Notes for Midterm 1
Notes for Midterm 1 3163
Popular in Industry Research
Popular in Communication Studies
This 44 page Class Notes was uploaded by Emma Dahlin on Tuesday March 8, 2016. The Class Notes belongs to 3163 at Ohio State University taught by Zheng Wang in Fall 2015. Since its upload, it has received 31 views. For similar materials see Industry Research in Communication Studies at Ohio State University.
Reviews for Notes for Midterm 1
Report this Material
What is Karma?
Karma is the currency of StudySoup.
Date Created: 03/08/16
01/14/2016 ▯ Comm 3163: Notes ▯ Craniometry 1800s: o Studies comparing size of skull with amount of intelligence Claim: Non-whites less intelligent than whites Many Claims o The earth is flat o Smoking causes cancer o Use of Facebook causes loneliness o Life existed on Mars billions of years ago o How do we decide what’s true? Communication as a social science means that… claims must be supported with evidence! One of the earliest questions in mass communication research: how often do people use television? Methods used: o Taking a survey o Sonar: ultrasonic audience measurement system and method o Modern day version? Thermal infrared (illegal) Two Main Running Themes Throughout the Course: o Different methods can be used to answer the same research question o All methods have weaknesses and strengths ▯ Intellectual Organization of the course: 1 half: learn how to conduct research, learn a variety of methods 2nd half: apply this knowledge to research questions in communication industry People lie! Racists who don’t want to appear racist Consumers who don’t want to admit using a specific product (sex- related products) What methods can we use to find out the truth? o We will go beyond the “lie detector test” o There’s a claim that people are not aware of the reasons for why they like/dislike a person, object, product, etc. o These reasons are “unconscious” o What methods can we use to find out these unconscious reasons? o How do we show cause and effect? o Does poverty cause crime? o Does smoking cigarettes cause cancer? o Claim that we can never really show that sex/gender or race can be causes: Ex: Being Black causes someone to be more likely to vote for Obama ▯ ▯ ▯ Theory and Hypothesis ▯ Does exposing smokers (who don’t want to quit) to anti-smoking ads increase their likelihood of quitting? o Finding: smokers exposing to anti-smoking ads were less likely to quit after exposure to an anti-smoking ad Why? Theory: a series of logically interrelated premises/assumptions that are meant to explain a phenomenon in the world (i.e answers the question of why) Example of a Theory o Cognitive Dissonance (Festinger 1957) o People are motivated to maintain their pre-existing attitudes and beliefs o Information that challenges pre-existing attitudes causes mental discomfort o To minimize discomfort, they engage in certain behaviors o Come up with favorable reasons to maintain existing attitudes o Denigrate or discount the information that challenges pre- existing attitudes ▯ Finding: Smokers (who don’t want to quit) exposed to anti-smoking ads were less likely to quit after exposure to an anti-smoking ad o Why? o According to Theory of Cognitive Dissonance These smokers want to maintain their pre-existing attitude of not wanting to quit Anti-smoking ad challenges this pre-existing attitude and causes mental discomfort To minimize this discomfort, they denigrate/discount the info in the ad and come up with reasons to maintain their existing attitudes ▯ Hypotheses: concise statements about what you expect to happen in the world (given the premises/assumptions that constitute the theory) o Hypotheses must logically follow from the set of premises and assumptions of the theory ▯ Example: Let’s derive a hypothesis using cognitive dissonance o Different domain: Politics o Hypothesis: If you show voters negative information about a candidate they like…they will like the candidate ▯ DON’T confuse a Theory and Hypothesis! Hypothesis: Texting while driving increases your likelihood of a car accident o If I ask you why… Your answer is essentially a theory o You only have a limited set of attention that you can devote to tasks o Texting takes away attention from driving o You need attention in order to drive well ▯ ▯ ▯ Concepts and Measures ▯ Concepts: abstract ideas meant to represent some facet of reality o Violence o Attitude o Intelligence o Attention o Love o Concepts are the building blocks of scientific language o Watching violent TV causes aggression o Education increases intelligence o Use of social media increases loneliness o (These hypotheses are constructed with concepts) Kim Peek o Diagnosed as mentally disabled as a child o Cannot button his own shirt o Cannot live on his own o Scored below average on IQ tests o Memorized at least 12,000 books o Can read two pages at once-one page for each eye o Can add phone numbers in his head until value reached trillions o Amassed massive amounts of information on 15 subject areas o Is he highly intelligent? o Researchers can disagree on the definition of a concept Is intelligence….? o Taking in and retaining information o Mathematical aptitude o Street smarts o Adapting to you environment o Knowledge about the world etc… o In research, you need to explicitly define your concepts o Important because… o Communication between researchers o Measurement ▯ Measures: the process of assigning quantitative values in a systematic way Suppose we define intelligence as “mathematical ability” o Possible measures: Arithmetic test (pencil and paper) Arithmetic test (cannot use pencil and paper) We want our measures to be VALID Validity: the extent to which a measurement procedure measures what it intends to measure o Are you measuring what you are supposed to measure? o If you define intelligence as verbal ability and measure it using a math