PSY290 EXAM 2 notes R. Stuetzle
PSY290 EXAM 2 notes R. Stuetzle PSY 290
Popular in Intro to Research Methods
Popular in Department
verified elite notetaker
One Day of Notes
verified elite notetaker
verified elite notetaker
verified elite notetaker
verified elite notetaker
One Day of Notes
verified elite notetaker
This 35 page Study Guide was uploaded by Eureka on Sunday January 17, 2016. The Study Guide belongs to PSY 290 at University of Miami taught by Rick Stuetzle in Spring2015. Since its upload, it has received 132 views.
Reviews for PSY290 EXAM 2 notes R. Stuetzle
Report this Material
What is Karma?
Karma is the currency of StudySoup.
You can buy or earn more Karma at anytime and redeem it for class notes, study guides, flashcards, and more!
Date Created: 01/17/16
Chapter 5 Experimental Research Experiment--Woodworth ◦ Investigator directly varies some factor(s), holds all else constant, and observes results of systematic variation ◦ Factor that is varied =independent variables ◦ Factor(s) held constant =extraneous variables ◦ Result that is observed =dependent variables ◦ Box 5.1--John Stuart Mill and the Rules of Inductive Logic (agreement/difference) Independent Variable ◦ Minimum of 2 levels/conditions to make comparison ◦ E.g. dose1 vs dose2 : a study requires at least two dosage levels in order to make a comparison. This study would be described as an experiment with “amount of marijuana” as the manipulated independent variable and “dosage 1” and “dosage 2” as the two levels of the independent variable. You could also say the study has two conditions: the two dosage levels. ◦ It is the factor of interest to the experimenter, the one being studied to see if it will influence behavior (the “watching violent TV” in the John Stuart Mill example). Varieties of Manipulated Independent Variables ◦ Situational variables--Variations in a feature of the environment participant encounters e.g. in a helping behavior study, IV is the number of potential helpers on the scene besides the participant, and the levels are zero, three, and six bystanders. Thus the experimenter has created three situations. e.g. Door in the face effect: L then S; S only; Simulan ◦ Task variables--Variation in type of task performed by participants e.g. levels of complexity of puzzle ◦ Instructional variable--Variation in instructions given to groups of participants E.g., “Go as fast as you can” vs. “Get as many right as you can” ◦ Combination of different types of IVs E.g. A study of the effects of crowding, task difficulty, and motivation on problem-solving ability Crowding = room size Task Difficulty = easy vs. difficult Motivation = win $1 vs. win $5 Multiple IVs allow for hypotheses re: interaction effects Control Groups as IVs ◦ Getting treatment or not getting treatment as independent variable In a study of the effects of TV violence on children’s aggressive behavior Group getting treatment = experimental group=children be shown a violent TV program Group without treatment = control group= others don’t get to see it, or see a nonviolent TV show ◦ Participants in control group are identical in all ways to those in experimental group except for experimental treatment--ideally ◦ Control groups are not necessary in all research. Control groups occur only in research when it is important to have a comparison with some baseline level of performance. E.g. the comparison between different gender ◦ Research example—experimental and control groups: experimental group(IV)= subjects were given reason to think they might be lucky control group(DV)= subjects were not given any reason to think that luck would occur. Result: support “self- efficacy,” did the kinds of things that might be expected to improve performance. Controlling Extraneous Variables ◦ Extraneous variable Any uncontrolled factors not of interest to researcher but which might influence the behavior being studied ◦ Confound = any uncontrolled extraneous variable that covaries with IV, and could provide alternate explanation of results The results could be due to the effects of either the confounding variable or the independent variable, or some combination of the two. Results from studies with confounds are uninterpretable. E.g. a verbal learning experiment: Two serious confounds: distribution of practice is confounded with total study hours/ distribution of practice is confounded both with total study hours and with retention interval Measuring Dependent Variables ◦ Dependent variable = measured outcomes of experiment o E.g., TV violence and aggression DV= some measure of aggressiveness ◦ Quality of study depends on choice and operational definition of dependent variable-- the behaviors are defined precisely, replication is impossible ◦ Dependent measures have been shown to be reliable and valid ◦ A brief pilot study can avoid these two problems (ceiling effect: depend measure is so easy that everyone gets high score, no difference/floor effect: too difficult) ◦ A particular construct could be an independent, an extraneous, or a dependent variable, depending on the research problem at hand. An experimenter might manipulate a particular construct as an IV, try to control it as an extraneous factor, or measure it as a DV. Manipulated versus Subject Variables ◦ Compare groups based on already existing characteristics of individuals (Gender, Age, Personality characteristic) ◦ Manipulated vs. subject variables How could you study the effects of anxiety on problem solving behavior with anxiety as a manipulated variable/ a subject variable? Manipulate anxiety directly by creating a situation in which one group is made anxious (told they’ll be performing in front of a large audience), while a second group is not (no audience). Using a subject variable, select two groups differing in their characteristic levels of anxiety and ask each to try the maze. ◦ Studies using independent variables that are subject variables are occasionally called ex post facto studies, natural groups studies, or quasi experiments (quasi meaning “to some degree” here). ◦ We can’t draw conclusions about causality when the IV is a subject variable rather than a manipulated variable-- When using subject variables, however, the experimenter can also vary a factor (i.e., select participants having certain characteristics) but cannot hold all else constant, all we can say is that the groups performed differently on the dependent measure. In a study-- how altruistic was affected by “self-esteem.” Manipulated variables: raised or lowered self-esteem temporarily— draw conclusion about causality Subject variables: personality test for self-esteem and select high and low score