Download Social Research Methods: A Comprehensive Guide to Understanding and Conducting Research and more Exams Sociology in PDF only on Docsity!
Bryman - Social Research Methods
Why is social research important? - Social research allows us to test ideas/ assumptions in a structured way. Knowing how to perform/evaluate research is important whenever you get the question: Why do you think so? Hypothesis - an informed speculation, which is set up to be tested, about the possible relationship between two or more variables. A hypothesis should be testable and falsifiable. Context of social research methods - social research and social research methods are embedded in wider contextual factors. They are not practised in a vacuum. Contextual factors -
- Existing theories (used by social scientist to understand the social world) influence what is researched and how the findings of research are interpreted.
- A researcher must be acquainted with the existing knowledge (literature) on the topic of interest, to build on it and avoid unwanted repetition.
- The choice to conduct either inductive or deductive research has implications on the relationship between theory and research and thus on the research.
- Epistemological and ontological considerations
- Values and ethical considerations
- Emphasize on implications for practice in social research (evaluation & action research)
- Political contexts
- Training and personal values of researchers Epistemology - a theory of knowledge - concerned with what should be accepted as knowledge & how knowledge should be acquired. Epistemological considerations - raise questions on how the social world should be studied and whether a scientific approach is the right stance.
- E.g. 'Is it necessary to focus on individual qualities of people and their social institutions and to avoid scientific research models?' Positivism - an epistemological position that advocated the application of natural science methods to the study of social reality; universal laws, experiments, and numbers. -> associated with quantitative approach Interpretivism - focuses on the importance of the interpretation and meaning of social reality. Subjective meaning plays a crucial role. -> associated with qualitative approach
Ontology - a theory of the nature of social entities - considers the nat. Ontological considerations - consider the nature of social phenomena.
- E.g. 'Are social phenomena inert and beyond our influence or the product of social interaction?' Objectivism - social phenomena exist independently from social actors (they are "mind- independent"). -> You can objectively observe social phenomena. -> E.g. culture (can be observed in an objective manner); emotions (hard-wired in the brains of all animals) Constructionism - social phenomena are dependent on social actors and social interaction; focus on change; subjectivity (also as a researcher!) -> You can not objectively observe. -> E.g. culture; emotions (depend on experiences of individuals). Deductive approach - research is conducted with reference to hypotheses that are inferred from theory (Theory/Hypothesis ---> Findings) -> Associated with quantitative approach Inductive approach - theory is generated from research (Findings ---> Theory) -> Associated with qualitative approach Quantitative research (commonly deductive) - research strategy that relies on the collection of numerical data, and the statistical analysis of those numbers. -> E.g. (i) Perform an online survey with a questionnaire about environmental attitudes among players and non-players. (ii) Perform an experiment in which people play a serious game in the lab after which you measure environmental attitudes. Qualitative research (commonly inductive) - research strategy with an emphasis on the description or interpretation of data, commonly expressed in words. -> E.g. (i) Become a member of a gaming community and have conversations with players. (ii) Form a focus group inviting people who play serious games. Interdisciplinary research - integrated data, methods, tools and concepts form different sciences in order to create a common understanding of a complex issue or question. -> E.g. integration of law and psychology in research on 'unbiased sentencing'. Elements of the process of social research -
Survey research - a cross-sectional design in relation to which data is collected predominately by self- administered questionnaires or by structured interview on a sample of cases drawn from a wider population and at a single point in time in order to collects a body of quantitative or quantifiable data in connection with a number of variables. These are then examined to detect patterns of relationships between variables. Qualitative methods of sampling - purposive, snowball, and theoretical sampling. Purposive sampling - non-random sampling of people who are relevant to the RQ. Snowball sampling - sampling where subjects propose other subjects/spread the word to other subjects. Theoretical sampling - sampling guided by the emerging theory. Data collection - Structured interview: a research interview usually in the context of survey research in which all respondents are asked exactly the same questions in the same order with the aid of a formal interview schedule. Participant observation: research in which the researcher immerses himself in a social setting for an extend period of time, observing behaviour, listening to what is said in conversations both between others and with the field worker, and asking questions (e.g. ethnography). Semi-structured interviewing: a term covering wide range of interview types. Data analysis - The application of statistical techniques to data that has been collected. Transcription: creating a text version of a recorded interview or focus group session. Thematic analysis: the extraction of key themes in one's data in connection with qualitative data. Coding: a process whereby data is broken down into its component parts and those part are then given labels. Codes: numbers assigned to data about people or other unit of analysis when the data is not inherently numerical. Quantitative analysis: - the application of statistical techniques to the data to test your hypotheses. Qualitative analysis: - analysis that focuses on the description, understanding, or interpretation of information. Writing up - the reporting and dissemination of results. The messiness of social research -
although we can attempt to formulate general principles for conducting social research, we have to recognize that things do not always go entirely to plan. Symbolic interactionism - a theoretical perspective in sociology and social psychology that views social interaction as taking place in terms of the meanings actors attach to action and things. Grounded theory - an iterative approach to the analysis of qualitative data that aims to generate theory out of research data by achieving a close fit between the two Empiricism - a general approach to the study of reality that suggests that only knowledge gained through experience and the sense is applicable. -> Ideas must be subjected to rigorous testing before they can be considered knowledge. Naive empiricism - the belief that the accumulation of facts is a legitimate goal in its own right. Naive/empirical realism - through the use of appropriate methods, reality can be understood. -> Assumes that there is a perfect correspondence between reality and the term used to describe it. Critical realism - a specific form of realism that recognizes the reality of the natural order and the events of the social world and which holds that the social world can only be understood when identifying the structures that generate these events. Retroductive research - making an inference about the causal mechanism that lies behind and is responsible for regularities that are observed in the social world. Operationalization (How do you make your research question testable?) - Translate the theoretical concepts in your research question into measurable and controllable variables (also called indicators). Variable - an attribute in terms of which cases vary. Constant - an attribute that does not vary. Dependent variable - what you measure or observe (this is not changed). Independent variable - what you change, control or manipulate in order to measure the effect on the dependent variable.