test- not a very valid measure o Example: Robert Yerkes o Developed intelligence tests for military recruits during WWI Administered to 1.75 million recruits Army Alpha Test (for literate English speakers) Army Beta Test (for illiterate people or non-English speakers) Defined intelligence as “native intellectual ability”: Mental faculties not affected by culture and educational opportunities 47% of Army recruits judged to have the mental age of a 13 year old: “A nation of morons” o Mental faculties not affected by culture and educational opportunities Reliability: the extent to which a measuring instrument consistently measures whatever it is that it is measuring o It is possible for a measure to be reliable BUT not valid o Ex: Body Mass Index Scale (BMI) o Supposed to measure obesity by looking at your height and weight o BUT doesn’t consider whether weight is fat or muscle ▯ ▯ Big Picture Stuf ▯ Know the difference b/t theory and hypothesis ▯ Important to explicitly define concepts ▯ We want our measures to be valid and reliable ▯ ▯ Causation ▯ X causes Y ▯ “Counterfactual conception of causation” ▯ it’s the dominant view of causation in the social sciences ▯ ▯ First, some terminology Unit: physical entity-person Treatment: an intervention whose effect we wish to assess Outcome: potential effect of treatment ▯ ▯ Example ▯ Unit=Lisa ▯ Treatment=smoking (one year), not-smoking ▯ Outcome=cancer, no cancer ▯ ▯ We want to assess if smoking causes cancer for a single unit (Lisa) ▯ ▯ Reality A: Lisa smokes (one year) ▯ Reality B: Lisa does not smoke ▯ ▯ If there’s a diference in Outcomes b/t Realities A& B, Then we can say that our treatment had a causal effect ▯ ▯ We only observe one of the realities! The other is counterfactual ▯ ▯ You cannot simultaneously observe a person in his or her treated and untreated states. (“The Fundamental Problem of Causal Inference”) ▯ ▯ Can’t determine the causal effect of X on Y for a single person ▯ ▯ BUT, we can figure out on average the causal effect of X on Y by comparing large groups of people ▯ ▯ Confound: factors other than your main treatment that may affect your outcomes ▯ ▯ Random Assignment: People you are studying have an equal chance of being assigned to your study’s conditions (in this case, treatment or no- treatment groups) ▯ ▯ What random assignment does: It distributes the confounds equally across your conditions If we find that people who smoked for a year had, on average more incidents of lung cancer, we can assay that it was b/c of smoking – AND NOT people’s diet! There is potentially a huge number of confounds that we don’t know about Random assignment distributes them equally across our conditions ▯ ▯ Hypothesis: Smoking Causes Cancer Independent variable (Smoking): the factor that’s going to influence another factor Dependent variable (Cancer): the factor that’s going to be influenced If you have a study in which the independent variable is randomly assigned, then you have an experiment Because experiments are designed to eliminate confounds via random assignment… o Most social scientists consider them to be “the gold standard for showing cause and effect” Understand the notion ▯ ▯ Lab Experiments ▯ “37 who saw murder didn’t call the police” – The New York Times o Hypothesis: when you are in a larger crowd, you are less likely to help others o Why? One theory: diffusion of responsibility: each person assumes someone else will take responsibility for it-so no one does How do we study this phenomenon? o One way: Laboratory Experiment Conducted in a very well-controlled environment (not necessarily a laboratory) Researcher has control over: Where it takes place What time Which participants Nature of treatment What point participant exposed to treatment Darley and Latane (1968) o Recruited college students as research participants o Told that they would talk to other students about personal problems o Random assignment: either talking to 1 student in another room (85% went out to get help) talking to 4 students – each in their own rooms (31% went out to get help) o Other “students” were confederates o Confederates=works for the researcher; pretends to be a participants o One of the participants pretends to have a seizure o Being part of a larger crowd decreases someone’s likelihood of helping someone Laboratory experiment o One of the main strengths: conducted in a very well- controlled environment o Used same confederates across the conditions—they all said the same things to each participant o Used the same “seizure script” for all participants Other examples of creative lab experiments o People do things that they know “are not good for them” when they are part of a group (smoking, driving while drunk, etc) o Hypothesis: group pressure can cause individuals to make “bad” judgments o Why? One theory: people are social and they want to be seen as part of a group (they want the group’s approval) o Ascher Line experiment When part of group 32% of participants gave wrong answer When by themselves, only 1% gave wrong answer Succumbing to group pressure Lab experiment on dangerous conformity: smoke-filled room study o Again, great control of the environment o Control over when and where smoke comes out (same location and time for each participant) ▯ External Validity ▯ Supposed we were interested in understanding this real world phenomenon: People can be peer pressured into doing things that they know are “not good for them” (smoking, driving while drunk, illegal drugs, etc) External Validity: the extent to which the research findings can be applied to settings, people, treatments, etc. in the real world Does the study actually help us to understand real world phenomenon? Smoke filled-room study is most applicable to real world settings o Staying in a burning building is dangerous. Smoking, driving while drunk, etc. are also dangerous behaviors. Shows