participants—cannot have causality DV: differences in volunteering ◦ Research example—using subject variables: using subject variables examines differences between cultures. Subject variables: gender and cult Result: Higher error means more field dependence People who exhibit field dependence tend to rely on information provided by the outer world, the field or frame of a situation and their cognition (toward other things) is based on this overall field. Box5.2 Classic studies—Bobo Dolls and Aggression ◦ IV: manipulated variable: the type of experience were the participants in the study, and participants in groups 1 and 2 were exposed to either a same- sex or opposite-sex model ◦ Controlling extraneous variables: the toys were always arranged in exactly the same way; participants in all four groups were mildly frustrated before being given a chance to aggress. ◦ Measuring DV: how closely they matched the model’s behavior Experimental Validity ◦ Chapter 4 mentioned the validity of measures-- A behavioral measure is said to be valid if it measures what it is designed to measure: content/criterion/construct ◦ Here we are referring to the validity of an entire experiment ◦ Statistical conclusion validity: concerns the extent to which the researcher uses statistics properly and draws the appropriate conclusions from the statistical analysis. o 3 ways to reduce validity: The wrong analysis-- need ordinal scale, but mistakenly use an interval data Fraudulent Reliability of the measure—Type II error ◦ Construct validity: referring to the adequacy of the operational definitions for both IVs and DV E.g. In a study of the effects of TV violence on children’s aggression questions about construct validity could be (a) whether the programs chosen by the experimenter are the best choices to contrast violent with nonviolent television programming, and (b) whether the operational definitions and measures of aggression used are the best that could be chosen. ◦ External validity: the degree to which research findings generalize beyond the specific context of the experiment being conducted To other populations: Example—convenience sampling: college students might not generalize to other populations (more cognitively/self-centered/ susceptible to social influence, and likely to change their attitudes on issues). To avoid, replicate finding on a variety of populations. Example—gender: Adequacy of women’s health initiatives based on studies will all male participants; Baltimore Longitudinal Study of Aging (all participants are male, but drawing conclusions as if they apply to everyone.) Example—generalizing results from one culture to another (individualist/collectivist) To other environments/ ecological validity: Applicable to other settings. Ecological validity—research with relevance for the everyday cognitive activities of people trying to adapt to their environment. Example: Memory Do memory processes identified in lab-based studies account for the way in which eyewitness memories are formed? No, whether principles discovered in the lab generalized to real-life memory situations was not clear Example: Drug treatment program Is drug treatment program that is effective in urban setting also effective in rural settings?]= To other times: Do the results generalize to other times? Age-Cohort Effects? Example: Development of Intelligence during adulthood Conclusion that intelligence begins to decline around age 30 In what ways (other than actual intelligence) do 10, 20, 30, 40, 50 year olds differ that might account for this finding? ◦ There are many examples of research, completed in the laboratory under so-called artificial conditions, which have great value for the understanding of human behavior. Example: research “ false memory” has relevance for eyewitness memory Box5.3 Ethics—Recruiting Participants: Everyone’s in the pool ◦ Internal validity: the degree to which an experiment is methodologically sound and confound-free. High internal validity means DV are directly associated with the IV and are not the result of confounding factors. ◦ Threats to Internal Validity: o Pre-Post Design: to test whether people will change over time as the result of some experience; one procedure is to evaluate people prior to the experience with a pretest. Then, after the experience, a posttest measure is taken. Experimental: pretest ➔ treatment ➔ posttest Control: pretest ➔ posttest In a program—give freshmen the anxiety treatment History: DV is due to the historical event of college’s change in grading policy rather than the program (anxiety caused by the low grade, drop in anxiety caused in policy) Maturation: DV is due to maturation of these students as they become accustomed to college life. Maturation, developmental changes that occur with the passage of time, is always a concern whenever a study extends over time. If the program is effective, the result is: Experimental: pretest Treatment posttest 90 70 Control: pretest posttest 90 90 Regression to the mean— extreme score at pre-testing, lower score would be predicted based on chance alone Students are selected for some treatment because they’ve made an extreme score on the pretest. Their anxiety scores might be lower than on the pretest, but the improvement could be a regression effect A control group of equivalent high-anxiety participants would enable the researcher to spot a possible regression effect. The following outcome would suggest regression might be involved: Testing: Just taking the pretest has an effect on posttest scores. Practice effects: be familiar, more confident Pretest sensitization: sensitize participants to something about the program.( Participants might be sensitized by the pretest to topics about which they seem to know nothing; they could then pay more attention to those topics during the course and do better on the posttest as a result. ) Instrumentation is a problem when the measurement instrument changes from pretest to posttest. Instrumentation is sometimes a problem when the measurement tool involves observations o Participant Problems Participant characteristics can affect the internal validity of your study Subject Selection Effects In an experiment need to be sure that participants in different conditions are equivalent to each other except for the IV E.g., effect of teaching method on performance o Students choose whether to take lecture or lecture/discussion format class o What can/can’t you conclude about effectiveness of teaching method at the end of the semester? o IV=teaching method, but there may has a confound due to the subject selection—interested in discussion, and perhaps better students Conclusion: performances