Sociology Research Methods Overview

Week 2 - sociology as a science/ overview of research methods part 1 How do we know what we know -​ Sociologists seek to contribute to and enhance knowledge about the social world -​ Research methods (methodology) -​ A systematic process of inquiry applied to learn about the social...

Week 2 - sociology as a science/ overview of research methods part 1 How do we know what we know -​ Sociologists seek to contribute to and enhance knowledge about the social world -​ Research methods (methodology) -​ A systematic process of inquiry applied to learn about the social world -​ The goal of research methods is to construct a defensible version of reality The science of sociology -​ Sociology is a science -​ What is science? ​ A way of knowing that attempts to systematically collect and categorize facts or truth ​ Twin pillars of science : logic and observation ○​ Scientific research operates on the theoretical and empirical levels. A scientific assertion must makes sense and by supported by empirical observation ​ Scientists gather information about facts in a way that is organized and intentional and usually follows a set of predetermined steps -​ The scientific method is central to making the gathering of empirical evidence (that is, evidence derived from observation) systematic because it -​ Established parameters or specific guidelines that help ensure that the findings are objective and accurate -​ Provides boundaries that focus a study and organize its results -​ Offer sociologists a shared basis for discussion and analysis -​ Why is this important? -​ What is the peer review process?? A word about revision -​ Sociologists should not be attached to any particular results. Research should be independent of personal beliefs (religions, politics etc) Sociology and the research process -​ Sociology research aims to find patterns of regularity in social life -​ Data collection methods are designed to detect these social regularities -​ Data analysis techniques are used to interpret them, and theory is used to help explain them  -​ Specify the research question (example) -​ Review the scientific literature -​ Knowledge is cumulative (meaning -​ Propose a theory/ state hypotheses (example) -​ Interrelated ideas that provide a logical explanation of empirical realities -​ Select a research design (what are examples) -​ A word about replication (why do it) -​ Collect the data (cases, sample, population -​ Analyze the data (how) -​ Draw inferences/ conclusions (generalization) -​ Disseminate results (peer review/ public sociology) Quantitative and qualitative methods -​ Quantitative methods result in data that can be represented by the condensed into numbers (aggregate. Compare, summarize data) -​ Less depth, more breadth - focus on a larger number of cases -​ Survey research - most common quantitative method in sociology -​ Qualitative methods are way of collecting data that yield results such as words or text (in depth interviews) -​ Gain in depth understanding of a relatively small number of cases -​ Potential for greater richness in meaning that quantified data -​ Sometimes qualitative and quantitative methods are discussed in a way that suggests they are in opposition - this is an inaccurate and unnecessarily limited perspective -​ Researchers may prefer one approach, either because their own research questions are better suited to one appeach or because they were trained more intensely in one specific method -​ Complementary, not competing : yes these methodological approaches differ… but it is more appropriate to see them as having different goals, strengths and weaknesses Basic and applied research -​ Basic : sometimes researchers are motivated to conduct research simply because they happen to be curious about a topic  -​ The goal of the research may be to learn more about a topic or test a theoretical perspective or solve an intellectual puzzle -​ Applied : refers to sociology that is conducted for some purpose beyond or in addition to a researchers interest in a topic -​ Find solutions to problems or discover ways of living better etc -​ Might involve research for a client -​ Evaluation research - often used to test the effectiveness of social policy or program ( is it successful or not) Variables -​ Variable : a characteristic or measure of a social phenomenon that can take different values -​ When one variable causes another, we have what researchers call independent and dependent variables Independent : is one that causes another Dependent : is one that is cause by another they depend on independent variables Relationship between variables -​ Relational statements connect two or more variables. Knowing the value of one variable provides some information about the values of another variable -​ Probabilistic relationship : two variables go together with some degree or level of regularity. As the level of one variable increases, the level of another variable also tends to increase or decrease To suggest causation three criteria must be satisfied -​ There must be a relationship or correlation between the indi and dep variables -​ The indi variable must be prior to the dependent variable -​ There must be no other variable that is responsible for producing the casual relationship - that is, some other X predicts both the independent variable and dependent variable in a way that helps explay why they are related Observation and evidence -​ Knowing through observation -​ Selective observation -​ Noticing only social patterns that one has experienced directly or wishes to find : confirmation bias (seek out or recall information that supports one's existing views) -​ Overgeneralization -​ Assuming that board social patterns exist based on very limited observation (vibes) Week 3 - overview of research methods: research questions/design part 2 Sociology and the research process -​ Sociology research aims to find patterns of regularity in social life (education and distress) -​ Data collection methods are designed to detect these social regularities (sampling surveys) -​ Data analysis techniques are used to interpret them, and theory is used to help explain them -​ Let's do a deeper dive into each of those elements Types of research 1.​ Exploratory : a type of research that examines new areas of inquiry, with the goals of (1) scoping out the extent of a problem or behavior ; (2) generating initial ideas about that phenomenon ; or (3) testing the feasibility of doing more extensive study (Pilot study) 2.​ Descriptive : a type of research directed at making careful observations and generating detailed documentation about a phenomenon of interest (unemployment rate) 3.​ Explanatory : a type of research that seeks explanations of observed behaviors, problems, or other phenomena. Explanatory research seeks answers to “why” and “how” questions Inductive vs. deductive approaches 1.​ Inductive : an approach to empirical investigation in which researchers start with a set of observations and use the empirical evidence they gather to create a general set of propositions about how the world operates 2.​ Deductive : an approach to empirical investigation in which researchers start with a social theory and then test its implications with data Reinforcing the essentials : the research process -​ Specify the research question -​ Review the scientific literature -​ Propose a theory and state hypotheses -​ Select a research design -​ Meanurment -​ Collect the data -​ Analyze the data and draw conclusions -​ Disseminate the result Research methods : some basics -​ A hypothesis is an assumption about how two or more variables are related: it makes a statement about the relationship between those variables -​ It is an educated guess based on theory, observations, or previous scientific discoveries (again - the importance of cumulative knowledge) -​ The hypothesis formulates this guess in the form of a testable proposition Hypotheses -​ Researchers rarely state that they have proven their hypothesis - rather, hypotheses are “supported” or “not supported” -​ The claim that one has “proven” a hypothesis is too bold; it suggests that one thinks a relationship exists with certainty -​ Caution about our findings allows for the possibility that new evidence or ways of testing a relationship will be discovered  -​ Researchers sometimes describe a null hypothesis, one that predicts no relationship between the variables being studies (education in unrelated to symptoms of depression) -​ “Reject the null hypothesis” ; the researcher claims that the variables being studied are related (education is related (negatively) to symptoms of depression). One type of research design: surveys -​ Surveys are used to collect data from study participants who respond to a series of questions about attitudes, beliefs, behaviours, and other variables -​ A survey targets a