PSY 313 EXAM 2 STUDY GUIDE
PSY 313 EXAM 2 STUDY GUIDE PSY 313
Popular in Intro. to Research Methodology
verified elite notetaker
Popular in Psychlogy
This 16 page Study Guide was uploaded by Bria Harris on Friday December 4, 2015. The Study Guide belongs to PSY 313 at Syracuse University taught by Amy Criss in Summer 2015. Since its upload, it has received 173 views. For similar materials see Intro. to Research Methodology in Psychlogy at Syracuse University.
Reviews for PSY 313 EXAM 2 STUDY GUIDE
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: 12/04/15
PSY 313 EXAM 2 STUDY GUIDE This study guide answers ALL of the questions on blank study guide she posted on BB Reliability and Validity Will Your Data Test Your Hypothesis Reliability Is the measurement consistent amp stable Will it produce the same result again and again Validity Does the experiment test what the experimenter says it does Reliability Types of Error Observed score or measure true value error Eg Exam score knowledge stress Random error just exists fine to have Can be in the instrument or in the person being measured Because it s random it cancels out with repeated measures Eg Weight Reliability Problems Error Random error a perfectly sound measure eg weight may produce different values Intrinsic noise drink half a gallon of water between 2 measurements you may weigh more Measurement error reading from a scale Because it s random it presumably cancels out with repeated measures of the same device Consistent error 0 Eg Scale always adds 2 pounds to the real weight How Do You Measure Reliability in a Research Study 1 Interrater reliability Same observations different raters Interrater reliability correlation between the rating of different judges Correlation doesn t mean the exact same values must be obtained But the same relative order is important 0 Take scores from 2 judges correlate them 0 Eg Give people an exam measuring verbal skills one rater gives person the score of 72 other gives a score of 98 low interrater reliability 2 Testretest reliability Same measurement different times Correlation between a test on different trails days weeks Eg Measure groups ability to block an air hockey puck 20 times to asses hand eye coordination Measure again 6 months later Correlate values r 027 weak 3 Split Half Reliability Same test different items that assess the same things Design a test that as different items that asses the same construct Eg 10th grade math test with 10 questions assessing trigonometry it is reliable Randomly split test into 2 halves question 15 then questions 6 10 Correlate scores against one another If reliable results will be positively correlated r 089 Reliability amp Validity Reliability refers to precision of data Are there systematic errors Validity refers to whether the data answers the research question Reliability is necessary for validity Validity Does the experiment answer the research question being asked Anything that makes you say umm maybe not thereat to validity Types External Do results generalize Construct Do operational definitions address constructs Internal Are alternative causes rules out External Validity All research is constructed at specific at a specific time in a specific place with a specific group of people How far can we generalize our results to different times places amp people Want to be broad amp global in our conclusions Therefore the question is Does the specific time place matter Could there be something that in uences our results Sometimes we study a specific group amp don t want to generalize then external validity isn t a concern Are the results more specific than suggested Population other participants cultures genders age etc Ecological from the lab to the real world Temporal to other periods of time of the day of the year or generations Does the study lack external validity Best answered by replication experiment in different time with a different population or outside of the lab Must be logical reason that some property of the situation affects the results 0 Threat eg volunteering at 2 AM volunteering at an unsafe location 0 No threat 0 Threat time period amp environment Example Study testing people s helping behaviors participant in the lab more likely to help someone that appears as though they need help if they re the only person around similar pattern emerges in the real world people are more likely to help if they re the only one around Construct Validity Does this experiment really measure what the research claims it measures Are the operational definitions reasonable measures of the construct Eg Elevation in heart rate good for stress Standardized test for teacher effectiveness Precautions Use logic Convergent validity shows that 2 measures of the same construct are correlated Divergent validity shows a lack of correlation to unrelated constructs Convergent Validity show that your operational definition is correlated with another measure of the same construct Eg Hypothesis Boys are more aggressive than girls 0 Operational definition of aggression Number of times child hits or kicks another person 0 Convergent measure different operational definition asking the teacher the overall rating of aggressiveness for each child should positively correlate Divergent Validity show lack of correlation with different potentially explanatory constructs Hypothesis amp Operational definition should do the same Want to rule out the possibility that ours is measuring a related construct activity level by counting total duration of running during recess should find no correlation Internal Validity Goal to say that manipulating the IV causes a change