of students improve, after taking the course including discussion and lecture, but not draw the causality. o Study#2: “ulcers(溃疡) in executive monkeys” Brady concluded that concluding the psycho- logical stress of being in command not just of one’s own fate but also of that of a subordinate could lead to health problems However, Subject Selection Effect took place in his study: Monkeys learning most quickly were placed in the executive condition Thus, he unwittingly placed highly emotional (and therefore ulcer-prone) animals in the executive condition and more laid-back animals in the control condition. Attrition: Some studies may last for a relatively long period, and people move away, lose interest, and even die. The problem is that if the particular the group of people drop out the study, the study is made up by the different group of people than beginning—subject selection effect One way to test difference between attriters and continuers is to look at the pretest scores or other attributes at the outset of the study for both groups； if they are indistinguishable, could strength the conclusion of the study, even though attrition happened Know who drop—compare with the rest of people Can differential attrition account for your findings? In sum, Internal validity refers to the methodological soundness of a study—it is free from confounds. External validity concerns whether or not the results of the study generalize beyond the specific features of the study; e.g. Criticizing the use of honors students is therefore not a problem with internal validity but a problem with external validity; the results might not generalize to other types of students. Chapter 6 Methodological Control in Experimental Research Between-subjects design— Different groups of participants contribute data for different levels of the independent variable Within-subjects design -– Same participants contribute data to all the levels of the independent variable Example: Effects of music volume on studying IV with levels A & B: Level A = quiet music/ Level B = loud music o Between-Subjects Design: Each participant is observed in either Level A OR Level B, but not both; That is, each group or level represents a different condition of the IV o Within-Subjects Design: Each participant is observed in both Level A AND Level B Between-Subjects design: In some conditions, it is the only way to design study: o IV is a subject variable: Introverts vs. Extroverts/ Males vs. Females Exception: the same person are studied at different ages/ marital status (he same people are studied before and after a marriage or a divorce.) o IV is a manipulated variable: sometimes when people participate in one level of an independent variable, the experience gained there will make it impossible for them to participate in other levels. This happens in most research involving deception E.g. the effects of the physical attractiveness of a defendant on recommended sentence length by Sigall and Ostrove (1975). 2 manipulated variables: the type of crime (burglary and swindle)/Barbara’s attractiveness Control group: didn’t see Barbara’s photo Result: when the crime was burglary, attractive Barbara got a lighter sentence on; while the crime was swindle, participants gave attractive Barbara a harsher sentence E.g., eyewitness memory for details preceding violent vs. non-violent events (Loftus & Burns, 1982) Viewing one version of a film makes it impossible to then have participants view the other version without bias/influence of prior experience o Pros Each participant comes in equally naïve about procedures o Cons Need a large number of individuals to fill all experimental conditions Differences between conditions may be due to independent variable (you hope so) BUT could also be due to pre-existing differences between the groups--To deal with this potential confound, create equivalent groups Creating Equivalent Groups: Groups that are the same as each other in all important ways except for the levels of the independent variable Ways to create equivalent groups: o Random Assignment A method of placing participants into different groups Every participant has an equal chance of being placed in each of the groups; The goal of random assignment is to spreads potentially confounding factors evenly throughout the different groups BUT...cannot ensure equal number of participants in each condition (think about if you flipped a coin to decide which group 20 participants would be in); The greater the number of subjects involved, the greater the chance that random assignment will work to create equivalent groups. Block Randomization: Used to ensure an equal number of participants per group; Each condition of the study has a participant randomly assigned to it before any condition is repeated a second time Random assignment is that the process is normally associated with laboratory research, random assignment is also possible in field research. E.g. Contact hypothesis—racial prejudice can be reduced when members of different races are in frequent contact with each other They found a university where roommates were randomly assigned and designed an experiment to compare two groups of white first-year students— those randomly assigned to another white student and those randomly assigned to an African American student. Over the course of a fall quarter, the researchers examined whether the close proximity of having a different-race roommate would reduce prejudice. In line with the contact hypothesis, it did. o Matching (see Table 6.1) Participants are grouped together on some trait and then randomly distributed to the different groups This is especially useful when there are too few participants available for random assignment to work well Two conditions needed for matching：You have a good reason to believe that the matching variable is correlated with the dependent variable； You need a good way to measure the matching variable One practical difficulty with matching concerns the number of matching variables to use. In matching, participants are grouped together on some subject variable such as their characteristic level of anxiety and then distributed randomly to the different groups in the experiment. In the memory study, “anxiety level” would be called a matching variable Example: Matching – Example Experiment comparing two forms of writing (emotional and control) on disease progression in HIV+ patients Match patients on the following: Each member of matched-pair then assigned to the different methods The matching variables selected above are good because they are variables that likely correlate with disease progression and if left uncontrolled