - You wonder whether radar detection and fine for over-speed help to reduce the number of accidents. - The tobacco shop near a school has been closed by the police. You want to know the students’ reactions. You ask each student if he/she supports or opposes the closing of the store, and how many times he/she has visited the store. - Which of the following will you consider to be a good hypotheses? Give reasons. 1. Socialization in childhood has a significant impact on adolescent gender-role identity. 2. Religiosity equals church attendance and praying 3. Actions are based in perceived costs and benefits 4. The greeted the level of education, the greater the tolerance of alternative lifestyles. 5. “Age is related to attitudes towards women’s liberation” - Types of Hypothesis: 1. Null Hypothesis (H0): NO relation between X & Y 2. Alternative or Research Hypothesis (H1): existence of a relationship between X&Y - As scientist, our aim is to REJECT H0 → limitation of science - A hypothesis is NEVER proved - If H0 is rejected then H1 is true and is SUPPORTED. Otherwise, it is false. Unit of Analysis - Unit of Analysis: refers to the unit which we collect data about, and we generalize the research to - The unit of analysis can be: 1. Individual (e.g., children, women, criminals) 2. Group (e.g., families, peer groups) 3. Organization (e.g., companies, schools, churches) 4. Social institution (e.g., religion, education)  5. Society (e.g., nations, tribes, countries) 6. Artifacts (e.g., books, movies, magazines) - What is the unit of analysis in each one? 1. Most Canadians believe in God 2. 10% of families in North Bay move within a year/ 3. Church records show that women give more money than do men 4. Number of children woman has 5. Average number of children for all countries in the world. 6. Girls watch TV more than boys 7. Deteriorating neighborhoods are more likely to organize protest marches than affluent neighborhoods. - Rule: Be clear about the unit of analysis. Once they are determined, do not switch from one to other. Otherwise, you will fall within the traps of ecological fallacy or reductionism in the analysis step Ecological Fallacy - A false reasoning resulted from a poor fit between units from which data collected and the units to which the findings are generalized - Studying groups but making inferences about the individuals - Collect data at a higher level or an aggregate unit, but generalize results to a lower level or disaggregate unit Ecological Fallacy - Example (Baker, 1988, p.99) - Research topic: causes of illegal drug use - Data: “rate of drug use” and “per capita income level” by province 1. Provinces with higher rates of drug use also have higher per capita income 2. Conclusion: ah ha! The rich people are often drug users - An example of ecological fallacy: NO empirical evidence on the incomes of the individuals who use drugs - Data collected at province level cannot be generalized to individual level - Because it could well be the case that poor people in the richer provinces more frequently use drugs Reductionism - The reverse of ecological fallacy - Collect data for a lower level, but the result to a higher level unit. - “The tendency to reduce a complex phenomena to a single cause” - It frequently occurs as data collection on individual levels are easier than higher levels Reductionism - Examples (Baker, p.100) 1. “Nazism took over Germany because German boys hated their fathers”  2. “The stock market crashed in 1929 because of the sexual looseness of people in the 1920s” - In each case, an aggregate-level event is being reduced to an explanation at the individual level. - Such complex phenomena cannot be reduced to hating fathers or sexual behavior. Surely other causal factors (most aggregate levels) are involved. Spuriousness - When the relationship is false, a mirage - Spuriousness: two variables are associated, but are not causally related because of an unseen third variable that is the cause of both X & Y - Example: 1) # of fire trucks (x) → extent of fire damage (Y) - X and Y are determined by an unseen third variable (size of fire) or control variable 2) Hair length (x) → TV programs - Unessen variable is Gender, associated with both X & Y Readings & Works for Today - Readings - Read Chapter 5 - Continue to work on Assignment #1: now you can do sections 3, 4, and 8 in Assignment #1 - Turn your research question into Research & Null Hypotheses - Determine independent and dependent variables in your hypothesis Lecture 4 (Chapter 6) Step 5: Measurement (Conceptualization & Operationalization) Introduction - To test your hypothetical answer to the RQ (hypothesis), you need to empirically measure the concepts in your hypothesis - E.g., consider this hypothesis: - Socioeconomic status (SES) is positively related with the health status (HS) - How to measure SES and HS Measurement Process - Measuring concepts in Natural Science is relatively an easy task, using well-developed and precise measurement instruments. E.g: - Measurement is a process, translating concepts into measurable forms linking concept to its corresponding empirical data through