-Interaction of history and treatment -Interaction effects of pretesting -Reactive effects of experimental setting Ecological validity - findings are applicable to everyday life. Measurement validity - indicators really measure the concept in question. -> face validity: does the measure reflect the concept in question, e.g. does it reflect depression at face value? -> concurrent validity: do we also see expected effects/correlations with variables not relevant to study? -> convergent validity: are the findings of a measure comparable with findings from other methods that measure the same concept? Replication/reproducibility - the degree to which the results of a study can be reproduced (vs. 'reliability' which shows the stability of a measurement at different times). Research design - provides a framework for the collection and analysis of data. -> The choice of research design reflects the priority being given to a range of dimensions in the research process. Cross-sectional design - a research design entailing the collection of data on a sample of cases in connection with two or more variables and at one point in time. -> Yields: Quantitative data -> Examines relationships between variables (i.e., correlation) -> Doesn't provide conclusions about causality (= weak internal validity)! Longitudinal design - surveys the same sample across two or more different time points. -> E.g. Panel study & Cohort study Comparative design - Comparing two or more cases, or two or more samples. Often quantitative in survey form, but can also be qualitative. -> E.g., cross-national research, cross-cultural research, cross-institutional research (Do people in different cultures recognize emotions similarly?). Panel study - longitudinal-design study of a large sample that is followed over time (study age and cohort). -> E.g. Understanding society; studying a panel of 40.000 households from the United Kingdom; adults are interviewed every 12 months on finances, employment, expectations, family and friends etc. Cohort study -
longitutional-design study of a sample consisting of a group that experiences some event (such as being born) in a selected time period (can only study age). -> E.g. The Up series: 8 documentaries spanning 49 years following the lives of fourteen British children since 1964. Classic experimental design (randomized controlled trial) -
- Presence of a control (or no-treatment) condition (i.e. placebo)
- Random assignment
- Potential to draw causal conclusions!
- Field experiments vs. laboratory experiments Random assignment - the random allocation of subjects to the experimental group and the control group. Quasi experimental design - studies that have characteristics of experimental designs but do not fulfill all requirements (e.g. no random assignment). -> E.g., when evaluating organizational programmes or interventions Natural experiment - experimental and control condition is determined by nature or other factors. -> E.g. fishing example Biases in research - experimenter bias, demand characteristics, reactive effect Experimenter bias - the expectations of the experimenter may influence the responses of the participants or the data collection in general. Demand characteristics - participants in experiments may adjust their behavior/responses according to what they believe is expected. Reactive effect - bias in responding because participants know that they are being studied. Single-blinde experiment - information that could bias the results is withheld from the participants. Double-blinde experiment - information that could bias the results is withheld from both the participants and the experimenter. Meta-analysis - a quantitative statistical analysis of several separate but similar experiments/studies in order to test the pooled data for statistical significance. Steps in the publication process - -Decide on journal
Pros and Cons of structured interviews - Pros: -interviewer can probe/prompt the respondent -interviewer can help clarifying -personal Cons: -interviewer can evoke social desirable responses -slow and expensive -interviewer effects -interviewer variability Pros and Cons of self-report survey (vs. interview) - Pros: -No interviewer effects/variability -Convenient -Quick & cheap -Sensitive questions Cons: -less room for open-ended/complex questions -Cannot ensure that the 'right' person answers -Cannot collect additional data -Risk for respondent fatigue -Greater risk of missing data -Lower response rates Pros and Cons of online questionnaires (vs. paper & pencil) - Pros: -Quick & cheap -Many options for formatting -Broad range -Data accuracy -Control the flow of the questionnaire Cons: -Lower response rate -Restricted to online populations Types of questions - -Personal-factual information (including behavior) What is your country of citizinship? -Factual questions about others How often does your partner cook dinner? -Normative standards or values Do you support the death penalty? -Knowledge Who is the current president of the United States? -Attitudes Did you like maths in high school? -Beliefs Should the EU have stricter immigration laws? -Behavioral intentions
Would you invite this person for a dinner party? -Feelings/sensations How angry does this movie clip make you feel? Open questions - a question that does not present the respondent with a set of possible answers to choose from. Pros: -respondents answer on their own terms -allow for new, unexpected responses -exploratory Cons: -time consuming -difficult to code Closed questions - a question that presents the respondent with a set of possible answers to choose from. Pros: -quick and easy -precoded -easy to compare answers Cons: -restrictive range -not exhaustive Common errors in survey questions -
- ambiguous terms (e.g. Do you run often?)