the power of group pressure in making people do dangerous things. o Neither study applicable. People don’t make judgments about line drawings or staying in smoke-filled rooms in their every day lives. o Both applicable. Show that people are generally group conformists. They will conform in both dangerous and non- dangerous situations. ▯ ▯ Internal Validity ▯ The degree to which a study has effectively demonstrated cause and effect (X causes Y) Are the changes to the independent variable responsible for the changes to the dependent variable? Experimental studies generally considered to be very high in internal validity (b/c of random assignment) “Threats” to Internal Validity o are the changes in the IV responsible for the changes in the DV? o Threats are factors other than the IV that can influence the DV Demand characteristics: potential problem for lab experiments o Participants form an interpretation of the study’s purpose and change their behavior to fit that interpretation o Potential solutions: Don’t tell participants the study’s hypothesis Deception (tell them a fake purpose of the study) Post-study questionnaires (ask them to guess the study’s purpose) Non-Random Assignment o Potential problem for all types of experiments o Arbitrary assignment DOES NOT MEAN random assignment Ex: if participant’s last name starts with A to L, assigned to condition 1 If last name M to Z assign to condition 2 Not random: distribution can vary by ethnic group (two conditions may have different ethnic composition) o Random means you have equal chance of being assigned to each condition Take-aways o Know the characteristics of a lab experiments ▯ African Americans twice as likely to be unemployed compared to Whites Earn 25% less when employed compared to Whites When employers are evaluating similar African-American and White applicants, do they favor the White one? Lab Discrimination Experiment-recruited White participants o Created 2 identical resumes o “James Sullivan” and “Jamal Jenkins” o Randomly assign the resumes o “James Sullivan” was rated more favorably than “Jamal Jenkins” even though the resumes were completely identical in terms of qualifications Field Experiment: experiment that occurs in a natural environment for the behavior being studied o Researcher does not have the same level of control over the study (compared to a lab experiment) o Independent Variable: “White-sounding” name resume and “Black-sounding” name resume o Dependent Measure: call backs for an interview o White sounding names received 50 percent more callbacks for interviews o Employers who list “Equal Opportunity Employer: in their ad discriminate just as much as other employers Advantages of field experiments over lab experiments o People often not aware that they are in a study (minimizes demand characteristics) o Studying people in their “natural” environments Hypothesis: Being perceived as African-American increases likelihood of discrimination o Which study is better at showing cause and effect? Lab study o Which study is more applicable to real world settings? Field study Do politicians racially discriminate against constituents? A field experiment on state legislators? Butler & Brookchman, 2011 Do violations of personal space influence biological functions? ▯ “Natural” Experiments: ▯ In lab and field experiments, researchers control random assignment Natural Experiment: treatments not assigned by experimenter (sometimes by “nature”) BUT assignment of treatments is random o Researcher usually has no control over any feature of the “study” (they try to find instances of “natural experiments”) Does noise inhibit learning? (Bronzaft & McCarthy, 1975) o Classroom far from tracks o Classroom close to tracks o School assigns teachers and students into classrooms in a random way Being female causes someone to be more likely to possess liberal political beliefs o IV has to be randomly assigned o This is impossible Observational Studies: studies in which the treatment (IV) is NOT randomly assigned Why use them? o Its not feasible (or in some cases, it’s impossible) to conduct an experimental study o It’s not ethical to conduct an experimental study o Example: Hypothesis: children who witness domestic abuse at home are likely to abuse their partners when they grow up Observational Studies: Different types of researcher control over the study o Researcher might not have control over any feature of the “study” (“find” observational studies in the real world) o Researcher might have control over some features of the study o In all cases, researcher does not have enough control to randomly assign the IV Many types of observational studies: We’ll only cover a small subset o Simple non-equivalent group design Hypothesis: Being female causes someone to be more likely to possess liberal political beliefs. Recruit Males and Females Measure then compare political beliefs Suppose we find that: Females more liberal than males Threats to internal validity Potentially huge number of differences b/t the groups This design makes no attempt to deal with confounds o Matching Designs For every treated unit, find another untreated unit with similar attributes Example: being female causes someone to be more likely to possess liberal political beliefs 41% of women identify as Democrats vs. 32% of men At least one potential confound: income Males earn more than females Higher incomes associated with increased conservatism Differences in income, not sex, could be affecting their political beliefs You can match on more than one attribute (confound) Ex: age Older people tend to be more conservative Weaknesses of matching You can only match on attributes that: (1) You are aware can be potential confounds! (2) AND you actually have a way of measuring the potential confounds “Threat” to Internal Validity Potentially a huge number of confounds that you