population, people who are the focus of the study -​ Researchers survey a sample: a manageable number of subjects who (ideally) represent a population - representativeness (examples:challenges?) -​ A study value often depends on how generalizable the sample-based findings (e.g., univariate,bivariate,multivariate) are to the broader population Unit of analysis, population and samples -​ Sampling bias : a types of bias that occurs when the elements selected for inclusion in a study do not represent the larger population from which they were drawn -​ Examples using the “education and subjective quality of life” study -​ When conducting research, we focus on studying a particular unit of analysis - a person, collective, or object that we want to learn more about. That unit of analysis encompasses a larger population we are interested in making claims about. It is usually not possible to include every unit of the population in a research study, so we must select a small subset of the population (a sample) from which we actually gather data Operationalization -​ In the research process, sociologists develop an operational definition; that is, they define the concept in terms of the steps it takes to objectively measure it. -​ The tools that sociologists use for measurement provide a bridge between conceptual and empirical levels -​ From the responses to the measurements (survey questions), we infer the extent that the concept exists at the empirical level. This involves translating a social phenomenon into a quantifiable or numerically measurable variable ( a measure with different values) -​ This lets sociologists formulate their predictions using statistical language, to present findings in graphs and tables, and to perform statistical techniques to show relationships -​ The quality of measurement can be a point of critique in peer review Correlation : education -> well-being? -​ At a basic level, researchers might wonder if there is a correlation between variables. What does this mean?  -​ When a change in one variable coincides with a change in another variable, we say there is a correlation -​ This does not necessarily mean that change in one variable causes a change in other variables only that they are associated To suggest causation, three criteria must be satisfied -​ There must be a relationship or correlation between the independent and dependent variables -​ The independent variable must be prior to the dependent variable -​ There must be no other variable that is responsible for producing the causal relationship - that is, some other x predicts both the independent variable and dependent variable in a way that helps explain why they are related (the usefulness of “control variables” in research) -​ The possibility of reverse causality : example : education and depression Week 4 : overview of research methods part III : sampling/ concepts/measurement Reinforcing the essentials -​ The word “systematic” is often used to argue that sociology is a science. We follow the scientific method -​ Establishes parameters that help ensure that the findings are objective and accurate -​ Provides boundaries that focus a study and organize its results -​ Offers sociologist a shared basis for discussion and analysis Sampling and measurement are essential pillars in that process Unit of analysis, population and samples -​ Sampling : the process of selecting cases that will be analyzed for research -​ Sampling frame : a list of members of a population that is available to researchers, which they use to select cases for their sample. Ideally, the sampling frame includes every single member of that population -​ Generalizable : when a study's result can reasonably tell us something about the larger population from which its sample was drawn (“representative sample”) -​ Nonrespondents : individuals who decline to answer questions for a survey or in depth interview or who cannot be reached by researchers Key terms 1.​ Sample statistics : means, medians, standard deviations, and other numerical values that describe a study’s sample and are usually meant to estimate parameters in the target population  2.​ Population parameters: means, medians, standard deviations and other numerical values that summarize the characteristics of a population. When researchers cannot observe the actual population parameters, they generate statistics from a sample of that population to estimate them Types of sampling -​ Nonprobability sampling -​ a type of sampling in which the researches do not know the likelihood that a person (or other unit of analysis) in the population will be selected for membership in the sample -​ Probability sampling -​ A type of sampling in which the researchers know the likelihood that a person ( or other unit of analysis) in the population will be selected to be in the sample (also called random sampling) -​ Sampling error : the difference between the statistics obtained from a sample and the actual parameters of a population. Probability sampling allows for the calculation of the sampling error that is expected given the size of the sample being used -​ Sample size relevant to the topic of sampling error -​ Because surveys only select a sample of the population, the result probably won't exactly match the true result that we would get if we interviewed everyone in the population -​ The margin of sampling error describes how close we can reasonably expect a survey result (sample statistic) to fall relative to the true population value (parameter) -​ Repeated sampling - can help sharpen our precision - and a higher volume of more credible samples across polls should improve the accuracy of the overall polling average Operationalization -​ In the research process, sociologists develop an operational definition; that is, they define the concept in terms of the steps it takes to objectively measure it -​ The tools used for measurement provide a bridge between conceptual and empirical levels -​ From responses to measurements (survey questions), we infer the extent that the concept exists at the empirical level. This involves translating a social phenomenon into a quantifiable or numerically measurable variable (a measure with diff values) Different types of sampling approaches -​ Simple random sampling : a sampling technique where the researcher gives all members of a population (more accurately, of a sampling frame) an equal probability of being selected  -​ Systematic sampling : involves systematically choosing sampling units from a list by defining an interval (k) and then selecting every kth element in our list. In an example of interval of 3 so you select every 3rd individual -​ Stratified sampling : a sampling technique where researchers divide the study population into 2 or more mutually exclusive subgroups (strata) and then draw a sample from each subgroup -​ Is used to ensure that the sample adequately represents the identified subgroups. To make sure that a small subgroup is adequately represented in our sample Proportional stratified sampling : carve up our sampling frame into strata then choose cases from each stratum in line with its numbers in the larger population -​ Disproportionate stratified sampling : an approach to stratified sampling in which the sizes of the subgroup samples do not match their relatives sizes within the population -​ Oversampling : when a subgroup represents a greater share of a sample than the same subgroup represents in the larger population -​ Weighting : adjusting for how much particular cases contribute to the statistics for a sample Different kinds of bias -​ Nonresponse bias -​ When respondents and nonrespondents differ in important ways, which means that the relevant characteristics observed in the sample differ from those in the target population -​ Self-selection bias -​ When certain types of people are more likely to volunteer for (or to be selected into) a sample. For example, people with strong opinions on an issues may be more likely to participate in a study -​ Attrition bias -​ When the participants who leave a study tend to come from particular subgroups, thereby undermining the representativeness of the study's sample Example Conceptualization : what is -​ A subjective feeling state of being alone, separated or loneliness apart from others, and has been conceptualized as an imbalance between desired social contact and actual social contacts ​ ​ ​ ​ Research question and -​ As levels of x increase levels of y increase/decrease hypothesis -​ Age associated with loneliness Operationalization -​ Loneliness : a subjective feeling state of being alone, separated or apart from others -​ 3 item index in the past month  -​ Feel lonely -​ Feel isolated -​ Feel left out -​ Items are positively correlated with each other An example of bivariate statistics: correlation -​ Correlation between variables -​ When a change in one variable