an indicator. That is: - Concepts (abstract) → indicator (a bridge) → Empirical Data (in number) - Concept: abstract summary of a behavior, attitude, or characteristic - Indicator (measure): tools used as observable representation of concepts - What about social concepts? Can we easily measure them? What indicators can we use to measure below social concepts? What empirical data will we produce? - Definition of social concepts are highly determined by time and context of usage - E.g., : The definition of “Health Status” varies from time to time and from society to society, compared with that of “weight” - Directly measure less abstract concepts (e.g., weight, height, age, sex, race, and education) - Indirectly measure more abstract concepts (e.g.,: magnetism, light speed, attitude, ideology, gender roles) - Conclusion: measurement of a social concept is a precise, careful, sensitive, and hard task. So, we need to practice the following steps. Steps in Measurement Process 1. Conceptualization: a process clarifying and defining the study of concepts - Define precisely the concept - What do you exactly mean by the concept - Use existing theories, literature (textbooks, dictionaries, encyclopedias, and journal articles) - Your definition is base for the next steps - Dimension of the concepts - Unit of Analysis  - E.g., : Socio-economic status is define: “Social and economic standing of a person in a social stratification system.” - Two dimensions: social & economic - Individual as unit of analysis - It is distinct from alternative concepts such as “social class” and “social stratification” - We must stick to this definition through the whole process of measurement SES 2. Operationalization: links a conceptual definition to indicators or measures (bridge) - You can use existing indicators or create new ones - Examples of SES: - Social standing measured by level of education - Economic standing measured by level of income - NOTE: you also need to define level of education and level of income before developing survey questions for them Measurement Process - Example 1 - Hypothesis: - “Socioeconomic status is positively related with the health status” - Conceptual Definition: - IV: Socioeconomic Status = “social and economic standing of a person in a social stratification system” - DV: Health Status = “an individual’s assessment of his or her general health” - Operationalization: - Operationalized IV = two survey questions on individual’s education and income - Social ranking = number of years of education - Economic ranking = level of income - Operationalized DV = a survey question - “In general, how would you say your health is?” (Excellent, very good, good, fair poor) Measurement Process - Example 2 - Hypothesis: - “Members of non-dominant racial groups are more likely than dominant racial groups to believe that the policing is racially biased” - Conceptual Definition: - IV: “dominant racial group” = Whites “Non-dominant racial groups” = Blacks - DV: “racially biased policing” = unequal treatment by the police against White and Blacks - Operationalization: - Operationalized IV = Do you consider yourself a white or a black? - Operationalized DV=  - A police usually treats blackbetter, the same, or worse than whites (better, same, worse) - A police usually treats Black neighborhoods better the same, worse than White neighborhoods (Better, same, worse) - A police is more likely to stop Black drivers than white ones (agree, undecided, disagree) Measurement Process - Example 3 - Hypothesis: - “Individuals who begin their first relationship in a cohabitation are expected to hold more positive attitudes to a non-marital childbearing than those who do not.” - Conceptual Definition: - IV: “cohabitation = “A living arrangement in which an unmarried couple lives together in a long-term relationship.” - DV: “attitudes to non-marital childbearing” = “mental orientation of individuals toward childbearing outside of marital union (i.e., cohabiting unions, same-sex unions, single parent) - Operationalization: - Operationalized IV = a survey question: - “Ever experienced cohabitation in the first relationship with a partner? Yes/no. - Operationalized DV = using six attitude survey question: 1. “It is acceptable for a divorced person to live with his or her children and a new partner without being married to that person 2. “Government should initiate giving the right for same sex couples [children]” 3. “A single woman should never choose to have a child” 4. “A single man should never choose to have a child” 5. “A child needs a home with both a father and a mother to grow up happily” 6. “When two people decide to have children, they should first get married” Alternative responses: (strongly agree, agree, disagree, strongly agree) Measurement Errors - Reliability: refers to what we measure, it is reliable? - Reliability refers to “consistency” of a measure (indicator) - If people answer a question (indicator) the same way on repeated occasions, then it is reliable - E.g. : if you read the same weight on your bath scale over 10 times → reliable scale (if no change in your body!) - E.g., the