SOAN 3310

16 Pages
Unlock Document

Sociology and Anthropology
SOAN 2120
Scott Brandon

Soan Exam Chapter 6 Survey Research What is a survey  measures many variables, tests hypotheses, and infer temporal order from questions about past behaviour, experiences or characteristics, expectations, self-classification, knowledge Steps for conducting a survey  Step 1: develop hypothesis, decide survey, write questions, and decide on response categories and design layout Step 2: plan how to record data, pilot test Step 3: decide on population, get sampling frame and decide sample size Step 4: conduct interviews and record data Step 5: enter data ad recheck it – review data Step 6: present findings How to Write a Good Question  3 principles: clear, simple and keep respondent’s perspective. Dilemma: everyone comes from different backgrouns so interpretation varies 12 things to avoid when writing survey questions: o Avoid jargon, slang, and abbreviations o Avoid ambiguity, confusion, and vagueness o Avoid emotional language o Avoid prestige bias o Avoid double-barreled questions (two or more questions joined together) o Do not confuse beliefs with reality o Avoid leading questions (one that leads the respondent to choose one response over another by its wording) o Avoid asking questions that are beyond respondents’ capabilities o Avoid false premises o Avoid asking about intentions in the distant future o Avoid double negatives o Avoid overlapping or unbalanced response categories (make response categories or choices mutually exclusive, exhaustive, and balanced)  Mutually exclusive: response categories do not overlap  Exhaustive: every respondent has a choice – a place to go  Balanced: outstanding, excellent, very good, satisfactory Aiding respondent recall  memory decreases over time. Provide aids ex: a fix time frame or location references Types of questions and response categories  Threatening questions: ask about sensitive issues or one that can threaten their presentation of self (ex. sex, drugs, alcohol, mental health). People try to protect their ego so they overreport good things (giving to charity) and underreport bad things (illegal stuff) Social desirable questions: bias occurs when respondents distort answers to make their reports conform to social norms. Knowledge questions: don’t use them. it can be threatening because respondents do not want to be ignorant Skip or contingency questions: two or more part question (the answer to the first part determines which of the two different questions a respondent next receives). Avoid asking irrelevant questions. Open vs. Closed questions  Open: questions to which respondents can give any answer. They are unstructured. Closed: gives respondent fixed responses to choose from. They are fixed and structured. Usually used in large-scale surveys (do not want many different answers) Partially open: there are fixed responses or respondent can offer their own answer (ex. Other) Best = to use both Non-attitudes and Middle Position  includes options like not sure (neutral), don’t know (middle) and no opinion (non-attitude) Three kinds of attitude questions: Standard-format  does not offer don’t know, respondent must volunteer it Quasi-filter  gives “don’t know” alternative Full-filter  ask if there is an opinion and then asks for the opinion for those who state that they do not have an opinion Wording issue  use simple vocabulary and grammar to minimize confusion & effects of specific words and phrases Length of survey/questionnaire  no proper length. They prefer long questionnaires because they are more cost effective Organization of questionnaire  sequence questions to minimize confusion. Make opening questions pleasant and easy. Do not ask boring questions first and do not end with threatening questions Order  can influence respondent answers. Context effects use funnel sequence of questions (ask more general before specific) OR divide number of respondents in half, and give half of questions in one order and the other half the alternative order, then examine results to see trends Questionnaire layout  easy to follow, do not cramp questions together, space it out, make a cover sheet, put personal information, give instructions, professional appearance Question formation  make responses ambiguous, list responses down a page, use matrix questions Non-response  you need cooperation or it weakens the survey Mail and Self-Administered questionnaires Advantages: cheapest, conducted by one researcher, done at respondent’s convenience, can give questionnaires directly to respondent, anonymity, and response rate normally high Disadvantages: people don’t always complete/return them, raise response rate by sending reminders but this adds time and cost, cannot control conditions, incomplete, cannot observe reactions, illiterate people Web surveys  Advantages: fast, inexpensive, flexible design and can use visuals and audios Disadvantages: 3 areas of concern (coverage, privacy, verification), unequal Internet access/use, self selection, provide clear instructions, avoid technical glitches Telephone interviews  Advantages: 95% of population