would threaten the internal validity of the study Within-Subjects Design o Every participant is exposed to every level of the independent variable Also called a repeated-measures design--Because everyone in this type of study is measured several times o Often used in studies of sensation and perception—physiological psychology E.g., Fig 6.1 Muller-Lyer illusions: Vary the orientations of the lines to see if the illusion is especially strong when presented vertically It makes more sense to make the orientation variable a within- subjects factor and give each participant a sequence of trials to cover all levels of the variable and probably duplicate each level several times o Pros Fewer people needed to fill all experimental conditions(convenience) E.g., 2 conditions and you need 20 people per condition Total N for between-subjects design=40 Total N for within-subjects design=20 Don’t need to worry about equivalent groups--any between- condition individual difference variance in different groups disappears. Comparing two golf balls for distance Study: (there is variability within each group, as reflected in the standard deviation for each) In first set: Recruit 10 professional golfers, randomly assign them to two groups of 5 each golfer hits one ball or the other—between subjects design: three possibilities: the golf ball, individual difference and chance In second set: used just the first five golfers, and each pro hit ball 1, and then ball 2—within subjects design, individual difference can be eliminated o Cons Sequence or order effects Once a participant has completed the first part of a study, the experience could influence performance in later parts of the study In the last study: the second set could be due to practice or warm-up effect, Or perhaps the pro detected a slight malfunction in his swing at ball 1 and corrected it for ball 2. Or perhaps the wind changed. Progressive Effects After trial 1 performance is better on trial 2 (practice effect) After trial 1 performance is worse on trial 2 (fatigue/boredom) Carryover Effects A/B has different effect than B/A: In a study, study with two basic conditions, experiencing the first condition before the second might affect the person much differently than experiencing the second before the first. Examples: the effects of noise on a problem- solving task Within-subjects design IV: Unpredictable noise(UPN) and predictable noise(PN) People in UPN first and then PN, if they do poorly in UPN, the poor performance might discourage them and carry over to condition PN People in PN first and then PN, if subjects might do well, some of the confidence might carry over to the second part of the study— UPN Thus, performance in condition UPN might be much worse in the sequence UPN–PN than in the sequence PN–UPN. Use the between- subjects design Controlling Sequence Effects o Counterbalancing Present conditions in more than one sequence Primarily used to minimize progressive effects Why is it less effective at minimizing carryover effects? o Types of counterbalancing Testing only once per condition Complete counterbalancing – every possible sequence of conditions is used at least once. Problem: complete counterbalancing need to recruit a lot of subjects The total number of sequences needed can be determined by calculating X! 4! = 4 × 3 × 2 × 1 = 24 Partial Counterbalancing – a subset of total possible sequences are used. This can be accomplished by sampling from the complete set of all possible orders or, more simply, by randomizing the order of conditions for each subject. Sampling from all possible orders guarantees that no one order will ever be repeated; randomizing the order for each subject does not carry that guarantee. The balanced Latin square: This device gets its name from an ancient Roman puzzle about arranging Latin letters in a matrix so each letter appears only once in each row and once in each column; he number of rows in a Latin square is exactly equal to the number of elements in the study that are in need of counterbalancing; When using Latin squares, it is important for the number of subjects in the study to be equal to or a multiple of the number of rows in the square. Here’s an example of a 6 × 6 square. Testing more than once per condition For participants to experience each condition more than once. This often happens in research in perception and attention Example: Müller–Lyer illusions: 4 conditions: A = horizontal; B = 45° to the left; C = 45° to the right; D = vertical Participants in the study are shown the illusion on a computer screen and make adjustments to the lengths of the parallel lines until they perceive the lines to be equal. Reverse counterbalancing – for each participant, present the conditions in one order and then again in the reverse order A-B-C-D D-C-B-A Example: J. Ridley Stroop when shown color names printed in the wrong colors, you were asked to name the color rather than read the word. Block randomization – every condition occurs once before any condition is repeated a second time. Within each block, the order of conditions is randomized. first ¼ chance to be selected, second 1/3, then 1/2, finally 1 B-C-D-A C-A-D-B Research Example—counterbalancing with block randomization A study(Hagemann, Strauss, and Leißing) suggests that hose who referee combative forms of athletics can be influenced in their scoring by the colors worn by combatants. IV: red or blue protective gear; within subjects design Block randomization: Each referee saw a block of the 11 videos in a random order and then saw a second block of the same 11 videos, also in a random order, but the researchers digitally switched the colors of the protective gear. Half the refs saw each block first. Thus, referees saw the same tai kwon do event twice, with players dressed in different colors. Other studies have proposed that red suggests aggressiveness and blue projects calmness, though, so it is conceivable those associations played a role in the biased judgments. Methodological control in developmental research-- developmental or child psychology o When independent variable is age o Two types of design to consider Cross-sectional design—A between-subjects design E.g., A cross-sectional study comparing the language performance of 3-, 4-, and 5-year-old children would use three groups of children. Pros: save time Cons: Problem of cohort effects: A cohort is a group of people born at about the same time. The differ is not only in chronological age, but also in terms of the environments in which subjects were raised, or different life histories Longitudinal design-- a single group of subjects is studied over time. A within-subjects