don’t know about o Pre-Post Test Without Comparison Group Measure dependent variable before treatment Measure dependent variable after treatment But, lack a comparison group that did not receive the treatment Example Hypothesis: Legalizing abortion can cause a decrease in crime Roe vs. Wade = treatment (court case that made abortion legal in US) Pre measurecrime rate before Roe v. Wade Post measurecrime rate after Roe v. Wade Suppose we find a decrease in crime rate from before and after the treatment Is it problematic to conclude that legalization of abortion causes a decrease in crime? YES, b/c there are lots of potential confounds (potentially huge number of other changes in the environment) o Reversal without Comparison Group Design Measure of dependent variable before the treatment Treatment Measure of dependent variable after the treatment Remove treatment Measure of dependent variable after treatment removal But, lack of comparison group Example: measuring % of fatalities when helmet law is put in and when it is appealed More compelling than pre-post test w/o comparison group design (more pieces of information) Does not completely rule out other changes to the environment o Pre-Post Test With Comparison Group (Diference-in- diferences approach) Measure of dependent variable before treatment Treatment Measure of dependent variable after treatment COMPARISON GROUP gets measured before and after on dependent variable but with no treatment in between Hypothesis: removing unhealthy foods from campus causes an increase in school performance Measure GPA, remove all junk food, measure GPA again (calculate pre and post test difference) Increase of .3: BUT, could be effect of national trend of eating healthy, not necessarily effect of the policy If there is also an increase in the control group that didn’t receive treatment than you may conclude that it is effect of national trend of eating healthy Being female causes someone to be more likely to possess liberal political beliefs o Some argue that sex is an “immutable characteristic” o You can’t change it – therefore prevents you from randomly assigning it o Some who think that experiments are the gold standard for showing cause and effect o Believe that you can never really show that sex can be a cause ▯ ▯ Surveys ▯ 2 types of information extracted from respondents o Factual knowledge Personal demographics, personal knowledge, “what is your sex?”, “how often do you use Facebook?” o Attitude/opinion about an issue, person, object, etc “do you favor or oppose allowing the US gov.t to torture terrorists?” 2 types of questions o Close-ended questions: questions with predefined responses “what is the most important issue facing this country? Economy, terrorism, energy policy, healthcare o Open-ended questions: respondents can answer in their own words “what is the most important issue facing this country?” __________ Basics of Survey Design o Question Wording Avoid double negatives “I don’t not know anything about that” “Do you favor or oppose not allowing gays and lesbians to legally marry?” “Does it seem possible or does it seem impossible to you that the Nazis extermination of the Jews never happened? 22% said it was possible that Holocaust never happened Avoid double-barreled questions Asking about two issues in one question and requiring one response “How satisfied are you with your pay and job conditions? Very satisfied, satisfied, neutral, unsatisfied, very unsatisfied “How much confidence do you have in Obama’s ability to handle domestic and foreign policy” Avoid leading questions Questions that are phrased in a way that makes the respondent feel that the researcher expects a particular answer “Like most Americans, do you prefer to purchase products manufactured in the United States?” Avoid words that may induce bias “Should the gov.t force citizens into a universal heal care system?” “Should the gov.t carry out a universal health care system?” “Are we spending too much, too little, or about the right amount on…” Avoid using Absolutes “always”, “never” Do you always go to the gym? Do you always eat breakfast? Should say: How often do you eat breakfast? Use Simple Words Avoid using complicated words Exhaustedtired Candidhonest Employmentwork Responsesanswers Keep Questions Short 1 to 2 short and clear sentences o Question Response Dichotomous Scale Two-point scale which presents options that opposite each other. Yes –No , Disagree-Agree Rating Scale Scale with more than two options Two heavily used in social science: Likert Scale & Semantic Differential Scale Likert Scales o Respondents rate their agreement with an item o Odd number of options with the middle option as “neutral” or “no opinion” o Evidence that acquiescence bias may occur in Likert Scales (especially among individuals low in socioeconomic status) o Acquiescence bias=tendency to want to agree with the statement Semantic Differential o Presented with two bipolar adjectives separated by (typically) odd number of spaces o Rating on a continuum (traits) o Not as susceptible to acquiescence bias Question Order Effects People’s responses to questions change depending on the order in which they answer the questions How to deal with it: randomize order of questions ▯ ▯ Social Desirability ▯ Social Desirability Bias: Tendency for people to answer questions in a manner that will be viewed favorably o People may give the “socially desirable” answer instead of the “real” answer o Bias most prevalent for “sensitive” topics o Ex: beliefs that may be viewed as racist, sexist, violent or illegal behaviors How do we know that it exists? o Compare what respondents say to “objective sources” o People will say they voted when they haven’t o People will report inflated incomes (when they are low) o People will report deflated incomes (when they are high) o People say that they don’t have a criminal