Notes for class for finals
Chap 11 - case studies
Chap 12 - Observations
Reasons for observational studies:
To describe nature
Generate and tyest hypothysis
Assess behavior
Other methods are impratical
Obersever methods:
Narritives - three boys
Ratings -realibility is often low.
checklists - sheets of paper listing behaviors - check frequency/duration (should have observers before hand to see what actions/behaviors are actually occuring.)
decide number of observations
Decide on recording procedures
-Big vs little (molecular vs muscle twitches)
Sampling of Observation sessions
-Timing of Observatrions - when do behaviors start/end/reoccur/watch for set time/etc.
-Setting - Where person is observed makes them act differently. - Try to observe where the client/observed is having problems to help figure out the problem.
Natural - what normally in the environmet
Contrived - Jane Goodall's hiding Bananaana's.
Obtrusive and Unobtrustive definitions may give different results. - woman hand washing.
Practice ovserving - observe the fish, observe the similarities and differences between these fish!
Just because observes agree doesn't mean it's realiable. - when unknown compairions they did what they saw, when known comparison would agree more with the other.
Control Drift - Have observers meet and compare notes.
-is possible video tape behaviors to be oserved. (score randomly at end)
-Randomly bring in newly trained observers and compare scores.
Problems w/observations!
-Confounding
-Computing Relative Frequencies - some beahaviors easier to observe then others.
-Feedback may lead to bias! - tapes labeled before and after (actually had equal measures of behavior) and told observers to watch differences! - given expectencies- observers would change their tune.
- Sloppy observing (teacher stuck finger in urine and licked it, asked students to do the same then told student he stuck his first finger in and licked second finger if they had been paying attention.
Chap 13 - Surveys
Writing the questions - specificity iscrucial!
-Dont measure attitudes
Measure things about the persons attitude (canidate! What is lied/disliked about the canidate?)
-Questions can mean different things to different people.
Assembling Questionaire!
-Start w/ nonthreatening, interest questions that can be answered w/ little effort
-Intro topics w/ general questions, then moev to specifics
-Keep all questions on single topic together.
-Use transitional phrases to intro new topics.
-Use several kinds of questions and vary their lengths (let yes be conservative answer to some questions and liberal answer to others)
-Use screening questions (Did you happen to read any mags in the past month?)
-Choosing individual subjects - can't generalize from this class to entire population.
Simple random sampling -
Students! - What do I want? Full time? Part time? Both? Undergrad? Grad? Both? Only from Hayward Campus?
Get list of students, assign each student a random number and then drop all numbers in urn and randomly pull out until aquired number needed!
-Telephone directories don't have all numbers - And some poeple have multiple listings!
-Strtified Sampling! - Crate with all sorts of fruit - pick ut random sample of each type of fruit, weight them and this will reduce standard error.
Non-response - (response rate = useable data) - can have not given useable data, may have not responded at all.
-Less likely to give telephone interview - african american, elderly, westerners city-dwellers.
-If return rate = small = NO big deal!
-If not.... then problem!
-rewards, intro letters, follow-up calls = all help w/ response rates.
-People who respond after follow-up calls = very different then origanl respondents!
Administering the questions!
-Methods! - mailing, phone calls, personal interviews, several computer-assited forms.
-Mail = covers wide area
-W/ mail no interviewer bias/effects.-
-Response rates lower!
-poorly suited for older, poor edu, others w/ trouble reading/answering.
Telephone = Most homes have phones, easy, wide coverage area, BUT OBTRUSIVE!
-Interviews = longer more probing questions, visual aides, more interesting answers.
Interviers commonly have no serious stake in research so they don't care if things get messed up!
Interviewer effects!
Computer-assisted self-interviewing (CASI) - private, has audio, can be done in free time, can type up long/short answers/pick answers.
-Easier to lie!
-Self-admin of a test leads to higher reports of drug use, abortion, degree of fault on autmotive accidents, suicidal thoughts/behavior, risk behaviors for HIV, & medical symptoms!
-Under reporting will always happen with socially stigmatized activities/thoughts!
Subtle differences can make a huge difference!
-Survay on alcohol and survay on diet!
-Men and woman reported higher booze intake on diet survay then on Alcohol survay alone!
Research on Survay Validity!
-Survey truth is in direct relation to social desirability to the actions and questioned in the survay!
Uniquie Qualities of Surveys -
-person asking questions rarely person who wrote questions
-respondedant has no say in direction of talk.survey schedule controls topics spoken about.
-respondeent that doesn't understand questions can't do more then ask the question be repeated.
Pressure to know!
-respondents believe questions to be about things they should now about!
-Will report knowing about knowing stuff that isn't even true (since they *should* know it (fictional political leaders, etc)*
-If given the 'don't know' option respondents assume they should use it.
-0in not given this option then they'll try to fake an answer!
Perspectives - carr accidents study. - changed words from hit to smashed to crashed.
-Changing the word made the respondents give higher speed estimats!
Week vs year questions...
-Week questions = easily remembered information
-Year questions = important or otherly more memorable events!
-Contexts establish perspectives! - ask about nixon & watergate, then ask about the trsut worthiness of american politicians. =will score lower when reminded of nixon!
Vauge terms - will automaticallt persume what the term means to them. "Few"
-my view vs. Serenity"
Desire to appear consistant - Boys and barbies vs girls and footballs.
Desire to be helpful - will omit brushing teeth is asked what you did, but will check it off if the activity is part of a checklist.
Retrospective Recall and Consistancy = COMPLETELY UNRELIABLE!!!!
Why survay houses disagree - when using same exact questions!
Sampling error - normally using cluster sampling for large surveys b/c it costs less.
Variauability in sample composition - different organizations use different methods (phone, mail, internet, etc)
-Dem's work weekdays so week-day only polls mostly repub.
Interviewer effects! - people will respond to the interviewer differently depending on the interviewers race/skin color/looks/ideals/attitude!
slight changes in wording, sequence and context of questions can change answers.
Chap 14 - Other Types of Self Report
-Self report may be true, but we'll never know because we can't evaluate it.
-If the dadta can be checked then the report can be made 'valid'.
Hypothysis generated can be falsified!
-Suicide rate is high among people who cl,aim to be deeply depressed.
Self report may provide useful insights into mental phenomenon:
-Herion addicts
-Self reports may be only possible DV's.
Phenomenon of deception
5 parts
self decetion & collusion
Ramification (deception extedned to more then one person)
percasiceness
feeelings
Motivations (explain why people remained in relations in which they were being deceived)
Self-reports about inner processes (self-esteem, love, anxiety, hostility) can have stronger inferences.
-reports can help the experimentor understand if the IV indeuced the intended state or not.
-focused interview method - have weird expericnce? Give us a discription!?
-listen in friendly manner w/o inbturuopting!
-Only talk to relive subjects anxiety or to reassure
CHAP 15 - Using Exsisting Data
-All great events have been distorted (only the winners write about wars)
-Experiments and Surveys - Can be re-evaluated to acess credibility. (ask for orginal findings and most researchers coulnd't/refused to give them.)
Meta-Anaylsis - combining all sorts of different research.
Organize a reviewing strategy - decide question and furmulate
shit... I don't have time to take notes over all of this...