same distribution in responses to “in general, how would you say your health is?” in 5 surveys → highly reliable indicator. Reliability - Four ways to improve reliability of measure: 1. Clearly conceptualize all concepts - Each measure should indicate one and only one concept - Remove “noise” (i.e., information related to other concepts) from the concept under study 2. Increase the level of measurement (will see in the next class) - Measured at the most precise level possible - Use the highest level of measurement (if possible) 3. Use multiple indicators of a variable: - A variable measured by multiple indicators is more stable, perfect and comprehensive - E.g.: Life satisfaction can be measured by more than one question (item) 4. Use pretests (pilot studies) and replication: - Poorly designed, vagues and general items or questions will be recognized through a pilot study or replication. - E.g., use Anomie scale from the past researches, but add or modify its items based on the study context and time - E.g., use anomie scale from the past researches, but add or modify its items based on the study context and time Validity - Validity: refers to whether we are really measuring what we want to measure - To what extent is the measure measuring the conceptual definition? - Validity tells us how much the measure is “true” and “correct” - A measure can be valid only for particular purpose and definition - E.g., a measure of “health status” designed for adults might be invalid for measuring health status of children Types of validity 1. Face Validity: judgment of specialities → whether the indicator can measure the concept 2. Content Validity: the extent that the measure captures the entire meaning of the concept - If conceptual definition as a “space” → content validity says whether the measure represents all areas of space - E.g., examples of non-marital births, using six indicators representing all different types of non-marital childbearing → this is a content-valid measure. 3. Criterion validity: Whether results of the measure agrees with that of another standard measure, widely accepted.  - Concurrent Validity: results of out measure must be associated with those taken from a preexisting measure - E.g., Results from your IQ measure are strongly associated with those from existing IQ test (assuming identical definition of IQ) - Predictive Validity: the results of your measure are strongly associated with the future behavior of the subjects in your conceptual definition - E.g., Students with high scores on Scholastic Assessment Test (SAT) in the US or with high scores on EQAO (Education Quality and Accountability Office) Test in Ontario will perform well in college, and those with low scores on the test will perform poor in college → if true, our measure has a high predictive validity. Relation Between Reliability and Validity - It is more difficult to achieve a validity than a reliability, as we work with abstract concepts - Reliability is necessary for validity - A measure can be reliable, but invalid. That is, it produces the same results, but measures something far from the conceptual definition - E.g., the results from your bathroom scale can be identical, but these results would be different from the result from an “official” scale which measure your “true” weight. Reading and Words for Today - Readings: chapter 6, pages 101-117 - Omitted pages: - Quantitative and qualitative measurement, p. 105 - Qualitative conceptualization and Operationalization, pp. 109-110 - Reliability and validity in qualitative research, pp. 114-115 - Other uses of the terms reliability and validity pp. 116-117 Week 5 (Chapter 6) Step 5: Measurement (Indexes & Scales) Operationalization - Practical Tips - In operationalizing any concept, be clear about: - Type of variable (Discrete/Continuous) - Level of Measurement of variable A. Type of Variable: - Discrete: Cannot be subdivided (e.g., Gender, Marital, Status, Number of Children) - Continuous: Can be subdivided infinity (e.g., time, Age) B. Levels of Measurement of a Variable - Conceptual Definition affects or determined - “Type of indicator or measure” - “Level of measurement” - “Statistical analysis” Levels of Measurement of Variable 1) Nominal: labels & classification: - Example: gender (Women/Men), marital status (Married, Single, Others), religion (Christian, Islam, Jew, Others) - Attributes (values) of a variable should be  - Mutually exclusive: only one category for each case - Exhaustive: Every possible situations should be covered (“other”) 2) Ordinal: classification + ranking (“higher/lower”, more/less); distance between scores are meaningless, E.g.: attitudes, social class, grade rank 3) Interval: classification + ranking + describing intervals as it has a real metric (equal interval). E.g.: IQ, Temperature - NOTE: the score of zero is arbitrary, not real - IQ=111>IQ=110 by one unit, if IQ=0 can we say one has “no intelligence”? - If temperature= 0 degree, can we say “NO temperature”? 