record The problem of social desirability bias o Can compromise the validity of your survey measure o In other words, you will get distorted and inaccurate responses ▯ Dealing with social desirability bias Within the context of a “regular” survey o Have you ever contracted an STD? – put the question near the end of the survey (people are scared if you put a threatening idea at the beginning and might not be forthcoming with other questions) o Change mode of survey: answering surveys in a lab or online (at home) More likely to tell the truth when doing online (no researcher present) o If the question involves asking about potentially racist or sexist beliefs, race and sex of researcher matters Ex: Do you thinks that Whites work harder than Blacks or Latinos? (Don’t use researcher that is Black or Latino) o If possible, write the question in a way that deflects attention away from respondent Ex: instead of What do YOU think about a female president? … what does your NEIGHBOR think about a female president? Using “specialized” surveys designed specifically to minimize/eliminate social desirability biases o “List Experiment”: (Kuklinski, 1997) 1970s and 1980s-“New South” thesis-South no longer a bastion of racial prejudice Recruit participants from the South-people randomly assigned to either control condition or treatment condition Control Condition: Now I am going to read you a list of things that sometimes make people angry or upset. After I read all three, just tell me HOW MANY of them upset you. I don’t want to know which ones, just hOW MANY. The federal gov.t increasing the tax on gasoline Professional athletes getting million-dollar contracts Large corporations polluting the environment Treatment Condition: Same Instructions The federal gov.t increasing the tax on gasoline Professional athletes getting million-dollar contracts Large corporations polluting the environment A black family moving in next door (critical item) Took average in Control and average in Treatment If average in Treatment condition is higher than average in Control condition that means people are picking the critical item in the treatment condition Average in control: 1.93 Average in treatment: 2.37 2.37-1.93=.43 43% of people in treatment condition were estimated to be angered by critical item (black family living next door) Have also done this test using critical item as : A woman serving as president 26% angry or upset in treatment group about a woman serving as President o Randomized Response Technique Would you be upset if your child told you he/she was gay or lesbian? Yes or No Give subjects a Randomization Device (ex: coin) If heads, automatically answer “yes” to any question If tails, answer question truthfully Subjects are only ones who know outcome of coin flip This is supposed to facilitate them answering question truthfully 200 took survey: 160 said “yes”, 40 said “no” You can assume that… 50% automatically said “yes” 50% answered truthfully Number of subjects who had to give true opinion: 100 Number of “yes” responses when giving true opinion (total yes-automatic yes) 160-100 60/100 means 60% truthfully said “yes” o Bogus Pipeline Procedure Participants strapped to what they think is a functioning “lie detector” “Lie Detector” machine does not actually work Assumption is that if people think you can detect their lying they more likely to tell the truth People tend to admit more socially uncomfortable things using this method ▯ ▯ Non-self-report Measures Open/close-ended questionnaires List experiment Randomized response technique o These are self-report measures: measure that ask people about their attitudes, beliefs, knowledge o There are also non-self-report measures: measures that still try to determine people’s attitudes, beliefs, knowledge … but they DO NOT ask people Hurricane Katrina o Photos of people taking food o Associated Press Captions African American man: “a young man walks through chest deep flood water after looting a grocery store” White man and women: “two residents wade through chest-deep water after finding bread and soda from a local grocery store” Explicit and Implicit Attitudes o Explicit Attitudes/Beliefs: conscious beliefs and evaluations of people, objects, ideas, etc. These are beliefs that you can deliberate about and report Do you like or dislike Donald Trump? Do you enjoy taking naps? Do you think that people in the Midwest are nicer than people from the East Coast? o Implicit Attitudes/Beliefs: unconscious evaluations or associations you have about people, objects, ideas, etc. These are evaluations and associations that you can’t consciously report As such, you can’t use surveys/self-reports to measure them o Major ways that explicit attitudes and implicit associations have been measured in psychology: Stereotyping o Explicit Stereotyping: conscious and intentional endorsement of a stereotype “sneaky goddamn Jews are all alike” “money grubbing Jews” When people purposely dress up as stereotypes Measuring it: surveys BUT, remember social desirability biases o Implicit Stereotyping: you have an unconscious association b/t a group and a trait What’s inside your head? Concepts are grouped/associated together Links and associations that they have are not random at all Associations are based on what you’ve been exposed to in your environment Think of stereotypes the same way Repeated exposure in cultural environment Family Friends Media o Bottom Line: Whether you like it or not, if you are exposed to them, you will form these unconscious associations o Measuring Implicit Associations: Implicit Association Test (IAT) Uses response time=speed at which someone responds Context of IAT, speed at which one presses the button Important: Association DOES NOT IMPLY endorsement Explicit attitudes do NOT equal implicit attitudes 50% of African-Americans implicitly associate: European Americans