in the DV A threat to internal validity is anything that prevents making that causal conclusion Early in experimental design Extraneous Variable any variable that is not a DV or a IV random error Confound an extraneous variable that systematically varies with the manipulated random variable AND explains that data Eg A 3rd variable but in experimental situation Eg Pepsi wins but label was confounded with beverage L always coke amp S always Pepsi In experiment when randomly assigned labels L or S people choose S 85 of the time regardless of beverage Confound is the label Example Does Darkness affect Memory Performance Undergrads given a memory test at 2 PM light outside amp 2 AM dark outside amp performance was worse at 2 AM Therefore darkness impairs memory Is this a valid interpretation Any confoundsgt 0 IV time 0 DV memory performance 0 Confound Everyone in the 2 AM condition is sleepier than everyone at the 2 PM condition typically you re more tired at 2 AM than 2 PM Threats to Internal Validity Confounds History uncontrolled events that happen mid experiment eg experiment starts in July and mid September they start doing construction in the room next door to the lab for 2 weeks Maturation participant changes over time eg gains loses weight gets older gets taller Instrumentation change in the ability to use instrumentation or in the measurement itself eg experiment using a scale scale starts malfunctioning during experiment giving false weights Selection selfselection or improper assignment to condition eg due to improper assignment all men in an experiment are placed in group 1 and all women in group 2 Testing effects change in performance due to practice or fatigue with the material eg experiment testing memory participants begin to remember questions on test after 25 trials therefore it looks like their memory is good Introduction to Experimental Methods Recall that Causation Requires Correlation DirectionalityTemporal Precedence Cause must come before the effect Experiment can enforce precedence by manipulating a variable Eliminate other explanations Through experimental control If successful then strong internal validity Elements of Experimental Design Manipulation Change variable to create multiple treatment conditions IV Measurement A variable is measured for each condition DV Comparison Score in one treatment is compared to scores in another treatment 0 Consistent difference evidence for causation Control All other variables controlled 0 Eliminated confounds Example experiment observing the effect of exercise on weight IV type of exercise light or rigorous DV weight Control Don t work out at all What is Experimental Control A set of tools to allow an experimenter to eliminate confounds amp to ensure that the IV is causing changes in the DV not something else Tools for eliminating systematic variation 1 Holding constant 2 Matching 3 Random Assignment Control groups Review Extraneous Variable EV variable not controlled or manipulated adds random noise Confound systematically varies with the IV amp can explain the results Controlled Variable held constant equal the same across levels of the IV because it is a suspected confound Example Does larger shoe size cause an improvement in reading ability 0 IV Shoe size 0 DV reading ability 0 Levels Size 4 vs Size 6 o Confound Age Holding Constant Hold the value of a potential confound constant across all levels of the IV All items have the same value or a restricted range Shoe size reading ability age everyone is 8 years old Matching If you have identified a possible confound Match participants or stimulus material on that confounding variable across the levels of the IV For every level of the IV there exists one item person with the same value of the potential confound Shoe size reading ability Size 4 5 of each 5yearold 55 Year old 6 years old 65 years old etc Size 6 Same as above Different levels of IV Random Assignment Sampling How we select people from the population Assignment How we place that sample into the conditions of the experiment levels of the IV randomly assign people to particular conditions For each participant from the sample randomly assign them to condition ip coin random number generator Any difference between individuals should bequot equally spread across conditions by chance Shoe size reading ability not possible because confound exists in the person can t change age Random Assignment Random Assignment Any difference between individuals should bequot equally spread across conditions by chance Matching A tightly controlled way to cancel out differences in a potential confounding variables Often times we use both Match one or 2 highly important variables Randomly assign the rest Control Groups Taking dance lessons DV measure waltzing ability Control Do nothing or watch Dancing with the Stars 0 Experimental group take dance lessons for 6 weeks Observation 1 amp 2 Measure waltzing ability 0 If we see high scores on the DV can we say that lessons improved performance 0 If so by how much l Observation 1 l Observation 2 Experimental Group Control IMPORTANT For some reason I can t draw lines on the graph There should be a big arrow going from the top of the highest point of the control group 35 to the highest point of the experimental group 75 shows effect of I Between Subjects Design Independent Measures Design IV level 1 IV level 2 One level of IV per group of people Measure DV Measure DV DV is the same for each group Select a sample of people A Assign them to a group Advantages Clean measures No practice of fatigue effects Participants are naive to the experiment Can ask WHO questions individual