4) Ratio: implies classification + ranking + describing intervals + an absolute zero point: E.g, Age, income, #of children - Absolute zero = “non-existence” - Make relative difference possible: - E.g.: 4 children is really double of 2 - Hierarchy of numbers: a higher level implies all the properties at lower level plus something unique - Nominal & Ordinal-level variable are discrete - Interval- & Ratio-level variables can be discrete or continuous: (e.g.,: # of children is discrete, but age is continuous) Exercise - Level of Measurement - For each one, indicate level of measurement & if is discrete or continuous. A) To measure the running time of 100 runners in the last Olympic. B) To find whether or not each household in 50 neighborhoods has cable TV C) To rate respondents’ agreement with each of ten statements about nuclear power on a scale of 1 to 7. D) To measure income as under $18,000, 18,000-30,000, 30,001-50,000, etc - Note: can we move from a high to a low or from a high level of measurement? It is always better to aim at the highest level. Specialized Measures: Indexes & Scales - Indexes and scales: techniques for measuring multidimensional concepts - They are also called “data reduction devices” - E.g.,: attitude concerning the “need” of social programs in Canada. - One way is to use and open-ended question: “What do you think about…” - Another way is to use structured items: - Social programs should be maintained - Social programs should be modified, or… - But a single question is insufficient Index - Index: a composite measure, adding indicators into a single score. - Often measured at interval or ratio levels \ - E.g.: Social Development Index (SDI) = Literacy rate + Life expectancy + % of houses having telephone - Check an index or scale for - Being mutually exclusive and exhaustive, - Unidimensionality: all indicators measure only one concept - E.g., indicators of SDI should only measure aspects of “social development”, not also some aspects of “economic development” Index Construction - To develop an index, follow these steps 1. Propose a “conceptual definition” for the concept 2. Recognize appropriate indicators 1. Develop the index by summing the scores on values of all indicators for each case. 2. Check for validity of the index, using statistical method, such as factors analysis - Example: “Social Development Index (SDI)” of provinces of Iran (Erfani, 2005) 1. Conceptual definition of “social development” : “improvement of life standards” 2. Operational definition: “life standards are measured by four indicators: life expectancy at birth (E0), literacy rate of women aged 15 and over (EDU), and urbanization (URB) 3. Index of SDI=(E0 + EDU + URB)/3 Index - Example 2 (Babbue, 1999:158) - To measure “Political Activism”, we operationalize “Political Activism” by seven questions 1. Did you write a letter to a public official? Y1 N0 2. Did you sign a political petition? Y1 N0 3. Did you give money to a political cause? Y1 N0 4. Did you give money to a political candidate? Y1 N0 5. Did you write a political letter to the editor? Y1 N0 6. Did you persuade someone to change his/her voting plans? Y1 N0 - NOTE: all of these different actions are assumed to represent the same degree of political activism - The scores will range from 0 to 6. Two persons getting a score of 4 might have been involved in different actions. Indexes - Good Examples - Some most commonly used Indices (by the UN, internationally - at aggregate level):  - Gender - related Development Index (GDI) - Uses 3 indicators: life expectancy, education, income - Gender - Empowerment Measure (GEM) - Uses 3 indicators: MP seats, managers, job access - Development Index (uses so many economic indicators!) - Human Development Index (HDI) - Uses life expectancy, literacy, living standard Scales - Scale: measures “how individuals feels or thinks about something or somebody” - Intensity of a respondent’s feeling or thinking (something emotional involved) - Essentially ordinal - Select items in such a way that they reveal the levels of intensity (response categories) - Highly dependent on the contexts. Do not use the scales for all societies and for all times! Likert Scale - A statement on which respondents agree or disagree with something or somebody (assertion vs. question) - Needs minimum two categories (e.g., agree/disagree), but better 4 to 9 - Balanced response categories (e.g., SA, A, U, D, SD) - Neutral category (i.e., undecided/don’t know)? - Summated rating scale: adding scores of Likert scales (just like an Index) Likert Scales - Practical Points 1. Add scores on item for each case 2. Avoid “problem of response set (RPS)” 3. To improve validity of concept, use multiple indicators 4. Create the items from reactions of people to the subject under study, found in news, press, conservations, etc. Readings and Works for Today - Readings: Chapter 6, pages 117-124 - Assignments - Complete assignment 1 - Start working on assignment 2 Week 6 Sampling Methods Introduction - So far you - Developed a hypothesis - Conceptualized and operationalized the concepts in the hypothesis - The Next steps 1. Method of Data Collection 2. Selecting a sample of population under study - How about the research which (whom) we want to collect data