and positive ideas African Americans and negative ideas Why? Implicit associations are reflections of what you are exposed to in your environment Why are implicit associations important? These can influence behavior o Gun or Tool Classification Experiment Shown either Black or White Face for 200 ms and then shown Gun or Tool Target for 200 ms Had to classify if it was a gun or tool Results: more likely to misclassify tool as a gun when African American vs. White Face Faster to classify gun as a gun when African American vs. White Face Stronger association b/t African Americans and guns than Europeans and guns o Shooter Bias Experiment Encounter people either holding gun or cell phone (African American of White Person) Shoot people with guns but don’t shoot if their holding cell phone Found that people are more likely to shoot when African American is holding cell phone vs. White Person Faster to shoot African American holding gun vs. White Person holding gun Non-self report measures continued o Measures that still try to determine people’s attitudes, beliefs, knowledge, etc o BUT they DO NOT rely on asking people Psychophysiological Measures o A form of non-self-report measure o Measure biological signals generated by different parts of the body o Eye tracking, skin conductance response, fMRI Eye tracking o Sensor technology that allows a device to determine what a person is looking at o Most eye tracking devices shine an infrared light at the eye o Eye movements composed of two events o Fixations – when the eye make quick stops o Saccades - rapid movements of the eyes in between fixations o Useful in measuring attention o What are people looking at? o How long are they looking at something? o Technology has been used in advertising o Determine which parts of the ad elicit the most attention Do adolescents pay attention to warning labels? (ex: Camel Cigarettes has Surgeon’s warning label on them) o Adolescents look at magazine tobacco ads while eyes are tracked o On average, spend 8% of total viewing time on warning Eye Tracking and Distracted Driving o Research: people using cell phones 4X more likely to cause accidents My Eyes are Up Here: The Nature of the Objectifying Gaze Toward Women o Males and Females evaluating photographs of women o Randomly assigned to assess either (1) Personality or (2) Attractiveness o Hypothesis: men would be more likely to focus on body when assessing both traits o Finding: Both men and women focused on bodies when making decisions on attractiveness o Men overall more likely to fixate first on body Skin Conductance o Used as a measure of emotional arousal (heightened excitation or agitation) o Two electrodes located on fingers o Low electric charge passed between the electrodes o As participant becomes emotionally aroused o Sweat glands produce sweat o Skin becomes a better conductor of electricity o Measures emotional arousal by determining how easy it is for electricity to pass through the skin o Larger “bump” indicates higher level of arousal o Caveat: emotional arousal could be due to: stress/anxiety, pleasure, disgust, joy, fear, etc. o Skin conductance can’t determine if arousal is due to positive or negative emotions Can playing violent video games desensitize people to violence? o Used skin conductance as measure of desensitization to violence o Randomly assign people to play non-violent video game or violent video game o Those who played violent video games were not as aroused when they were looking at violent footage aka not as disturbed by it Skin conductance and polygraph: o Most widely used format is the Control Question Test (CQT) o 3 types of questions: Irrelevant Qs=no real relation to what is being investigated (ex: “What’s your name?”, “How old are you?” “Do you live at ____”) Relevant Qs=pertain to what is being investigated (“Did you embezzle funds?”) Control Qs=questions about misdeeds that are similar to those being investigated, but refer to subject’s past and are usually broad in scope (“Have you ever stolen anything in your life?”) o Assumption: Innocent people are going to be more afraid of Control questions than Relevant Questions Level of arousal to relevant questions should be less than the control questions o Guilty person Level of arousal for relevant questions should be higher than the control questions fMRI: functional Magnetic Resonance Imaging o Very powerful magnet o Can measure amount of oxygenated blood going into brain regions o Subjects typically (at minimum) do 2 tasks o Brain region “x” is “active” o More oxygenated blood is in brain region “x” while subjects are doing Task 1 than Task 2 o Study question: do individuals generate a negative emotional (fear) response to African-Americans? o Tasks : Task 1-Target is African-American Task 2-Target is white Result: more oxygenated blood is in the right amygdala when participants are looking at an African-American target than a Caucasian-American target What’s so special about the amygdala? Amygdala associated with fear conditioning If you surgically remove amygdala in rats, they are no longer susceptible to fear conditioning Patient SM: 44 years old: Caucasian: Female: Amygdala destroyed when she was really young Evidence that she has not been able to experience fear for decades She cannot pick out a scared face Her drawings of emotional responses One interpretation of the fMRI data: participants have a fear response to African-Americans Problem: brain regions are usually associated with many different psychological processes Amygdala activity has also been associated with: Novelty detection Thinking optimistically Positive emotions We DON’T know which psychological process was actually engaged when people were looking at African- American photos ▯ ▯ Sampling Sampling: procedure used to select research participants Smokers exposed to anti-smoking ads will be more likely to