can be reached; it is quick to do long distance, high response rate, more expensive but more flexible, control sequence, use of probes Disadvantages: higher cost, limited time, cannot reach those without telephones, inconvenient times, reduces anonymity, hard to use open-ended questions, cannot use visual aids, tone of voice Face to face interview  Advantage: higher response rate, can use longer questions, observe surroundings, and ask complex questions Disadvantages: high cost, training, supervision, interviewer bias, appearance, tone of voice, wording Role of interviewer  make everything clear (only half understand questions), obtain cooperation and build report but remain neutral and objective, nonjudgmental, monitor pace and answers Stages of an Interview   Introductory and entry – show authorization and explain  Asking and recording  Exit – thanks respondent and leave Probe = request to clarify unclear answer, to complete an incomplete answer or obtain relevant response Interviewer bias  social setting can affect answer and expectations can create bias Ethical survey  invasion of privacy, voluntary participation, exploitation, pseudo survey = use survey format to persuade someone to dos something Lecture: Survey Research Survey research:  Mode of observation  Select sample and give them questions  Used for descriptive, explanatory and exploratory purposes  Best way to collect original data for describing a population too large to directly observe Questionnaire construction  general format, format for respondents, contingency questions, matrix questions, put the easiest questions first, instructions, pretest the questions 3 main methods of administering survey questionnaires  Self-administered: computer generated or paper, cheap and quick, respondents can answer more freely, more effective for sensitive issues Face to face interviews  produce fewer questions, ensure respondents don’t skip questions, more effective for complicated issues Telephone surveys  time, money, impact of interviewer on responses is less Structured questions  Single response items: one response (ex. what year were you born) Categorical response: gives categories to choose from  response items must be exhaustive and mutually exclusive  make sure there is no category overlap Secondary analysis  analyzing someone else’s survey. Good because it is cheaper and faster than doing original but bad because it may not be valid Techniques for increasing response rate  follow up, inducement (prize or money), sponsorship, introductory letter, method of return, make sure it is party of their day, format, and respondent selection Structured interview  questions in the exact same order in the same way, no probing questions Focused interview  takes place with respondents known to have been involved in experience, refers to situations that have been analyzed before interview, interview guided specific topics relate to hypothesis, focus on experience on specific topics Non-directed interview  most skill from interviewers/researchers, most flexible form, do not have schedule or questionnaire, may have probes, subject talks and researcher probes them, more of a discussion Chapter 7 Experimental Research Experiment = when you modify something in a situation then compare an outcome to what existed without modification 3 things to do in experiments  begins with hypothesis, modify something in a situation, compare outcomes with and without modification Experimental research  strongest for testing causal relationships because the 3 conditions for causality are best met Research questions  allows research to focus on causal relations, best for issues that have narrow scope or scale, limits one’s ability to generalize to larger setting, isolate and target impact Random assignment  facilitates comparison in experiments by creating similar groups Why randomly assign  Random assignment = assigning cases to groups for the purpose of making comparisons. It is a way to increase one’s confidence that the groups do not differ in a systematic way. It is a mechanical method, the assignment is automatic.  Each case has an equal chance of ending up in one or the other group How to randomly assign  begin with collection of cases, divide it into 2 or more groups by a random process Matching vs. Random Assignment  matching is an alternative to random assignment but not frequently used. Problem = cases differ in many ways and researcher cannot know which is relevant Parts of the experiment  Part 1: treatment or independent variable = what the researcher modifies Part 2: dependent variable = outcomes in experimental research are the physical conditions, social behaviours, attitudes, feelings or beliefs of subjects that change in response to treatment Part 3: pretest = measuring of the dependent variable prior to introduction of treatment Part 4: posttest = measurement of dependent variable after the treatment has been introduced Part 5: experimental group = group that received the treatment or in which treatment was present Part 6: control group = the group that does not receive the treatment Part 7: random assignment Steps in