coincides with a change in another variable, then there is correlation -​ This does not necessarily mean that change in one variable causes a change in another variable, only that they are associated A word about causation -​ To suggest causation, 3 criteria must be satisfied -​ There must be a relationship or correlation between the indi and dep variables -​ The indi variable must be prior to the dep variable -​ There must be no other variable that is responsible for producing the causal relationship - that is, some other X predicts both the indo and dep variable in a way that helps explain why they are related Quantitative and qualitative research -​ Using loneliness as an example, what would be an example, what would be an example of qualitative data? How could that qualitative data inform quantitative data Tut 1 sample measurement and error How do we decide on who to sample -​ First considering the population of interest, the total group who share some behaviour or trait we are interested in -​ A single part of that group are considered the unit of analysis ( typically individualism groups, organizations and countries) - level we are looking at -​ The sample is a smaller collection of units of analysis from the larger population of interests ( selected subset for study) -​ Target population ( complete group of interest) Why do we sample -​ Because most situations we do not have enough information about the entire population of interest  -​ Sampling affords a way to learn about the larger population of interest. What what we learn from samples can at best be only partially generalized to a larger population How can a sample represent a population -​ Sampling can be divided into 2 types -​ 1. Probability sampling ( random selection- equal chance of selection) -​ Non-probability sampling ( convenience?purposive selection - selection based on convenience) How do we measure -​ A measurement is a variable that represents the broader phenomenon we are interested in -​ Measurement requires translation concepts (bigger ideas, like stress and it underlying physiology) into things we can observe systematically (such as the difference between someone saying “i'm stressed: or “i'm not stressed” this is called operationalization Abstract concept (broader phenomena we want to understand -> operationalization -> concrete measurement (self report (1-10) heart rate (BPM) cortisol (ng/ml) -​ Observable indicators that can be measure systematically OPERATIONALIZATION on midterm What makes a good measure -​ Measurement develop through innovation and testing of their validity and reliability -​ A valid measure is one that captures what we intended (e.g. agreements to “im frequently stressed” related to other distress and biomarkers) -​ A reliable measure consistently produces the same results from the same phenomenon - consistent measure over time -​ Valid but unreliable -​ Reliable but not valid -​ Valid and reliable How do we know we went wrong -​ If the aim of sampling is to represent the target population and the aim of measurement to represent the broader concept, then deviation from either goal can meaningfully be called error -​ Systematic error - is bias in one direction due to a cause and random error is unpredictable variation -​ In the social science, the two most common systemic forms of error are social desirability bias and acquiescence bias ( compliant or compliant - don’t understand and just choose a answer or compliant with the survey) -​ Random error ( unpredictable variations in both direction around the true value -​ Systematic error ( consistent bias in one direction due to a specific cause) Week 5 - midterm review How do we know what we know -​ Sociologists seek to contribute to and enhance knowledge about the social world -​ Research methods (methodology) -​ A systematic process of inquiry applied to learning about the social world -​ The goal is to construct a defensible version of reality The science of sociology -​ Sociology is a science -​ A way of knowing that attempts to systematically collect and categorize facts or truths -​ Twin pillars of science : Logic and observation -​ Scientific research operates on the theoretical and empirical levels. A scientific assertion must make sense and be supported by empirical observation - scientists gather information about facts in a way that is organized and intentional and usually follows a set of predetermined steps -​ The scientific method is central to making the gathering of empirical evidence (evidence derived from observation) systematic because : -​ It establishes parameters or specific guidelines that help ensure that the finders are objective and accurate -​ Provides boundaries that focus a study and organize its results -​ Offers sociologists a shared basis for discussion and analysis -​ Why is this important? -​ What is the peer review process? A word about revision? -​ Sociologies should not be attached to any particular results. Research should be independent of personal beliefs (religion / politics) Sociology and the research process -​ Sociological research aims to find patterns of regularity in social life -​ Data collection methods are designed to detect these social regularities -​ Data analysis techniques are used to interpret them, and theory is used to explain them 1.​ Specify the research question (example) 2.​ Review the scientific literature -​ “Knowledge is cumulative (meaning?) 3.​ Propose a theory/state hypotheses (example?0 -​ Interrelated ideas that provide a logical explanation of empirical realities 4.​ Select a research design (what are examples?)  -​ A word about replication (why do it?) 5.​ Collect the data (cases, sample, population) 6.​ Analyze the data (how?) 7.​ Draw inferences/conclusions Quantitative and qualitative methods 1.​ Quantitative methods -​ Results in data that can be represented by and condensed into numbers (aggregate, compare, summarize data) -​ Less depth, more breadth - focus on a larger number of cases -​ Survey research - most common quantitative method in sociology 2.​ Qualitative methods -​ Are ways of collecting data that yield results such as words or text (ex. In-depth interviews) -​ Gain in depth understanding of a relatively small number Variables ​ Variables : a characteristic or measure of a social phenomenon that can take different values -​ When one variable causes another, we have independent and dependent ​ Independent : is one that causes another ​ Dependent : is one that is caused by another they “depend” on independent variables Relationship between variables -​ Relational statements connect 2 or more variables. Knowing the value of one variable provides some information about the values of another variables -​ Probabilistic relationship: two variables go together with some degree or level of regularity as the level of one variable increases, the level of another variable also tends to increase or decrease -​ To suggest causation 3 criteria must be satisfied : -​ 1. There must be a relationship or correlation between the independent and dependent variables -​ The independent variable must be prior to the dependent variable -​ There must be no other variable that is responsible for producing the causal relationship - that is, some other X predicts both the independent variable and dependent variable in a way that helps explain why they are related Observation and evidence -​ Knowing through observation -​ 1. Selective observation : noticing only social patterns that one has experienced directly or wishes to find: confirmation bias (seek out or recall information that supports one's existing views)  -​ 2. Overgeneralization : assuming that broad social patterns exist based on very limited observation (vibes) Week 7 : methodological focus : survey research substantive topic :immigration and adolescent achievement Essentials of survey research -​ In survey research, a researcher poses a set of predetermined questions to a sample of individuals (respondents) -​ Those questions (items) are listed in a questionnaire - a researcher can read to a respondent, or the respondent can answer on their own through a paper or online form -​ 2 or more answers (response categories The goal of surveys is to compare the responses of many individuals. Surveys provide breadth, while in-depth interviews provide depth Survey research is an ideal method to gain a representative picture of the population that is often too large to observe directly Research can cover a large area with online or telephone surveys (computer assisted telephone interviewing) Rigorous probability sampling techniques boost confidence the results from the sample are generalizable to the population of interest -​ Surveys are standardized: they ask the same questions, phrased in the same way. This makes it possible for researchers to compare responses across individuals (trends) -​ Other methods (in-depth interviews) do not achieve the same degree of consistency and comparability as a survey. -​ Not all surveys are always reliable. A poorly phrased question can cause respondents to interpret its meaning differently, leading to inconsistent answers -​ If you construct your questionnaire carefully, however, it is likely to produce reliable results -​ FENCE SITTERS are respondents who choose neutral responses instead of disclosing their real preferences  -​ Floaters are respondents who choose a substantive answer when in actuality, they don't understand the question or have a opinion -​ Floaters can be motivated by the social desirability of having an opinion on an important issue. This can be more of an issue if a question does not provide a neutral response Surveys for explanatory research -​ Explanatory research : surveys are often the method that sociologists use to engage in causal inference - that is, determining whether and to what extent a cause-effect relationship exists between 2 or more variables -​ Surveys are not as good as experiments for internal validity - that is, whether a change in the independent variable truly does cause a change in the dependent variable. With proper statistical control (control variables), the analysis of survey data can do an adequate job of inferring causality -​ Most surveys collect data from a sample of the population of interest, so the findings from this method tend to have a higher degree of external validity : that is , we tend to be more confident that any patterns and relationships we find in our samples will reflect the realities in the larger population of interest Essentials of survey research -​ RESPONSE RATE : the percentage of people who actually completed the survey out of everyone asked to participate -​ 2,529 out of 5,600 people contacted completed this survey - response rate is 45% -​ NONRESPONSE BIAS: a type of bias in a study when respondent and nonrespondents differ in important ways, which means that the relevant characteristics observed in the sample differ from the target population -​ Prefer not to say is a non repose 6 ways to improve participation and the response rate Personalize Personalize your questionnaires Highlight Highlight your credibility in describing your study Short Keep your questionnaire short and simple Persist Be persistent in your recruitment efforts Reminders Send out pre-questionnaire notices and post-questionnaire reminders Offer Offer compensation or at least a token of appreciation CROSS-SECTIONAL SURVEYS : are surveys that are administered at one point in time. They provide a “snapshot” about how respondents think, feel Longitudinal surveys  -​ Make observations over an extended period of time. There are several types : TREND, PANEL, COHORT -​ Trend surveys focus on how people's attitudes or behaviours change over time. They do not require interviewing the same people at each time point because the interest is in trends - not specific people Panel surveys -​ The same people participate in the survey each time the researcher fields it. -​ Can have sampling bias as they often lose respondents over time (die or drop out) losing them can lead to attribution bias as the people who remain have specific characteristics - having more interest can skew the study. Cohort survey -​ A researcher identifies a category of people who are of interest and then regularly surveys people who fit that category Panel studies -​ A benefit is it is easier to infer causality - that is, evaluate whether a cause-effect relationship exists between 2 or more variables -​ Experiments are ideal for this because they randomly assign subjects into control and experimental groups and then introduce a treatment. This design helps us infer that change in the indi variable caused any outcomes in the experimental group. The subjects were randomly assigned to the two groups and ruled out the possibility that preexisting differences between those groups explain those observed outcomes. -​ In a panel study, you have a type of built in control and experimental groups : the participant earlier in time is the control group, and that participant later in time (after the change in the indi variables) is the experimental group The 5 most important things related to sociological inquiry 1.​ What are research questions and why are they important for sociological inquiry 2.​ What are hypotheses and why are they important 3.​ What is the role of theory in developing hypotheses 4.​ What is research design and why do we do it as part of sociological inquiry 5.​ What are variables and how are relationships between variables usually described in hypotheses? The problem and research questions -​ Sociologists are curious if “high achieving students of color” are “socially sanctioned” -​ Do they become less popular among their peers if they display academic excellence? -​ Why is this a problem? -​ It might undermine the motivation to do well in school -​ This in turn could amplify inequalities between groups Research design -​ The researchers use an existing dataset from a survey : the national longitudinal study of adolescent to adult health Probabilistic relationship #5 -​ Two variables go together with some level of regularity -​ “ each point increases in GPA is associated with a 16% increase in the odds of receiving any friendship nominations” (as GPA goes up so does popularity -​ The effect of academic achievement on popularity among peers is negative for black and american indian youths from native born families and nonexistent or positive for immigrant minority students of any race/ethnicity Reinforcing the essentials : relationships among variables -​ We conducted the first empirical analysis of how race/ethnicity and immigrant status jointly moderate the effect of GPA on adolescents popularity among peers -​ The relationship between GPA and popularity depends on both race/ethnicity and immigrant status -​ There was no evidence that any group experiences social sanctions for academic achievement -​ Penalty TUT 2 Why do we survey -​ Surveying allows us to directly sample and compare through identical questions and answer structures  -​ It is designed for representative samples, making it able to generalize the results of the sample to the larger population -​ Survey bias : occurs when the survey, for reasons of design or randomness, deviates systematically from the population -​ Target pop - complete group of interest -​ Sample - selected subset for study What is a survey -​ Survey design : creating instruments (questions, order, response choices) to collect self-reported data on a given phenomenon -​ Survey mode : how the survey is delivered to respondents (in person one on one, to a class by a teacher, online) -​ Response rate: critical for ensuring the data represent the target population -​ Self reported data : valuable yet susceptible to recall errors and social desirability bias How to capture change in surveys -​ Divided into 2 types -​ 1. At a moment in time, cross sectional -​ 2. Across time points, longitudinal What can surveys teach us -​ Chart examines immigrant status and the social returns to academic achievement -​ Independent GPA -​ Dep - Popularity Basic known, what can still go wrong -​ Like all other methods, surveys have random and systematic error -​ Among the major errors, however, are low response rates (particularly What are some questions you could ask in a survey? Which response choice would you have? Week 8: in-depth interviews and focus groups (multi-methods with surveys) Cross-sectional surveys -​ Are administered at one point in time, they provide a snapshot about how respondents think, feel etc Reinforcing the essentials : discovering, explaining and interpreting social patterns -​ Sociologists focus on levels of variables -​ For example: what is the overall level of psychological distress in the population? What proportion of the unemployed are classified as temporary layoffs? -​ Sociologists focus on associations between variables -​ For example : how is being temporarily unemployed associated with levels of psychological distress?  -​ Sociologists focus on the causal mechanism that “explain” the associations between variables -​ Unemployed -> financial strain -> distress -​ A word about causal direction: (social causation hypothesis vs social selection/drift hypothesis) -​ Qualitative methods help us to acquire more details about the meanings and processes behind the quantitative pattern - this can help enrich our understanding and interpretation of the data -> the stories behind the numbers Multi-method example : unprecedented shock and a causal chain of events and outcomes -​ How did this pandemic induced shock - being temporarily laid off affect workers’ mental health Final exam practice question -​ What is a research question you might ask about the effects of the trade war? -​ Following the steps of the research process, describe on way that you might go about testing that question -​ Explain how could you make your study mixed-methods with an example Theoretical model -​ The stress process model : when a disruptive event occurs, individuals often struggle to reestablish homeostasis, and the process can be wearing and taxing on individuals -​ Theoretically, a stressor like being temporarily laid off should elevate distress because it “ is involuntary, undesired, and unscheduled”. Being laid off from a job can also lead to a series of other chronic stressors An extraordinary context -​ In his elaboration of the consequences of stress, wheaton (1999) noted the importance of social context in shaping stress exposure - potentially altering its meaning and threat level -​ Given the extraordinary context surrounding the pandemic and the temporary nature of the job disruption, an alternative view - the “forced vacation” hypothesis - argues that people who become TLO might report lower distress compared to people who continued working during the same period -​ Social comparison processes : what happens when you believe that other people are “in the same boat” The data : the canadian quality of work and economic life study (C-QWELS) Reinforcing the essentials -​ Random samples were selected from a large national panel of individuals (maintained by the research firm angus reif) to match “a balanced sample matrix” by gender age and region -​ This is an example of proportionate stratified sampling: an approach where the sizes of the subgroup samples match their relative sizes in the population. This method is used to ensure that we have a representative sample of working adults Quantitative results -​ The first part of the study focused on the set of research questions in april 2020 survey data -​ Being TLO is not associated with financial strain -​ Being TLO is not associated with mastery -​ Being TLO is associated with lower levels of distress relative to their counterparts who remanded employed -​ These findings do not confirm the core predictions of the stress process model -​ By the time of the may 2020 survey, patterns shift: -​ Being TLO now becomes associated with elevated levels of financial strain and lower levels of mastery -​ This is evidence that an “honeymoon phase” of a forced vacation observed in april 2020 started wearing off by may -​ However being TLO is still not associated with elevated levels of psychological distress as predicted by the stress process model Multi-methods to help understand why empirical observations do not conform to theoretical predictions -​ Triangulation : using one research method to evaluate or extend the findings discovered with another method -​ Why did workers who were TLO not report more distress than their peers who continued working? -​ To reconcile this puzzle, the researchers add qualitative evidence to help better understand the quantitative puzzle -​ The researchers argue that the pandemic represents a rare context in which the stress process models predictions might not play out as expected A SHIFT TO QUALITATIVE APPROACH -​ FINAL EXAM TIP : not the shift in the approach and design from the quantitative part of the study - and the interplay between the quantitative and qualitative approaches Reinforcing the essentials: semi-structured in-depth interviews as a common qualitative approach to data collection  -​ In-depth interviews are ideal to probe how individuals interpret events in social life and unpack complex processes behind the nuances of experiences and perceptions -​ We want to understand the experience of being temporarily laid off : interviewed 47 participants [by zoon or phone] who were exploited at baseline and then were temporarily laid off during the period of april and may (or longer) in 2020 (not the difference in sample size and data collection method) qualitative/in-depth interviews -​ Semi-structured and unstructured interviews focus on generating rich qualitative detail about a topic -​ During in-depth interviews, researchers prioritize open- ended questions (which give respondents more flexibility to discuss what they think is important) and probes (predetermined or improvised follow-up questions) The basics: why is it called “semi-structured” the development and contents of the interview guide 1.​ Background information -​ First think back to before you were laid off, tell me about your job, job title -​ Think back to jan 2020. Sense of job security before pandemic 2.​ How you learn that you would be laid off -​ Can you recall when you first learned you would be laid off -​ How stressful was the experience? What would you have said [probe : reasons; meaning of stress 3.​ Experiences while being laid off -​ What was it that you missed most about work [probe : the people, structure, work] -​ Anything else that relates to your experience of being laid off - anything you missed that really matters Open-ended questions -​ Interview questions that give respondents more space to answer as they wish. In in-depth interviews, open-ended questions encourage interviewees to answer at length -​ In surveys, open-ended questions allow respondents to using their own words rather than choosing from categories ‘ Compare quantitative and qualitative approaches Structured survey In-depth interview To measure the sense of control, we asked : During the past year, did you ever feel like how much do you agree or disagree with the you had little control over things? Did you following statements feel like you had little control over things? -​ You have little control over the thighs Did you feel powerless? that happen to you a.​ How did being laid off shape that -​ No way of solving some problems you feeling? have b.​ Did anything help you keep a sense of  -​ Feel helpless with problems in life control? Index range from 1(low) to 4 (high) FINAL EXAM QUESTION : HOW COULD YOU PROBE ON THE STRUCTURED SURVEY ITEMS? Nonverbal cues -​ A respondents facial expressions, gestures, and other types of body language, which can give hints about their mood or their underlying feelings and thoughts about a topic. These can be data points -​ Verbal cues : nervous laughter, crying, etc Asking questions -​ Rapport : the sense of connection a researcher establishes with a study participant, which may encourage them to speak in greater detail about their thoughts, feelings, and experiences or to act more naturally while being observed -​ Simplify language : think about the theories and concepts you are interested in and how you might break down abstract ideas into statements that are clear, concrete, and vivid. Avoid jargon and academic language -​ “Sense of control” “work-life conflict” Using NVivo ( a data analysis tool), the researchers separately analyzed interview transcripts, coded themes, met to discuss themes, and continued reviewing transcripts and refining the core themes Qualitative findings 1.​ A temporary state : many had a pause from work and were protected from ongoing work stress; they suddenly had more time for self-care 2.​ Reduced internal attribution : the normativity of being TLO allowed workers to perceive job disruption as a societal-level phenomenon affecting the labour market, rather than due to personal qualities 3.​ Alleviated financial strain : financial strain linked to TLO was reduced by increased personal savings/reduced spending, government support, and employer-provided pay/benefits (countervailing forces and buffers?) Key take-away points -​ Guided by the stress process model, workers who were TLO should have experienced higher levels of distress relative to their counterparts who remained working. No confirmation of this pattern -​ Using qual and quan data, the authors advance the forced vacation thesis and identify mechanisms for why they found this paradox -​ The experience of being TLO- at least early in the pandemic - unfolded in social conditions that might have buffered against adverse mental health -​ The study provides an example of using inductive versus deductive Focus groups an example -​ Interviews conducted with a group of respondents at the same time. During a focus group session, one or more moderators will typically ask the group question about a particular issue or topic In-depth interviews versus focus groups -​ In-depth interview : the interviewee is more active while the researcher plays the role of listener and question-asker -​ By contract, focus groups are designed to spark discussion/debate in the group. Rather than just asking questions, researchers are actively trying to get participants to talk to each other. They observe interactions among focus group participants in addiction to listening to what each person has to say TUT 3 : interviews and depth A forced vacation uses mixed method design - this is a representation of triangulation Why triangulation? The forced vacation study -​ The forced vacation study combines in-depth interviews and surveying to understand why job disruption at the beginning of the pandemic did not lead to distress -​ The forced vacation in-depth interviews revealed 3 reasons why the layoff did not result in distress 1.