quit smoking Some Terminology: o Population: the whole set of individuals you are interested in (defined by researcher) ex: all smokers in the US o But, you cannot study all smokers in the US o Obtain a sample: a subset of individuals from the population of interest When we run a study: o When we make conclusions about the population based on information we acquire from our sample (generalization) o Ideally, we want a representative sample: sample is similar to population in features relevant to research o Online reviews: example of non-representative sample Usually from people who either have very strong or very negative opinion of product o Two general ways to obtain samples: Probability sampling: more likely to obtain representative sample Non-probability sampling: more convenient to implement, Less likely to obtain a representative sample How you would do probability sampling: o You first need a sampling frame: list of individuals that make up your population of interest o Sampling frames are NOT going to be perfect (i.e the list won’t have every member of your population of interest) o Want to get a good approximation Example: population of interest: Ohio voters Good sampling frame: Ohio voter registration records Population of interest: homeowners in Columbus Good sampling frame: Numbers in phone book o Undercoverage: individuals in population NOT covered by sampling frame o Overcoverage: individuals in sampling frame that are NOT part of population of interest o BUT, there can be really bad sampling frames Ex: Population of interest: homeless Bad sampling frame:: voter registration lists (homeless people typically don’t vote) Some populations may not even have a known sample frame Population of interest: marijuana users in Ohio o (1) Simple Random Sampling: each individual in the sampling frame has an equal chance of being included in the sample Ex: sampling frame: Ohio voter registration record w/ 100 individuals Want to run a study with 25 ohio voters Each person in sampling frame has a 1% chance of being selected to be part of our study o (2) Stratified Random Sampling: way to make sure that sample represents population in some relevant dimensions Ex: 60% of voters in Ohio are female, 40% are male Want sample to have same distribution of males/females If voter registration list is sampling frame, create male and female lists Obtain 60% of sample by simple random sampling from list of females Obtain 40% of sample by simple random sampling from list of males o (3) Systematic Random Sample: Sampling frame is a list of individuals Pick a random individual as the starting point Every nth element of the list is selected Ex: Have a list of 100 peopth Randomly start at the 12 person of the list Decide on an interval of 4 Pick 16 , 20 , 24 , 28 person, etc. One possible problem: Periodicity: the order of items in the list may bias the selection process Ex: sampling broadcast TV shows from a list, with an interval of 10 th What if NBC aired every 10 show on the list? Sample would not actually be representative of all broadcast TV, but just of NBC Non-Probability Sampling o Much easier to implement than probability sampling o Less likely to obtain a representative sample o (1) Convenience Sample Research participants are selected if they can be accessed easily and conveniently Ex: Undergraduate study pools Ex: Mall intercept – researchers in a shopping mall stop people passing by to ask them if they would be willing to participate in a research study Huge in terms of research! o (2) Volunteer Sample Participants have volunteered to take part in the study (sometimes give payment) Ex: Call-in or online polls in News or Talk shows o (3) Purposive Sample Participants selected due to characteristics they possess Individuals that do not meet the necessary criteria are not used Ex: want a sample of business executives Interview executives from a company located close to the University o (4) Quota Sample Researcher selects people according to some fixed quota (non-probability version of stratified random sampling) Ex: Want gender distribution of sample to reflect population (51% female, 49% male) Select 51 women and 49 men (using mall intercept) o (5) Snowball Sample Participants are encouraged to tell people they know about the study Word of mouth is then used to recruit further participants/ “referrals” Sometimes an incentive is offered to those whose friends do participate WEIRD Samples o “The weirdest people in the world” – article published in 2010 o Main argument: behavioral social sciences make generalizations about people (i.e human race) based on samples drawn from WEIRD societies o WEIRD=Western, Educated, Industrial, Rich, Democratic o Small slice of human race o Claim: when it comes to fundamental aspects of motivation, psychology, and behavior, WEIRD subjects are unusual from rest of species o Example from the field of psychology: From 2003-2007, 93% of studies in the main 6 psychology journals used only Americans and Europeans as research participants 68% from US 99% of US participants were from undergraduate research pools (convenience sample) Evidence that American undergraduate students re: VERY different from everybody else NOT AT ALL representative of humans o Western vs. Non-Western societies More analytic reasoning than holistic reasoning More motivation to view self in positive light Less conformity More desire for choice o Americans vs. Other Western Societies More analytic reasoning More motivation to view self in positive light Less conformity More desire for choice o College Americans vs. Other Americans More analytic reasoning More motivation to view self in positive light Less conformity More desire for choice o Claim: if you want to understand human beings Relying largely on American undergraduate student samples is an incredibly bad decision ▯ ▯ Semi-Structured Interviews, Focus