design E.g., a single group of subjects is studied over time. Cons: Attrition can be a problem Feasibility of collecting data for 3 years? maynot o To balance cohort and attrition problem-- cohort sequential design: A group of subjects is selected and retested every few years, and additional cohorts are selected every few years and also retested over time. Comparing the data in the rows gives you longitudinal designs, while comparing data in columns (especially 2020, 2025, and 2030) gives you cross-sectional comparisons. Comparing the rows enables a comparison of overall differences among cohorts. Example: Seattle Longitudinal Study— K. Warner Schaie (2005): The initial cohort had 500 people ranging in age from their early twenties to their late sixties. The study has added a new cohort at 7-year intervals ever since 1956 and has recently reached the 50- year mark. The study: box 6.1—the record for repeated measures Controlling for the effects of Bias: o Bias = preconceived expectation about what should happen in an experiment Two categories: those affecting experimenters and those affecting research participants. These two forms of bias often interact. o Experimenter Bias: As experimenter you might (inadvertently) interact with participants in a way that will make them behave in a way that will confirm your hypotheses Example – Rosenthal studies of picture perception with experimenter expectancy as independent variable Positive bias expectancy --- Participants give higher ratings Negative bias expectancy – Participants give lower ratings How is this possible? Experimenters can innocently communicate their expectancies in a number of subtle ways Participants read a frown or a smile of experimenter Some ways experimenter biases might be communicated Ways in which instructions are given Describing anchors on scale Facial expressions when answers given Participant changes response to try to get “positive” reaction from experimenter Personality characteristics E.g., preschoolers’ perform better on cognitive task when experimenter is “caring” versus “indifferent” Experimenters can be shown to influence the outcomes of studies in ways other than through their expectations. The behavior of participants can be affected by the experimenter’s race and gender as well as by demeanor, friendliness, and overall attitude o Ways to control for experimenter bias Minimize contact between experimenter and participant Computerize assessment--mechanize procedures as much as possible Double-blind procedure: neither the experimenters nor the participants know which condition is being tested on any particular trial— hence the designation double. Experimenters are kept in the dark (blind) about what to expect of participants in a particular testing session. Have research assistants who are “blind” to hypotheses Hard in practice though…. E.g. Williams and Bargh (2008)’s study-- experiences of physical warmth would increase feelings of interpersonal warmth, without the person’s being aware of this influence IV: a cup of either hot or iced coffee Ran the study using a double blind: they created physical warmth and coldness in subjects in a way that kept experimenters uninformed about which condition was being tested. Result: Participants whose hands had been warmed subsequently judged a neutral third person as warmer in personality than those given cold hands. E.g., High Positive vs. High Negative Feedback Research example: if the cognitive decline, as the day wears on, could be neutralized by America’s favorite drug—caffeine. IV: a caffeine group or a decaf group, and morning or afternoon Double blind: the experimenters administering the memory tests did not know which participants had ingested caffeine and the seniors did not know which type of coffee they were drinking. Result: Time of day did not seem to affect a short-term memory task, but it had a significant effect on a more difficult longer-term task completed immediately after the short-term memory task o Participants(subjects) bias: Knowing that you are in an experiment can make you change your normal behavior Hawthorne Effect (read Box 6.2) Belief of participant that they are part of special group and focus of attention—They were part of an experiment. Regardless of whether changes in independent variable are positive or negative get positive outcomes NOT due to independent variable BOX6.2-- Relay Assembly Test Room study: the purpose of the study was to investigate factors influencing worker productivity. (Five workers did the actual assembly, and the sixth supplied them with parts. At various times there were changes in the scheduling of rest periods, total hours of work, and bonuses paid for certain levels of production. Supposedly, the workers remained productive because they believed they were a special group and the focus of attention—they were part of an experiment). Efforts to present self as a good subject Demand characteristics: If hypotheses are too obvious to participants, they no longer act naturally; instead, they behave the way they think they are supposed to behave, making it difficult to interpret the results; Participants serving in all of the conditions of a study have a greater opportunity to figure out the hypothesis. Hence, demand characteristics are potentially more troublesome in within-subject designs Evaluation apprehension: participants want to be evaluated positively, so they may behave as they think the ideal person should behave. In a helping behavior study: Demand characteristic-- astute participants might guess they are in the condition of the study designed to reduce the chances that help will be offered—the experimenter doesn’t want them to help. Evaluation apprehension: altruism is a valued, even heroic, want to help One study has suggested that when participants are faced with the option of confirming the hypothesis and being evaluated positively, the latter is the more powerful motivator Controlling participant biases Minimize demand characteristics Use of deception to get participants to behave more naturally Use of placebo control group: This procedure allows for comparison of those actually getting some treatment (e.g., a drug) and those who think they are getting the treatment but aren’t. If the people in both groups behave identically, the effects can be attributed to participant expectations of the treatment’s effects. Everything same except content of treatment As much special attention etc. Manipulation Check: Ask participant what they think the hypothesis is Conduct field researches Chapter 7 Experimental design I: Single-Factor Designs Single Factor—Two levels o Single-factor designs have one