​ Temporary state : expectations of return 2.​ Exemption from work stress : break from high pressure work environment 3.​ Lack of choice : no need to act Key takeaways for the importance of qualitative research 1.​ Qualitative interviews can provide more detail than quantitative surveys, by focusing on how individuals process, navigate and experience a given phenomenon 2.​ Address unanswered questions about the chosen phenomenon 3.​ Provide richer, more detailed data What is the depth in in-depth interviews -​ Depth is going beyond surface-level responses to explore participants experiences, feelings and perceptions -​ It's more than just asking questions - it is about probing from more specific information based on contextual cues and an established trust What is the deep versus shallow conversation to you -​ Probing, create a connection to ease the tension, build relationship to get more info, staying silent to volunteer information What are some ways researchers attempt to have deeper conversations Interview guides and probes -​ Interview guide are the sets of topics and questions that the interviewer uses to steer the conversation to points of interest -​ Probes are the follow ups to a question designed to gather more information or redirect the conversation back to the topic of interest Two types of interviewing -​ Interviews often take place one on one or in focus groups 1.​ One on one typically allows for more depth and structure to a conversation 2.​ Focus group risk, ironically, loss of focus. But they potentially gain more authentic interaction Week 9 : to learn about experimental research designs with a classic example Final exam practice questions -​ Compare and contrast the different types of research designs using example from the assigned readings -​ Surveys -​ In-depth interviews -​ Focus groups -​ Experiments Experiments : the basics -​ Researchers introduce a variable (independent variable, also known as the experimental stimulus or treatment) and then observe what happens to another variable afterward (dependent variable, also known as the outcome or response variable) -​ A deductive approach : “what effect does the independent variable have on the dependent variable?” experiments excel at answering the effect part of the question -​ Internal validity : how well a study shows if changes in the indep variable causes changes in the dependent variable -​ External validity : a study's ability to generalize and results obtained from its sample to its target population or beyond -​ Confounders : a variable other than the independent variable that may be influencing the dependent variable -​ Spurious relationship: a situation in which a relationship between 2 variables seem to exist but, in reality, they are correlated because of a third variable a confounder -​ Observational data : information that sociologists gather in the real work from surveys, in-depth interviews, ethnographic observation, and other methods that do not use an experimental design -​ Quasi-experiments : studies that approximate a true experiment design to improve their internal validity beyond what they could typically do with observational data. These  designs lack random assignment to experimental and control groups, which means they are more vulnerable to related threats to internal validity Experiments the basics -​ Random assignment : participants are assigned to either experimental or control groups at random - that is, an equal chance to be assigned to either group -​ Experimental group: participants who do receive the experimental stimulus or treatment. Their scores on the dependent variable can be compared to the control group to gauge the impact of the stimulus or treatment -​ Control group : participants who don't receive the experimental stimulus but are included to measure their scores on the dependent variable. Their outcomes can be compared to the experimental group to gauge the impact of the treatment Getting a job: is there a motherhood penalty? 5 important things 1.​ What are the main concepts and theoretical ideas that frame the hypotheses in this paper 2.​ What research designs did this study use to test the hypotheses 3.​ What are the main components in the procedures section of the paper 4.​ How does this study operationalize focal variables 5.​ How do researchers present or talk about their results? Key examples from this study. The puzzle (and problem) -​ In the workplace, mothers experience a disadvantage in rewards like pay and perceived competence -​ One study found that, visible pregnant women were judged to be less committed to their jobs, less dependable, warmer, more emotional -​ Key questions (mechanisms) : why would being a parents lead to disadvantaged in the workplace for women but not men Key concepts -​ What is status -​ A position in a set of things that are rank-ordered by a standard of value. Status structures or hierarchies are rank-ordered relationships between individuals -​ Status beliefs shape perceptions, interaction, and decision-making, with higher status actors potentially perceived as more worthy, esteemed, valued, and competent. Key theoretical idea -​ Status characteristics theory : -​ Defines a status characteristics as a categorical distinction among people such as a personal attribute (gender) or a role (motherhood) that has been attached to it widely held beliefs in the culture that associate greater status worthiness and competence with one category of the distinction Key theoretical ideas  -​ Performance expectations : the expectations that group member for their own and other performance on task -​ Group members who have a more highly valued state of a status characteristic are assumed by themselves and other group members to be more competent in general at most things than those who have a less valued state of that characteristic (stereotyping process) -​ Status characteristics (gender race education) systemically shape the appearance of competence and influence - and the standards observers use to determine ability -​ High0 status actors are expected to offer more competent performances, so they are often given behavioural and evaluating advantages -​ Performance expectations imply that people with status disadvantage in groups much work harder and perform better than status-advantaged members to attain status in the group (proving themselves worthy) Claims : status beliefs can create biases in organizational practices -​ If motherhood is devalued status in workplace settings, the researchers hypothesize that mothers will be judged harsher -​ If motherhood penalty exists because cultural understandings of the motherhood role conflict with cultural understandings of the “ideal worker” (versus the intensive mothering ideal) -​ Tension between incompatible cultural understandings (family devotion vs work devotion) leads evaluators, perhaps unconsciously, to perceive mothers as less competent, committed, or promotable -​ Important point: the tension between family and work roles occurs at the level of normative cultural assumptions and not at the mothers own commitment to work roles. It is the perceived cultural tension between the 2 roles. It is the perceived cultural tension between these 2 roles that leads to the hypo that motherhood is a devalued status in work settings Hypotheses -​ Mothers are perceived as less committed to work, employers will subtly discriminate against mother when making evaluations that affect hiring, promotion, and salary -​ Don't expect the same treatment to fathers because of cultural understandings of what it means to be a good father are not seen as incompatible with understandings of what it means to be a good worker Research design : experiments -​ We other think of experiments as being conducted in laboratories. In a lab setting, the researcher creates an artificial situation to manipulate variables -​ As a method, this is useful for testing if-then statements : if x then y will result -​ What is the “if x happens then y will result” in this study Procedures  -​ Study participants went to the lab individually, read a description of a company hiring for a marketing position, and examined application materials for 2 applicants who differed on parental status but were otherwise similar Procedures -​ The researcher paried application matericals by race and gender to generate four experimental conditons where participants rate one parents and one non-parent applicant who are either black men, black women, white men or women -​ How did the researchers manipulate parental