Groups, and Wiki Surveys ▯ ▯ Surveys Limitations o If closed-ended: researcher needs to write all questions and all answers o Might not know the most important things to ask o Might not know the most important responses o If open-ended: can’t ask follow-up questions ▯ Potential ways of obtaining more information: Interviews o Interview: Verbal exchange where one person (interviewer) elicits information from another person (the interviewee) through a series of questions o Defining characteristics of interviews Open-ended, avoids “yes” and “no” questions Flexible (can ask follow-up questions based on answers) Focus on participant’s thoughts/experiences Build relationship with interviewee o Semi-structured Interview: Have a set of pre-determined questions BUT, you can ask additional questions based on their responses (questions may differ based on respondents’ answer) Different from the structured interview Always the same set of pre-determined questions Types of Questions in Semi-structured Interview: Essential Questions Central focus of study Ex: What’s you’re view on immigration? Throw away Questions Used to develop rapport, demographics, cooling off Ex: Where do you work? Probing Questions Ex: Could you tell me more? What happened next? Sequence of a semi-structured interview: (1) Introduce yourself, read rights, explain aim of what you are doing (not explain everything) (2) Warm-up, simple questions, factual questions throw-away questions, gets them relaxed generate feeling of trust (3) Main issue- shopping list of essential questions and probing questions (bulk of interview) (4) Cool off questions, more throw away questions (5) Closure, thank them for participation (6) Recorder is turned off – off the record Suggestions: Listen more than you speak Use straightforward, non-threatening questions at the outset Build trust Avoid questions with yes or no answers Eliminate cues from interviewer – try to avoid giving away whether you agree with their opinions They need to feel like they are NOT being judged Allow respondents to think (don’t cut them off) Allow participants to provide additional information when they see fit Allow for the interviews to evolve organically as they see fit Can be conducted in diferent mediums: Telephone Interviewing over email – can take a very long time Video conferencing (Skype) can also do recording Advantages: Not limited by pre-determined answers Allow respondents to come up with issues that you may not have previously considered May be able to use this information to construct a close-ended survey Disadvantages: Takes a long time to collect data Tends to last 30 minutes-1.5 hours Transcription can be slow and time consuming (need to transcribe audio recording) On average and depending on skill level, it can take 5-6 hours to transcribe 1 hour interview Shares some weaknesses with regular surveys People can only share reasons they can consciously verbalize People might not tell you the truth Focus Groups o Group of people asked about beliefs, and attitudes towards a person, product, etc. o Group of people with certain characteristics generate data in a focused discussion o Members of group can speak to each other o Designing a focus group: Size: 5-10 participants per group Fewer than 5, not as much discussion More than 10, difficult for moderator to keep discussion flowing well All participants may not have enough time to speak May decide that some groups composed of certain individuals (ex: all men, all women) Group composition important Try to avoid power/status differentials (ex: one person has PhD, rest only finished high school) Decide how alike or different you want the group to be Ex: strangers vs. acquaintances, experts or novices o Data Collection: 8-12 questions, fewer is better 5-10 minutes per question (open-ended, clear) Drop a pebble, see a ripple o Moderator Skills: Strong interviewing skills Ability to control and guide discussion (in moment judgment) let it go but control it Ability to suppress personal views o Advantages: Interaction and group dynamics Can reveal how people can talk about an issue Wider range of responses Activate forgotten details (experiences) Group exchange can remove inhibitions o Disadvantages: Might give a distorted view of people’s attitudes Attitudes become more extreme over the course of discussion Groupthink-everyone agrees with the dominant speaker One of the reasons to avoid power/status differences among group members Transcribe group discussion (time consuming) Wiki Surveys o Tries to have: Some of the openness of a semi-structured interview Ease of data collection characteristic of close-ended surveys o A single question associated with many responses Respondents can add response options Respondents usually “vote” for options two at a time o Example: When Bloomberg wanted to find ideas how to make NY a cleaner city What do you think is a better idea for creating a greener, greater NYC? o Advantages: Some of openness of semi-structured interview (respondents can add responses) Data collection is relatively easy Potentially very useful for generating ideas (products, policies, etc.) Crowdsourcing idea generation and obtaining people’s opinions on what they like o Disadvantages: Maybe of limited use for many questions in social sciences ▯ ▯ So far, we’ve been studying and collecting data directly from people We’ve learned how to use surveys, response times, psychophysiological techniques, interviews, etc. BUT, we might also be interested in studying messages (nature and content) o (commercials, TV shows, political ads, speeches, tweets, etc) For example…Do sports news websites offer the same amount of coverage for male and female athletic teams? Is the nightly news more likely to portray non-Whites as criminals compared to Whites? Do video games portray females in a negative manner? All these questions are a