Sampling & Research Methods: Understanding Experimental Designs in Social Sciences - Prof., Study notes of Introduction to Sociology

Download Sampling & Research Methods: Understanding Experimental Designs in Social Sciences - Prof. and more Study notes Introduction to Sociology in PDF only on Docsity! Sampling 09/27/2013
* The Basics  The unknown (e.g. population) & what is to be studied (e.g. the sample)  Historical blunder: FDR
I. Why sample?  1. Representing the many with a few o a. Extract a small collection of cases from a larger population  does NOT have to be people  2. Efficient & cheap way to do research  3. Samples are ALWAYS estimates o a. The goal: maximize sampling accuracy

II. Types of Sampling Schemes  1. Representative (or Probability) sampling o a. Quantitative research o b. Based on “probability” theory for statistics o c. Allows for generalization to a larger population, or reliability  2. Non-probability o a. Qualitative research o b. Seek in-depth discussion from sample studied o c. Low reliability, higher validity

III. Non-probability Sampling (types)  1. Available subjects (convenience or Haphazard) sampling o person-on-the-street; whoever’s available o high risk of error in the sample  2. Purposive (or Judgmental) sampling o researcher’s need for unique cases o helpful for difficult to reach populations (handicapped; homeless)  researcher’s theoretical interests  3. Snowball sampling (Network, chain referral) o Social associations: one contact produces many (socio-grams: matrix of connections) o Helpful for difficult to reach populations  4. Quota sampling o sample relevant characteristics in the population (create variation) o Nonprobability or probability w/in each category  5. Informant o specific/targeted sampling  6. Deviant Case: Extended Case Method o Purposive (or theory-driven) sampling that seeks out deviant cases o Challenges theory by extending the analysis  7. Sequential Sampling Experimental Research 10/4/13
*introduce stimulus, observe the response, provides nice environment for measurement  stimulus (independent variable), response (dependent), measure change that takes place
I. The Power of Experimentation  1. Useful tool for determining causal explanations o A tool of the natural sciences & quantitative research, and Behaviorism  2. Maximizes the researcher’s CONTROL over the environment o Offers the test of causation by…  Controlling temporal ordering  Eliminating rival explanations  Allows for statistical measurement
II. Logic & Experimental Components  1. Start with a hypothesis (specific AND testable)  2. Sampling/Random assignment of “subjects” into groups o Experimental “treatment” group o Control “comparison” group  3. Pretest: establish a benchmark  4. Introduce stimulus or treatment (IV)  5. Observe (measure) changes (DV)  6. Posttest: make before and after comparison
III. Why randomize?  Eliminate researcher bias  Use of probability theory (randomize error)  Allow for unbiased comparison b/w experimental & control groups  Random assignment vs. matching o What is relevant?  Double-blind experiments (groups are randomized to the researcher)
IV. Internal Validity in Experiments  1. Internal Validity: eliminate alternative explanations for the dependent variable  2. Threats to internal validity o a. Selection bias: violation of equivalence o b. History: event b/w pre & posttest o c. Maturation: changes in subjects o d. Testing: pretest biases the experiment o e. Mortality: dropout of subjects o f. Instrumentation: change in instrument over time (new statistical technique) o g. Statistical regression: regression to the mean o h. Contamination (Compensatory Behavior)  Subjects in comparison groups will communicate  Control group emulates the experimental group o i. Experimenter expectations (demand characteristics) Survey Research 09/27/2013
*Survey as Self-Reporting  a. One’s behavior  b. Attitudes & Beliefs  c. Characteristics and Attributes  d. Expectations, Hopes, Desires  e. Self-Classification  f. Cultural Knowledge

I. What is survey research?  1. Research tool used to gather data based on self-reporting from “respondents” o a. What people say NOT what they do  2. Opportunity to research a broad or large population  3. Provides ability to compare across populations & time o a. Standardize instrument to produce variation in responses  4. Technique has a very long history (think the U.S. Census), but it was popularized around WWII o a. Government commitment to large-scale & statistical social science research o b. Crowded out case study & field-based research o c. Today a problem of overuse  Can become non-theoretical “market research”

II. Logic of a Survey  1. A “positivistic” method usually coupled w/ sampling and used in quantitative research  2. Sample many respondents asking the same question or questions o a. Survey questions ARE measures of variables (operationalizing)  3. Seek statistical correlations through “control variables” o a. Elaboration Model

III. Types of Survey Research  1. Mail: most popular o a. Advantage: cheap & confidential o b. Disadvantage: low control & response rate  2. Telephone o a. Advantage: easy access to respondents o b. Disadvantage: costly, limited time for open-ended questions o c. LAN line “extinction”  3. Face-to-face o a. Advantage: highest response rate & in-depth o b. Disadvantage: very costly & labor intensive  4. Web Surveys o a. Adv: Immediate returns, inexpensive, & becoming popular o b. No control over WHO responds; misses those w/out internet access

IV. Steps in Survey Research  1. Develop theory, research question, & hypothesis  2. Identify target population & sampling scheme  3. Construct a survey “instrument” or questionnaire o a. Questions should avoid confusion & be written w/ respondents in mind  4. A researcher operationalizes their concepts/variables as SURVEY QUESTIONS o A. Where your hypotheses are tested  5. Question types: Closed-ended; open-ended; and contingency questions  6. Conduct a “pilot” study (optional)  7. Send out surveys or conduct an interview “schedule” for administering of questions  8. Plan the recording and coding of responses o a. Consider coding AS you write questions  9. Collect and analyze data o a. 75% is a GOOD response rate

V. Questioning Survey Questions  1. Keep questions: clear, simple, and directed to your respondent  2. Avoid: o a. jargon (prestige/expert) o b. ambiguous or leading questions o c. emotional language & prestige bias  Disadvantages:  Variation can be difficult to interpret  Difficult to code  Time consuming to analyze

VI. Interviews & interviewer Role: “conversation w/ a purpose”  1. Stages in Interviewing o a. Establish a special relationship (unfamiliar w/ the need for disclosure) o b. Asking questions…probing o c. Record answers (can be intimidating) o d. Exiting the encounter  2. Avoid interviewer bias o a. Nonjudgmental: be neutral & objective o b. Appearance: fit in o c. Confidentiality: explicit o d. Sensitivity: take advantage of the human interaction

VII. Survey Ethics  1. Ethics in Survey Research o a. Confidentiality: the privilege of privacy o b. Always voluntary o c. Pseudo-survey: manipulation of respondent, misuse of information o d. Intentional designs (question phrased with answer revealed)

VIII. Index and Scales  1. Measure dimensions and indicators of concepts (intensity & rank- ordering) o a. Index a composite of several scores/attributes (or variables) – think SES: Income, Education, Occupation o b. Scales (preferred): assign scores to the pattern of survey responses  2. Depending how complex a concept, it may require multiple indexes and scales to measure VIIII. Types of Scales  1. Bogardus Social Distancing o a. Social Matrix (dimensions of interaction) o b. Based on “level of comfort” o c. Willingness to interact or be close  2. Thurstone: numerical scale (e.g. 1-10) o a. A group of “judges” are given a set of indicators of a concept to rank  the agreement is used in a survey  interval level: equivalence  3. Likert: agree…disagree o a. Ordinal level: additive & ranked o b. A measure of intensity o c. include a “neutral” response…?  4. Semantic Differential Scale o a. Use opposite adjectives to measure feelings or thoughts on a topic (e.g. Obamacare: compassionate vs. obtrustive) o 09/27/2013