conducting an experiment  Step 1: decide on topic, narrow it into testable problem and develop hypothesis Step 2: plan a specific design Step 3: decide on groups, treatment conditions, and number of times to measure dependent variable Step 4: locate subjects and randomly assign them to groups Step 5: measure dependent variable in pretest and then expose one group to the treatment and then measure the dependent variable again in a posttest Step 6: interview subjects before they leave and record measures of dependent variable and examine results Control in experiment  eliminate alternative explanations to establish causality definitively. Researcher uses deception to control setting Pre-experimental designs  used when it’s difficult to use the classical design. They have weaknesses that make it hard to determine causal relationship because of their poor internal validity Advantages: experimental and control subjects are identical in their genetic characteristics, income level, attitudes, beliefs – at the beginning of the experiment Disadvantage: possibility of multiple treatment interference One Shot Case Study Design  one group, treatment and a posttest. No random assignment One group pretest posttest design no control group and random assignment Static group comparison  two groups, posttest, treatment Quasi-experimental and special designs  make identifying causal relationships more certain, they are variations of classical experimental design. Researchers have less control over independent variable Two group posttest only design  same as static group (two groups, posttest, treatment) BUT the groups are randomly assignment Interrupted time series  one group, multiple pretest measures before and after treatment Equivalent time series  pretest, then a treatment and a posttest, then a treatment and a posttest, then a treatment and posttest etc. Solomon four-group design addresses the issue of pretest effects. It combines classical experimental design with two group posttest only design and randomly assignments subjects to one of four groups Design notation  system for symbolzing parts of experimental design  O = observation of dependent variable  O1 = pretest & O2 – posttest  X = treatment, independent variable  R = random assignment Internal validity  ability to eliminate alterative explanations of dependent variable Logical of internal validity  rule out variables other than the treatment by controlling experimental conditions though experimental designs Selection bias  threat that participants will not form equal groups because subjects can have characteristics that affect the dependent variable History  an event is unrelated to treatment that will occur during experiment will influence the dependent variable Maturation  threat that some biological, psychological or emotional process within subjects and separate from treatment will change over time Testing  threatens internal validity because more than the treatment alone affects dependent variable. Solomon four group design helps detect testing effects Instrumentation  when instrument or dependent variable measures changes during experiment Mortality  when a subject does not continue throughout experiment Statistical regression  when extreme values or for random errors to move group results toward the average. It happens when subjects are unusual with regard to dependent variable or if there is a problem with the measurement of instrument Diffusion of treatment of contamination  threat that participants in different groups will talk to each other and learn about the other’s treatment. Must isolate groups. Experimenter expectancy  when researcher’s expectations influence the participant’s behaviour to act a certain way. It threatens internal validity and double blind experiments help design this External validity  ability to generalize experimental findings to events and settings outside the experiment Reactivity  participants might react differently in an experiment than in real life because they know they are in study Hawthorne effect modify aspects of working conditions and measured productivity Lecture: External Validity External Validity  to whom and under what circumstances can results be generalized? Internal validity  is the experiment responsible for observed changes?  Threats to internal validity are potential explanations for observed changes  If an experiment has excellent internal validity, it can still have threats to external validity Chapter 8: Analysis of Quantitative Data Coding  systematically reorganizing raw numerical data into a format tht is easy to analyze using computers Codebook  documen
More Less

Related notes for SOAN 2120

Log In


Join OneClass

Access over 10 million pages of study
documents for 1.3 million courses.

Sign up

Join to view


By registering, I agree to the Terms and Privacy Policies
Already have an account?
Just a few more details

So we can recommend you notes for your school.

Reset Password

Please enter below the email address you registered with and we will send you a link to reset your password.

Add your courses

Get notes from the top students in your class.