independent variable Independent Groups o The independent variable is manipulated o Effectiveness of weight-loss programs Independent variable = Weight loss consequence Level 1 = self-reward for weight loss Level 2 = self-punishment for weight loss Dependent variable = pounds lost over 8 week program What would you expect data to look like after 8 weeks? Why? Why is this an example of a study that has to be conducted as a between subjects design? Between subjects, manipulated, randomly assignment Research example: Kasser and Sheldon (2000) :investigated the relationship between insecurity and one’s materialist leanings. Independent group 1-factor design, between subjects design, randomly assignment, IV: Those in the “mortality-salience” group were asked to write an essay on the “feelings and thoughts they had concerning their own death” Those assigned to the second group wrote an essay on their “thoughts and feelings regarding listening to music” DV in the study included estimates they gave of their future income, the value of their future homes and possessions, and how much they’d spend per year on travel. Result: The prediction was that those participants thinking of their own future demise (the mortality-salience group) would feel temporarily insecure, and the insecurity would trigger greater estimates of future financial well-being. Matched Groups o Simple random assignment is used to create equivalent groups o Effects of sleep deprivation on responses to interrogation questions o Independent variable = sleep deprivation Level 1 = Deprivation (21 hours awake in lab) Level 2 = Normal sleep pattern at home Matching variable = ? To control for what? How does this help clarify conclusions? Dependent variable = score on suggestibility tests Research study: Kroeger, Schultz, and Newsom (2007). They developed a video peer- modeling program designed to improve the social skills of young children (ages 4 to 6) with autism. IV “Direct Teaching” seeing a video of children modeling social skills and “ Play activities” group did not see the videos Matching variable: children with autism. Due to different levels of autism, a matching procedure was followed by random assignment. . Using a standard scale for measuring autism, the Gilliam Autism Rating Scale, they created pairs of children with matching levels of functioning, and then randomly assigned one of each pair to each group Nonequivalent groups-- Ex post facto design o Ex post facto design: the subjects in the study are placed into the groups “after the fact” of their already existing subject characteristics. o Matching followed by random assignment creates equivalent groups; matching in an ex post facto design makes the groups more similar to each other, but we cannot say equivalent groups are the result because random assignment is not possible with the subject variables o Are cognitively gifted children also gifted socially and emotionally? Independent variable = degree of giftedness Level 1 = Gifted (IQ = 130 +) Level 2 = Average (IQ = 90-110) Dependent variable = Social/Emotional Problem Solving Test Score Research study: Examines the effects of brain damage that results from an accident. McDonald and Flanagan (2004) investigated the abilities of 34 subjects with TBI (traumatic brain injuries)to process and understand social exchanges. the researchers tried to select subjects so the two groups would be as similar as possible, except for the brain damage; in this case they selected control group subjects that were “matched on the basis of age, education, and gender” Repeated Measures—within subject o Within subjects design: (a) requires fewer participants, (b) is more sensitive to small differences between means, and (c) typically uses counter- balancing to control for order effects. o Each participant in the study experiences each level of the independent variable o Research Example: a study of motion perception and balance (Lee and Aronson) within subjects, single factor, and repeated-measures design; alternating sequence What are the effects of a moving environment on children’s balance? Independent variable = direction of motion of “Moving Room” Level 1 = forward motion (10 trials) Level 2 = backward motion (10 trials) Dependent variable = direction of infants’ body lean or fall Between subjects multilevel design: Box7.1 study “stroop effecr”: a within-subjects design with one independent variable, tested at two levels and using reverse counterbalancing Study1: IV RCNb(Reading color names printed in black) and RCNd(Reading color names where the color of the print and the word are different); using reserves counterbalancing to avoid sequence effects. DV: naming time. Study2: naming the colors rather than reading color names IV: NC( naming color test) and NCWd(naming color of word test where the color of the print and the word are different)DV: naming time. Interested in the difference between the mean scores. o Does one level of your independent variable “do better” than the other? Inferential statistics (like a t-test) will tell you whether this difference is larger than you expect due to chance alone o The t test examines the difference between the mean scores for the two samples and determines (with some probability) whether this difference is larger than would be expected by chance factors alone t test for independent groups Independent groups design Ex post facto design t test for dependent groups matched groups design repeated-measure design Single factor—more than two level: o single-factor multilevel designs. o Pros: enable the researcher to discover nonlinear effects. multilevel designs are more informative and often provide for more complex and interesting outcomes than two-level designs. Box7.2 Yerkes-Dodson law: nonlinear effect, U shape, in a study with arousal level as an independent variable Research example: Balloon study--Bransford and Johnson’s (1972) study illustrates these context effects. In their study, participants were asked to comprehend a paragraph. Bransford and Johnson found that comprehension could be improved by adding some context (meaning). They designed a single-factor multilevel independent groups study with five levels of the independent variable. IV/DV: Adding the other is able to rule out (i.e., falsifying) alternative factors: o No context 2reps rules out that 1. It isn’t context that matters, it is just practice. o Context after rules out that It is just showing the cartoon that matters, because the cartoon provides retrieval cues during the memory task that are not present for the repetition groups. o Partial context rules out that: The cartoon content just provides additional