about? - Can we interview/observe the whole population? - Or, could we be content with a sample form the population? Basic Ideas - Problem: The population is too large to gather from its every element. - Solution: choose representative sample from the population - But, how to select a sample? And what sample size is good? Sample Designs - Two main approaches in selecting a sample: - Probability sample - Based on mathematical theory of probability - Two motivations: 1. Saving cost and time  2. Accuracy of collected date - Aim: to generalize results to the population - Mostly used for quantitative research - Non-probability sampling: - Aim: to find informant cases (e.g., individuals) → - Provide deep and rich information about the social reality - No idea of generalization Non-Probability Sampling Methods - Convenience Sampling: selects people in/her convenience - Ineffective and highly misrepresented - Cheap and quick - E.g., Person-on-the street interview conducted by TV - Quota Sampling: is the improved version of convenience method - Preserves the proportions as in the population - E.g.,: select 500 cases based on fixed, real proportions of Male/Female and Age groups of 18-34, 35-59, 60+ in the population No-Probability Sampling Methods - Purposive (judgment) sampling Research use judgment in selecting samples from “specific purposes”: 1. To find unique and informant cases - E.g.,: to select “fashion” magazines purposely to find cultural themes about clothing → helps researchers to design a good survey questionnaire studying “cultural norms of clothing” in Canada 2. To find members of a difficult-to-reach population - E.g., talking with people associated with prostitutes (e.g., police, social groups) to locate population of prostitutes 3. To identify particular types of cases for in-depth investigations. - E.g., women performed clandestine abortion, where abortion is illegal - Mostly used for explorative research - Snowball sampling: - Used to identify subjects in a network - To show the network by a Sociogram - Linked directly or indirectly by a specific linkage - Used to select elements of hidden population - E.g, illegal drug users, homosexuals, people with HIV/AIDS, or homeless people Probability Sampling: Terminologies - Population or “target population” (N): contains ALL element relevant for study - Census: a study which observes all elements - Sample (a): a carefully chosen subset of the population - Element: each unit or case in the population sample - Sampling Frame: List of names of all elements in the population - Sampling element: every element in the sampling frame - Sampling ratio: ratio of the size of the sample to the size of the population - E.g.,: 1000/1,000,000= 0.001 or or 0.1 percent (0.001 * 100) NOTE: define precisely target population in your study by 1. Element being sampled, 2. Geographical location 3. Temporal boundaries of population - E.g.: “All men aged 18+ living in North Bay, Ontario of Canada from September 1 to December 31, 2024) - Basic Logic: selection operates randomly (i.e., each element has an equal chance or probability of being selected). - Probability sampling Methods: 1) Sampling Random Sampling (SSR) 2) Systematic Sampling (SYS) 3) Stratified Sampling (STRS) 4) Cluster Sampling (CLS) Simple Random Sampling (SRS): - N units in the population are independent - We select n element into the sample - Elements have equal probability of selection - (i.e., probability of selecting each unit = 1/N) - Assumes a sampling frame is available - Follows the steps using the table of random numbers, or use a random number generator - Should one use the random numbers? Yes ALWAYS! - Steps in SRS: 1. Have ready or create a sampling 2. Number each element in the sampling frame 3. Decide on a sample size 4. Count the largest number of digits needed for the sample 5. Locate a random-number table 6. Begin to select a random number everywhere on the table while moving in a specific direction 7. Ignore the repeated and out-of-range numbers Systematic Sampling (SYS) 1. Number each element in the sampling frame 2. Decide on a sample size 3. Compute a sampling interval (N/n; e.g., 20/5=4) 4. Select randomly a starting point, using random-number table while pointing blindly at a number 5. Select the samples systematically based on the sample interval (e.g., select every 4th (e.g., select every 4th number of the 20 numbers.) - Limitation: systematic errors (e.g., cyclical data in Table 7.3 of Text, P.143) Stratified Sampling (STRS) - Based on the idea of heterogeneity - More representative of population than SRS - steps: 1. Divide population into sub-population (strata) of interest, based on supplementary information of the population which are not always available! 2. Draw a random sample for each strata by SRS or SYS, based on the proportion of population in each stratum - When to use STRS? 1. To avoid missing of the stratum of interest, with a small size in the population (e.g., population of PEI, Aboriginals), in a national sample. 