status? -​ On resumes for the parent, they listed “parent-teacher association coordinator” under the heading “other relevant activiies” the non prent was described as the fundraiser for the local neighborhood association -​ The memo for the parent included “mother/father to tom and emily. Married to john/karen” the nonparent was described as simply married to john/karen’ -​ Study materials presented applicants as similar across a range of different qualities like career goals, educational history, past work experience, and described them as highly productive Procedures -​ They manipuate gender of the applicatn in the labratory experiment so that they can evaluate the claim that men are not penalized in the workplace for being fathers -​ They manipulate race of the applicant in the labratory experiment so that they can evalute whether their argument holds for both black and white applicants. -​ Male and female study participants were randomely assigned to one of the four eperiemntal conditions Measurment : dependent variables -​ Competence : participants rated applicants on seven point scales ranging from “not at all” to “extremely” capable, efficient, skilled, intelligent -​ Committed: participants rated how commited they thought the applicant would be relative to other employees in similar positions at the company -​ Ability : in what percentile would the applicant need to scare on his/her management profile exam in order for you to consider him/her for employment -​ Evaluation : what salay would you recommend for each applicant? What is the likelihood that an applicant would be subsequently promoted if hired? Would you recommend hiring the applicant Finding : mothers/nonmothers -​ Mothers were rated less competent and less committed than non mothers -​ Mothers were held to harsher performance standards than nonmothers -​ Mothers reommended starting salar was 11,000 less than starting salard ofered to nonmothers -​ Mothers rated less promotable and less liekly endorsed for management  -​ Participants recommended 84% of nonmothers for hire; they recommended 47% of mothers for hire Findings : fathers/nonfathers -​ Findingd consistent with the theoretical predictions of no fatherhood penalty -​ Fathers were actually advantaged on some measure -​ Applicants who were fathers were rated more committed to their job than nonfathers -​ Fathers were allowed to be late to work more often than nonfathers -​ Fathers were offered higher salaries than nonfathers Another key finding -​ African-american applicants were offered 6,800 lower salaries, on average, compared with those offered to white applicants -​ This difference i striking, especially since african-americans were not judged to be less competent or committed to work- they were simply offered lower salaries -​ The resumes in the african-american conditions were the same, except for the first names, as those used in the white scenario descriptions, so differences in qualifications do no explain the finding More findings -​ The researchers also conducted an AUDIT STUDY to test for hiring discrimination based on parental status -​ They adapted the applicantion materials from their labratory experiment to apply to over 600 real job ads -​ They found that prospective employers called mothers back about half as often as nonmothers -​ fathers , by contrast, were no disadvantaged in the hiring process -​ The findings of the audi study corresponded to the labratory study, providing converging evidence for the motherhood penalty across two studies employing different methods and samples The methods of the audit study -​ Resumes and cover letters from a pair of fictitious, equally qualified, same-gender applicants (both male or both female) were sent to employers advertising for entry- and mid level marketing and business job openings at a larger city newspaper over an 18 mother period -​ Job openings were randomly assigned to either M pair or F pair conditions. The same sex pairs contained one parent and one non parent, and parental status was counterbalanced across members of the pair -​ Monitored whether gender or parental status affected the odds that an employer called back an applicant Tut 4 : experiments What is an experiment -​ An experiment is a research design that involves manipulating one or more independent variables and randomly assigning (giving group one father resumes and group 2 mother resumes) participants to different conditions -​ The purpose of an experiment is to determine whether that manipulation, the treatment or change in the key independent variables ( whether a job applicant is a mother or a father) causes change in another variable (getting a job) Control vs experimental groups -​ Control groups is the participants or cases that do not receive the treatment -​ The experimental group is the one who does receive the treatment Internal and external validity -​ Internal validity ensures that the observed effects are due to manipulation (that the resume is from a mother or father) rather than external factors (labor market conditions, mood etc) -​ External validity refers to the degree to which the study can be generalized to a larger population Lab vs field vs natural experiments -​ A laboratory experiment is conducted in an artificial setting -​ Field experiment is conducted in a real-world setting with controlled manipulations -​ Natural experiment observes effects that naturally occurs where the treatment is not assigned by the researcher Internal - lab Survey research finds that mothers suffers a substantial wage penalty, although the CAUSAL MECHANISM producing it remains elusive Selection bias -​ Selection bias occurs when participants or cases are not representative of the population intended to be analyzed -​ You want to do a laboratory study on whether wearing bright colours reduce negative interactions Week 10 : review essential elements of sociological inquiry (EXAM QUESTIONS LOOK BACK) PRACTICE QUESTIONS BASED ON THE ASSIGNED READING What is an abstract -​ A part of academic articles that summarizes the main components of its contribution Why should we care about this topic -​ By understanding neighbourhood preferences, community groups, policy makers and planners may be able to adequately respond to community needs. However, rigorously studying such a dynamic and complex mix of subjective and objective processes is highly challenging -​ Its a multi-faceted research topic that can produce policy-relevant knowledge that affects quality of life (eviction rates) Main elements of the research process (cause, predict, explain) - (limitation and future research directions) Sociology sometimes relies on layers of different methodological approaches in the research process. An example is defining neighborhoods Microcosm - a small place, society, or situation that has the same characteristics as something much larger What is a conjoint survey experiment -​ Survey-based techniques that help determine the attributed people value an object or action -​ Repeatedly presents study participants with pairs of profiles made up of different attributes and asks them to choose between two profiles. Attributes are randomly varied -​ This randomization is crucial: it means that respondents see unusual combinations of attributes, as well as common combinations, thereby allowing researchers to untangle the causal effect of attributes that typically co-occur (safety and school quality) What are variables? And different ways researchers talk about relationships between them -​ They find that residents in low SES neighbourhoods share many of the same priorities as residents in higher SES neighborhoods like safety, transit, school quality, neighbourliness, public spaces, and building types -​ SES-based differences appear across a range of preferences including bike usages, local commercial spaces, and cultural and recreation facilities -​ They find a consistent, robust inverted relationship between place alienation and neighborhood satisfaction -​ Moreover, this relationship is not mitigated by socioeconomic factors, neighborhood conditions When considering place alienation and neighborhood satisfaction, they find a consistent, robust inverted relationship - this means that as place alienation decreases, neighborhood satisfaction increases  -​ Place alienation is defined by the proximity between a respondent perception of their actual neighborhood and the ideal neighborhood attributed revealed by the conjoint experiment The so what question - illustrate distinctions between independent, dependent and moderating or mediating variables The inverted relationships between place alienation and neighborhood satisfaction is not mitigated by socioeconomic factors, neighborhood conditions (control variables) Week 11: review essential elements of sociological inquiry using the assigned reading as an example EXAM QUESTIONS