information that helps subjects, during the test, recall more of the content. Within subject multilevel design: Research example: Retest “ Mozart effect”: Improvements in spatial skills could follow from listening to music by Mozart 3level IV: listening to Mozart for 10 minutes/ listening to a recording of soothing environ- mental sounds for 10 minutes/ not listening to anything This is a single-factor, multi- level, repeated-measures design. Authors chose to use a 3 × 3 Latin square, with 12 participants randomly assigned to each row of the square; The average number of digits correctly recalled was virtually identical for all three condition, but find the practice effect Presenting the data Three options: o Write the numbers out in the text of your report (e.g., “Women offered help more quickly (M = 38 s, SD = 12 s) compared to men (M = 64 s, SD = 22 s).” Alright when independent variable has only 2 or 3 levels…but hard to present clearly with > 3 levels o Make a table of the results Condition Mean Score Standard Deviation Female 38 12 Male 64 22 o Make a graph Dependent variable on Y axis (vertical) and levels of independent variable on X axis (horizontal) 100 80 60 40 20 0 Males Females Graphs can be especially striking if there are large differences to report or if nonlinear effects (e.g., Yerkes–Dodson Law) occur or if the result is an interaction between two factors. Tables are often preferred when data points are so numerous that a graph would be uninterpretable or when the researcher wishes to inform the reader of the precise values of the means and standard deviations; they have to be guessed at with a graph. o Tips for graphing: Bar graph or line graph? If continuous variable: Line graph preferred; bar graph acceptable. If discrete variable: Use a bar graph; a line graph is inappropriate. Analyzing Single-Factor, Multilevel design o Problem with computing multiple t-tests to compare all possible pairs of conditions? The difficulty is that completing multiple t tests increases the risks of making a Type I error—that is, the more t tests you calculate, the greater the chances are of having one accidentally yield significant differences between conditions. o In Single-Factor, Multilevel design, we use ANOVA Oneway ANOVA Followed up with post-hoc tests RECALL: Statistical decisions made based on ratio of two general types of variability: Ratio: Between conditions variability (systematic + error) Within conditions variability (error) Want larger b/w conditions vs. within conditions to show effect of treatment on outcome o One-way ANOVA for independent groups: multilevel independent groups design /multilevel ex post facto design o One-way ANOVA for repeated measures: multilevel matched groups design / multilevel repeated-measures design Control group design: Placebo Control Group o Placebo = substance that appears to have a specific effect but in fact is pharmacologically inactive o Placebo control group Participants led to believe they are receiving some treatment when in fact they aren’t In a study to determine if alcohol slows reaction time. We need placebo control group, because participants might hold the general belief that alcohol will slow them down, and their reactions might be subtly influenced by that knowledge. Waiting List Control Group o Used in treatment outcome studies E.g., effect of therapy on ADHD symptoms E.g. In the Miller and DiPilato (1983) study, he effectiveness of two forms of therapy (relaxation and desen- sitization) to treat clients who suffered from nightmares. DV: relaxation, desensitization, control; both forms of therapy produced a reduction in nightmares compared to the wait list control subjects. At the end of 15 weeks, those in the wait list control group began treatment. o Those in wait list control group always get the treatment at the end of the study o Why is a random control group not sufficient? Some might argue it is unethical to put people into a wait list control group because they won’t receive the program’s benefits right away and might be harmed while waiting. Research example: Merikle and Skanes (1992) evaluated the effectiveness of self-help weight loss audiotapes DV: the placebo control group (n = 15) thought they were getting a subliminal tape, but in fact they were given one designed to relieve dental anxiety; The experimental group participants (n = 15) were given a subliminal self-help audiotape; the wait list control (n = 17), was told “that the maximum number of subjects was currently participating in the study and that... they had to be placed on a waiting list” In this study, the wait list control group had the effect of evaluating the strength of the placebo effect and providing an alternative explanation for the apparent success of subliminal tapes. Yoked Control Groups o Used when participants in experimental group are exposed to varying number of events or for a variable amount of time o Each member of control group is “yoked” (= matched) to a member of the experimental group o RESULT = for groups as a whole, time spent participating or types of events encountered is kept constant Research example: the study was designed to evaluate the effectiveness of a psychotherapy technique that was popular (but controversial); The therapy is called “eye movement desensitization and reprocessing,” or EMDR. It is said to be effective as a treatment for anxiety disorders. using a matching procedure to create equivalent groups (matching them for age, sex, and the type of traumatic event they reported), randomly assigned them to an experimental and a yoked control group; Participants in the control group were yoked in terms of how long the session lasted, so if a subject in the EMDR group took 25 minutes to reach a SUD level of 0-1, a subject in the yoked control group would participate in the control procedures for 25 minutes. The control group did everything the experimental group did. Result: a placebo effect is probably lurking behind any alleged success of EMDR. Chapter 8: Experimental Design II: Factorial Designs Factorial design o A study with more than one independent variable (or factor) o In a study of social anxiety in adolescent boys you are interested in the effects of (1) social partner familiarity (familiar vs. unfamiliar), and (2) sex of social partner (male vs. female) Identifying Factorial design: o a numbering system that simultaneously identifies the number of independent variables and the number of levels of each variable. o This design is called a 2 x 2 FACTORIAL DESIGN: Each number indicates the number of levels of each independent variable So the first factor has 2 levels (familiar vs. unfamiliar) and the second factor has 2 levels (male vs. female) o a 2 × 3 (read this as “two by three”) factorial design has two independent variables; the first has two levels and the second has three. A more complex design, a 3 × 4 × 5 factorial, has three independent variables with three, four, and five levels o The combination of levels of the independent variables define all of the possible conditions of the experiment The term levels refers to the number of levels of any one independent variable. In factorial designs, the term conditions equals the number of cells in a matrix like the one you just examined. 