2. To avoid overrepresentation or underrepresentation of population of a given stratum, happening by SRS or SYS, by giving a fixed proportion to each stratum in STRS (e.g., Box 7.6 of Text) - E.g.,: Population: N= 20,000 individuals aged 18+, where 90% Christian (N1), 8% Jewish (N2), 2% Muslim (N3) - N=200 → n1=180 (0.9 * 200), n2=16 (0.08 * 200), n3= 4 - If we use SRS, we may miss Muslim strata, which is the focus of our study Cluster Sampling (CLS) - Addresses two problems: 1. Lack of a good sampling frame for a dispersed population 2. High cost for reaching the sample elements, specially when the elements, or itself as a sampling element - Steps: 1. Select randomly decided number of clusters out of all listed clusters in the population 2. Randomly select decided number of elements within each selected cluster, after creating a sampling frame for each selected cluster - Advantages of CLS:  1. Available good sampling frame of clusters 2. Creating a sampling frame within each selected cluster is manageable 3. Lower costs of reaching to elements within a selected cluster 4. CLS can be multistage (e.g., Provinces, counties, towns, blocks, households,...) - Disadvantages of CLS: 1. Less accurate, because each stage introduce sampling error: - A tradeoff between cost and accuracy → how many clusters and how many elements? - Few clusters, with many elements in each one → less representativeness (similarity of cases within a cluster) and low cost - Many clusters, with few elements in each one → more representativeness and high cost How large should a sample size be? - To compute a sample size, we need to use some statistical calculations which are beyond this course - However, as a rule of thumb and based on experience, if everything else being equal: - For small population (i.e., less than 1000), choose 30% of the population as a sample size (i.e., 300) - For moderately large population (i.e., 10,000), choose 10% of the population as a sample size (i.e., 1000) - For large population (over 150,000), choose 1% of the population as a sample size (i.e., 1,500) - For very large population (over 10 million), choose 0.025% of the population as a sample size (i.e., 2,500) Readings Readings: chapter 7 - Omitted topics - Sequential sampling (p. 133) - Random-digit dialing (pp.148-150) Lecture 7 (Chapter 11) Analyzing & Interpreting Data Analyzing the Data Making sure that the data are clean, you can start analyzing data by using an appropriate statistical method  A researcher usually analyzes the data at three levels to obtain further findings: 1. Univariate Analyzes: analyzes one variable 2. Bivariate Analyzes: analyzes two variables 3. Multivariate Analyzes: analyzes more than two variables Univariate Analysis Frequency Distribution: summarizes data into a table Graphic Presentation: illustrates the overall shape of distribution and highlight any clustering of cases ○ For discrete variables at any level of measurement 1. Bar Graph (NOTE: Separate the bars, because…) 2. Pie Chart Mostly for continuous, interval/ratio variables: 1. Histogram (Note: connect the bars!) 2. Line Charts or frequency Polygons (using the midpoints only) Bivariate Analysis Examine whether two variable are independent Cross-tabulation: used for N, O, R, I var. ○ Contingency table: title, cell, row & column %, X & Y ○ Hypothesis testing by chi-square and significant level ○ Measures of Association: measure strength and direction of a relationship Phi or Cramer’s V and Lambda range between 0 and 1, used for nominal level variables Gamma ranges between -1 and +1, used for ordinal-level variables Income BY Age (GSS 2001) Interpretation of results: Column percentage show that with the increasing age, income levels increase Since significant level (P-value =0.0005) is smaller than a=0.05, so H0 is rejected. So, there is a statistically significant relationship between age and income. Measure of Association Guidelines to interpret Gamma: Range Strength 0.00-0.29 Weak 0.30-0.60 Moderate Greater than 0.60 Strong Interpretation of results: The positive sign of Gamma = 0.121 indicates a positive relationship between age and income level. That is, income increases with rising age. The value of Gamma =0.121 shows a weak relationship between age and income Multivariate Analysis Statistical Control: check for spuriousness ○ Eliminate alternative explanations ○ Constructing trivariate table → partial tables ○ Compare the original with partial tables → traces changes in direction and strength of the original relationship 1. Partials replicate the same relationship between two variables in the original table → no spurious relationship 2. One partial does not replicate the original relationship, but other do 3. No relationship in original table, but relationships appear in partial tables (e.g., : see next slide: income BY age BY sex) Income BY Age BY Sex Interpretation of Results: Partial tables (male and female tables) replicate the same relationship between age and income observed in the origin table. (Total table). So, there is no spurious relationship between age and income, when controlling for the effects of sex of respondents Midterm N → N use Phi, Cramer’s V, Lambda only Pattern of relationship is interpreting the percentage