- long questions
- double barreled questions (e.g. Do you agree with laws regulating euthanasia and abortion?)
- general questions (e.g. Are you happy with your life?)
- leading questions (e.g. What should we do to stop the problem of people smoking in their private homes?)
- double negatives (e.g. Would you not choose to not eat genetically modified food?)
- technical terms/jargon Characteristics of questionnaire answers -
- mutually exclusive
- exhaustive Likert item - a statement combined with an answer scale that holds a number of ordered and labelled answer possibilities. -> Variability in: -Number of scale points -Scale point labels -Presence of a "neutral" scalepoint Visual analogue scale - response on a continuous line with marked (end)points.
Random sampling - a sample that reflects the population accurately because each unit has an equal probability of selection. E.g. the sample has similar demographic characteristics as the population. Sampling bias - a distortion in the representativeness of the sample that arises when members if the population (sampling frame) stand little or no chance of being selected for inclusion in the sample. -> Caused by the use of a non-random sampling method Probability sample - sample selected using random sampling. Types of probability samples: simple random sample, systematic sample, stratified random sample Systematic sample - each unit is selected from the sampling frame according to fixed intervals (e.g. every 5th unit). Stratified sampling - each unit is randomly sampled from a population that has been divided into categories (strata). Sampling error - error in the findings derived from differences between random sample and population. -> Note that sampling error can occur even with a probability sample! -> As sample size increases, sampling error decreases. -> The greater the heterogeneity of the population the larger a sample will need to be. Non-sampling error - error in the findings due to the differences between the population and the sample that arises either from deficiencies in the sampling approach (e.g. due to inadequate sampling frame) or non-response, or other methodological & analytical mistakes. Non-response - refers to the event where some members of the sample refuse to cooperate, cannot be contacted, or for some other reason cannot supply the required data. Census - the enumeration of an entire population. Attrition - sample gets smaller in number because participants drop-out of the study for various reasons. Convenience sample - the researcher simply uses what he/she can get (i.e., what is available)
-> Convenience samples are rarely representative Convenience samples in psychology - it is very expensive, and impractical to use a true representative sample in psychological laboratory experiments as psychology seeks principles of behaviour that should hold for all humans. Purposive sample - researcher may look for subjects with certain characteristics, because these characteristics are relevant to the research question. Matching - the matching of subjects across different samples on certain characteristics (i.e., gender, age, economic status). When do you need a representative sample? - when the research explicitly aims to generalize its findings to a real-world population. Content analysis - an approach to the analysis of documents and text that quantifies content in terms of predefined categories. Sources for content analyses - -Mass media (e.g., TV, newspapers, internet) -Written sources (e.g., books, plays) -Oral sources (e.g., speeches, radio) -Visual data (e.g., photographs, paintings) -State documents (e.g., policy documents) -Personal documents (e.g., tweets, personal archives) Types of content that can be analysed - -Significant actors -Words (e.g., frequency) -Subjects and themes -Dispositions (e.g., ideology) -Context (e.g., placement of the article) Intra-coder reliability - the consistency with which a researcher codes. Inter-coder reliability - the consistency (amount of agreement) between the coding of different researchers. Characteristics of coding schemes - -Mutually exclusive categories -Exhaustive categories -Clear instructions to coders Structured observation -
Characteristics of experimental research - -Study effects in isolation -Control important factors -Draw conclusions about causality (experiments allow for conclusions about causality) -> Note the importance of internal validity for experimental designs! Steps in experimental research - -Literature review -Concepts and theories -Formulate a research question -Formulate a hypothesis (a priori) -Decide on your sample -Collect data -Analyze data (test your hypothesis) -Report results Confirmatory research - research that tests priori hypothesized relationships between variables. Exploratory research - research that explores data for possible relationships between variables. Control group/condition - group in which variables that also occur in the experimental group/condition are kept the same, except the independent variable of interest. -> If you find a difference between groups, then you have to be sure that this difference can be attributed to the manipulation, and not to other factors. -> Counters threats to internal validity such as: maturation, expectation, learning, habituation, etc. Random assignment to conditions - each subject has an equal probability to be assigned to one or the other group/condition. -> Neutralizes the effect of individual factors (e.g., personal experience, individual traits, severity of symptoms). Thus, counters threats to internal validity by selection. Between-subjects variable - different participant groups get different manipulations. -> E.g. varying the following between subject groups: -Films/music/vignettes/puzzles/newspaper articles -Sensations/food/drugs -Instructions Manipulation check - test to see whether a concept was successfully manipulated. Within-subjects variable - every participant gets all of the different manipulations. -> E.g. varying the following for every subjects:
-stimuli (words/pictures/sounds/sensations) -tasks (perspective, strategy, judgment) -time Stimuli can be presented randomly or counterbalanced. Counterbalancing - vary the order within manipulations to counter order effects. -> Can also be applied to questionnaires, response buttons, stimulus position, etc. Complex designs - (i) Experimental designs can combine multiple (within & between!) independent variables: -> E.g. 2 (agency vs. no agency) x 2 (life worlds intersecting vs. detached) (ii) Experimental designs can combine manipulable and non-manipulable independent variables: -> E.g. 2 (angry vs. happy facial expression) x 2 (male vs. female subjects) Special case: Deception - Researchers sometimes deceive participants about: -the truth of a manipulation -the goal of the study as part of the manipulation or to counter demand characteristics. Coverstory - informing the participant about a fictional goal of the study, to conceal the true goal of the study. Advantages of laboratory experiments - Controllable, thus: -practical for random assignment -practical for the implementation of your independent variable (easier to ensure internet validity) -easier to replicate Advantages of online experiments - Online communities (e.g., facebook) and crowdsourcing internet marketplaces (e.g., MechanicalTurk) provide with new opportunities to perform experiments. Disadvantages of experiments - -Lack of external validity -Lack of ecological validity Pitfalls: -Bias through demand characteristics -Experimenter bias Nudging - little "push" towards certain behaviour based on insights from psychology and behavioural economics. -> Practiced by gov't agencies and NGOs, etc. -> E.g. Placing target into toilet bowl
-aim to understand rather than to generalize Characteristics of successful interviews - -Knowledgeable -Structuring -Balanced -Ethically sensitive Focus group - a form of group interview in which there are several participants. -> Emphasis on a fairly narrow topic & on interaction within the group to construct joint meaning Ethnography - research in which the researcher immerses himself in a social setting for an extend period of time, observing behaviour, listening to what is said in conversations both between others and with the field worker, and asking questions. Divide between Quantitative and Qualitative research - Divide is not that strong/clear
- Quantitative researchers also study the social meanings people give to the world
- Reports of qualitative research rarely demonstrate that interpretative understanding has been achieved -> Mixed methods are often applied. Also, researchers may perform a quantitative analysis of qualitative data, or vice versa. ETHICS - Content: Overview of history of ethics in science, important components of ethical research -Identify ethical issues in the social sciences -Understand the importance of informed consent -Understand and reproduce important aspects of scientific integrity Ethical issues in research area - -Research involving vulnerable groups e.g., children, elderly, prisoners -Research involving sensitive topics e.g., drug use, sexual deviance -Research involving secure/confidential information e.g., social security or health information -Research involving risk e.g., pain, humiliation, stress Ethical issues in research methods - -Open interviews: unstructured asking about sensitive topics -Ethnography: what to do with sensitive information? -Observation: privacy (people don't know they are observed) -Laboratory research: people are manipulated Nuremberg code (1947) -
Voluntary and informed consent of human subjects is absolutely essential. Participants need to be informed of the nature, duration and purpose of the experiment, including the risks and benefits. Participants have the right to stop the experiment at anytime. Deception - -Deception is often used (in mild form) in social psychology experiments. -People cannot be informed fully before the start the experiment. -Participants are informed (i.e., debriefed) about the deception and the studies goals and hypotheses when the study is finished. Scientific misconduct - -Fabrication of data -Falsification of data -Plagiarism -Report duplication -Violation of ethics Confirmation bias - The tendency to search for or interpret data in such a way that it confirms your expectations/predictions, and to pay less attention to data that disconfirms your own expectations/predictions. Replication - repeating a scientific study to test whether the findings are stable. Direct replication - testing the same hypothesis while aiming to fully reproduce the conditions of the original study. Conceptual replication - testing the same hypothesis while changing key elements in the design, such as the independent or dependent variable. The importanceof replication - "Replication is a means of increasing the confidence in the truth value of a claim" Conflict of interests - a conflict between the researcher's obligations concerning ethics and scientific integrity and financial relationships with industry, sponsors, etc.