Reading for the week - what is the main topic of focus in this study -​ What is discrimination? -​ It involves treating people differently, often negatively, based on their group affiliation rather than individual merit -​ It often targets individuals or groups based on characteristics (race age gender) -​ It can be overt (direct) or subtle (indirect) -​ It can be intentional or unintentional Why should we care about this topic -​ Discrimination affects a range of workplace processes and outcomes (life chances) -​ Hiring -​ Compensation -​ Performance evaluations and rewards -​ The quality of everyday interactions Practice question : describe the main points of similarity and difference between the study in today's assigned reading and the study from week 9: getting a job: is there a motherhood penalty -​ Research shows that people often feel distress when they experience a potentially discriminatory incident but cannot classify it conclusively A central puzzle  -​ Some incidents of discrimination against women are obvious: it is immediately clear that a person is treated negatively because she is a woman -​ Some incidents of possible discrimination against women are not obvious : it is not immediately clear that a person is treated negatively because she is a woman or for some other reason In this study, the researchers propose that the effects of such ambiguous incidents extend beyond personal emotional costs to include socially consequential action or inaction at work Components of mixed-method design -​ Prof doering and her colleagues used a mixed-method design to create and refine their research tools, with an initial qualitative approach (in-depth interviews) informing subsequent quantitative approaches (survey and vignette experiment) Research questions -​ Their mixed-methods approach allows they to ask -​ How do women see ambiguous incidents as affecting their work experiences? -​ And what actions do they anticipate taking in response to more versus less ambiguous incidents? Practice question -​ The first substantive lecture slide in this course (week 2) was titled “how do we know what we know” based on what you've learned in this course, how would you answer that question? How do we know what we know -​ Sociologists seek to contribute to and enhance knowledge about the social work -​ Research methods (methodology) -​ A systematic process of inquiry applied to learning about the social world -​ The goal of research methods is to construct a defensible version of reality Quantitative and qualitative methods -​ Quantitative methods: result in data that can be represented and condensed into numbers (aggregate, compare, summarize data) - -​ Less depth, more breadth - focus on a larger number of cases -​ Survey method - most common quantitative method in sociology -​ Qualitative methods are ways of collecting data that yield results such as words or text (in-depth interviews) -​ Gain in-depth understanding of a relatively small number of cases -​ Potential for greater richness in meaning than quantified data -​ Sometimes qualitative and quantitative methods are discussed in a way that suggests they are in opposition- this is an inaccurate and unnecessarily limited perspective -​ Complementary, not competing: yes these methodological approaches differ… but it is more appropriate to see them as having different goals, strengths and weaknesses Practice question : make an argument that quantitative and qualitative methods are not in opposition to each other - that they are complementary, not competing. How is the study by professor doering and her colleagues a good illustration of this Sampling for the in-de[th interviews -​ A sample of 31 professional women. They selected professional women because they tend to report higher levels of discrimination and have the realistic (but difficult) option of confronting discrimination, which allows the researchers to observe responses -​ They began recruiting participants by asking acquaintances and colleagues for referrals and then used snowball pampling to ensure sample variation -​ Snowball sampling is a form of convenience sampling and a nonprobability sampling approach : a technique in which a research draws a sample from part of the population that is convenient to obtain - for instance, because potential interviewees are located near the researcher or otherwise are readily available The purpose of the in-depth interviews -​ 31 in-depth interviews to gain knowledge about how women experience and respond to uncertainty about ambiguous incidents -​ They assessed whether women consider ambiguous incidents to be an important aspect of their workplace experiences and careers. They also wanted to understand how women made sense of ambiguous incidents -​ What contextual information did they consider? -​ Whom did they consult, and how did this shape their interpretations? -​ What actions and workplace changes occurred following ambiguous incidents? -​ They compared details from reports of both ambiguous and obvious incidents Survey -​ In-depth interviews (N=31) showed that ambiguous incidents were common and occurred more frequently than obvious incidents. But maybe their interviewees’ impressions reflected that of a small, select group of individuals -​ Based on interview findings, the researchers then designed a survey of professional women (N=600) to evaluate whether respondents in a large, diverse sample experienced ambiguous incidents as regularly as their interviews revealed. In this way, their survey approach serves as a confirmatory extension of their interview-based findings (triangulation) Ethnography -​ Qualitative method of studying a phenomenon within its social context by doing first-hand observations and providing detailed (thick) descriptions -​ The setting where an ethnographer conducts observations is called a field site, and the act of visiting that space to do research is “going into the field” -​ In the field, they take extensive notes and describe their experiences, trying to capture a detailed portrait of the locations and people they observed  -​ When they leave the field, ethnographers analyze the rich information they gathered, often with an interpretive approach - that is, making sense of what people say or do, and the social significance of those words and actions An unexpected qualitative finding within the survey design -​ In their survey data, they asked an optional open-ended question in which they invited respondents to describe their experiences of ambiguous or obvious gender discrimination. Respondents provided detailed descriptions of ambiguous incidents, many resembling the content from the in-depth interviews The vignette experiment -​ Interviews showed that uncertainty shapes how women respond to potential discrimination, and survey data showed the frequency of ambiguous incidents in a larger sample, but neither of these approaches could isolate the causal effect of uncertainty from the influence of other situational factors. -​ They designed a vignette experiment to measure the effects of uncertainty in perceptions of discrimination on anticipated behaviour while holding all other aspects of the incident constant (random assignment) -​ This approach made it possible to examine how women anticipate responding to incident that differ only in their degree of ambiguity but are otherwise identity (HINT motherhood penalty comparison) -​ From a purely empirical perspective, the ideal experiment to study how women respond to discriminatory incident would be a field experiment so researchers could monitor women's behaviour in a natural workplace setting -​ Exposing people to actual discrimination and monitoring their behavior would be unethical. So they used a vignette experiment. What is a manipulation check -​ Participants read a vignette with either the less ambiguous manipulation or the more ambiguous manipulation and then indicated whether gender discrimination occured in the situation using a 100-point sliding scale (0=obviously yes, 50 = unclear, 100= obviously no) -​ Within each vignette, the average response in the ambiguous condition was near the “unclear” anchor, whereas the average response in the ambiguous condition was closer to the “obviously yes” anchor Summary -​ What are 3 main findings from the different research approaches? These show the usefulness and power of mixed-methods for producing knowledge -​ Interviews - show that women struggle with how to interpret and to ambiguous incidents -​ Survey - data show that women experience ambiguous incidents more often than incidents they believe were obviously discriminatory -​ The vignette experiment - with incidents that are obviously discriminatory, women anticipate taking actions that make others aware of the problem. By contrast, with ambiguous incidents, women anticipate changing their own work habits and self-presentations