2x3x4 =24; 5x2x3= 30 Outcomes—Main effects and interactions Main Effects: o Main Effects = the overall effect of a particular independent variable o A main effect is the difference between the means of the levels of any one independent variable. o The main effect of one factor involves using the data for all levels of the other factor o Main effects have nothing to do with levels, only related with IVs 5IV 5 main effects o It takes an analysis of variance (ANOVA) to make a judgment about whether the differences are significant statistically or due to chance. Research Example: test this “closing time” idea and to determine whether it applies to both men and women Two main effects (gender and time), The ratings, made at 9:00 p.m., 10:30 p.m., and 12:00 midnight; Both main effects were significant (remember that main effects are determined by examining row and column means) Result: a closing time effect occurred for men, but it did not occur for women. Interaction: o The real advantage of factorial designs is the ability to show interaction effects o The effect of one independent variable on the dependent variable differs depending on the level of another independent variable o Example: Means are same, no main effects; has an interaction: the effect of one variable (course type) depended on the level of the other variable (major). Factorial designs can be more informative than single-factor designs. o How many interaction effects are possible in a factorial design? Research example: Row and column means were close (12.8 and 13.5), and they were not significantly different from each other. So there were no main effects.Has the interaction: he effect of one factor (where they recalled) depended on the level of the other factor (where they studied). Interaction sometimes trump main effects Interactions sometimes make the main effects Example: the effect of caffeine on the memory of elderly subjects who were self- described “morning people.” In this experiment, both main effects were statistically significant. In fact, caffeine’s only advantage was in the afternoon. The interaction is the only important result. Combinations of main effects and interaction 1. There is a main effect for type of training factor; There is no main effect for presentation rate. No interaction 2. There is a main effect for the presentation rate factor. But there was no main effect for type of training. No interaction. 3. Main effects for both factors occur, no interaction. 4. Two main effects, but interactions can trump main effects when the results are interpreted, an interaction between type of training and presentation rate. the interaction may have been influenced by a ceiling effect, To test for the presence of a ceiling effect, you could replicate the study with 50-item word Two main effects, an interaction. The interaction is to say that at the fast rate, the imagery training is especially effective Lines on the graph are parallel, then no interaction; If the lines are nonparallel, however, an interaction probably exists. Whether an interaction exists (in essence, whether the lines are sufficiently nonparallel) is a statistical decision to be deter- mined by an ANOVA. Mixed factorial design o Factorial designs can have: all between subjects variables; all within subjects variables; at least one of each ( = MIXED factorial design) o Think about control issues related to each design type (i.e., equivalent groups and sequence effects) o Decision about whether to use counterbalancing when have a within subjects variable depends on purpose of that variable No counterbalancing when sequence (i.e., trials or time) are variable of interest o Text Case Study 18 = Mixed Factorial WITH counterbalancing A study with two manipulated independent variables: self-efficacy and “looming.” 2 x 2 mixed factorial design; random assignment IV1 = self-efficacy= sense of competence in dealing with life’s problems , which is manipulated by (1) can leave room and swat spider vs (2) can’t move away; IV2 = looming, which is manipulated by film of spider(1) not moving or moving away or (2) moving towards you (=looming); DV = self-reported fear Counterbalanced order of presentation Both main effects were statistically significant; an interaction; Interactions are sometimes more important than main effects o Text Case Study 19 = Mixed Factorial WITHOUT counterbalancing o E.g., effects of old knowledge on ability to remember new information (= proactive interference) Especially strong effect if type of knowledge is the same (e.g., phone number) between old and new but interference effect should be reduced if type of stimulus changes Each person gets 4 trials Each trial = view 3 news stories, work on distractor task, recall as much about stories as possible Randomly assigned to release versus control group Release condition (domestic politics, domestic politics, domestic politics, foreign politics) versus control (domestic, domestic, domestic, domestic) Reason for repeated- measure:The effect only showed after third trail Before 3 trail the effect is due to the sequence of trail ; After 3 trail the effect release, shows significant relationship Factorial Design with Subject AND Manipulated Independent Variables o P x E factorial design: designs can yield an interaction between the type of person (P) in the study and the situation or environment (E) created in the study P = person or subject variable; Subject variables are always a between subjects factor E = environment created by experimental manipulation; E can be either a between subjects or a within subjects factor Neither main effect would be significant, but an interaction clearly occurred. One effect happened for introverts, but something different occurred for extroverts. Specifically, introverts performed much better in
Are you sure you want to buy this material for
You're already Subscribed!
Looks like you've already subscribed to StudySoup, you won't need to purchase another subscription to get this material. To access this material simply click 'View Full Document'