differences Disadvantages Needs lots of resources more participants Risk differences between participants in each group that are not related to your IV Dealing with individual differences Potential confounds People Are Different Sometimes the hypothesis is about those differences Individual difference study also called quasiexperimental Exploits the differences that naturally exist between individuals 0 Male vs female 0 Young vs Old 0 Disabled vs Abled o Athlete vs Nonathlete Individual Differences Design Example Are left handed people better at remembering words Left Handed DV R h 1 t Subjects PSY e ersa COFFGC 205 Pool Right Handed DV Rehersal correct BUT sometimes you want to assume that everyone is the same amp ignore those individual differences Controlled difference study Attempts to eliminate or minimize difference between differences Use experimental control matching holding constant or random assignment 0 Example After Only Design We are only measuring the DV once at the end of the experiment Experimental Group Form sentences Subjects PSY 205 Pool Conrtol Group Rehersal Example Before After Design Pretest Posttest Test before and after the IV manipulation Measure difference and compare across groups Post test measure Leadership Ability Pretest measure Leadership Ability Experimental Group Subjects PSY 205 SU Students Posttest measure Post test measure Leadership Conrtol Group Leadership ability Ability Basic Controlled Difference Designs using Matching Example Dr Smith is studying he usefulness of a math tutoring program amp is concerned that IQ Is a confounding variable that higher IQ students might receive less benefit because they already do well on such tests doesn t rely on random assignment Subject Pool determine IQ for each sampled individual put them into pairs of matched IQ Randomly Split each pair No Tutoring Tutoring Effective but time consuming to find matching subjects Matching vs Random Assignment Is it worth matching a variable Safest but when 0 You have a small of participants 0 and concern about 1 potential confounding variable Or letting random assignment do the job for you Less work but typically safe when 0 Lots 40 of participants 0 And you aren t concerned with a specific IV Pitfalls to Avoid Differences between Environmental Condition control all other aspects of the experiment ONLY difference between groups should be level of IV Avoid observing your 2 groups at different times of day or different months of the year etc Experimenter Bias if experimenter is aware of the groups he or she may inadvertently in uence the results Gestures or tones of voice reinforce desired behaviors Misinterpret behavior in the direction of what is expected Use a single or double blind study Single blind experimenter unaware of which group each participant is in Double blind experimenter amp participants both unaware of group Between vs Within vs Mixed Subject Between each participant receives 1 level of the IV Within each participant receives every level of the IV Mixed multiple IV s at least 1 is between and at least one is within Within Subjects Design Repeated Measures Each participant is every condition receives every level of the IV Measure DV Treatment 1 after Level 1 IV One sample of participants Measure DV Treatment 2 after Level 2 IV Advantages No worries about differences between groups because there s a single group Statistically more powerful because differences between people is controlled for each person is their own control Disadvantages Prior behavior decisions may affect later behavior decisions 0 Changes over time see maturation in internal validity 0 Testing effects practice or fatigue 0 Carryover effects sometimes people drop out of the experiment over time amp may not experience every level of the IV Example Advantage 2 Baseline heart rate Elevated Difference 1 95 105 10 2 65 80 15 3 70 85 15 4 80 90 10 Individual differences are eliminated because everyone is in the same group eg every participant contributes to every level of the IV Within Subjects Experiment Disadvantages Previous Behavior Affects Later Behavior Changes over time outlined in internal validity eg maturation Testing effects Practice or fatigue Practice improvement from experience with DV Fatigue Decrement from experience with DV Carryover effects Experience with level of an IV impacts later performance Participant attrition Participants who start but don t finish becomes a selection problem Example of Testing Effects Look at slideshow on BB for images Repeated practice within DV can help Example A or harm Example B performance B 2 hour task where you press Z for even numbers and M for odd numbers Example of Carryover Effects Measure Measure Dr1nk Alcohol Driving Ability Dr1nk Water SUbject P001 Drviing Ability Drinking alcohol amp then drinking water one affects the other Avoiding Carryover Effects for Counterbalancing Design Balanced method for counterbalancing An equal of subjects receive each fixed order of conditions levels of IV Random method for counterbalancing Subjects get the different conditions levels of IV in a random order Both of these distribute any order effects evenly between the conditions levels of the IV Example Balanced Do blue shoes make you run faster Start with 100 runners o 50 run one mile in red shoes then one mile in blue shoes 0 50 run one mile in blue shoes then one mile in red shoes 0 Everyone is slower for 2nd mile regardless of shoe testing effect Example Random All subjects get conditions in a random order Eg on each trial a runner runs in a red or blue pair of shoes chosen randomly How Do You Decide if a Between or Within Subject Design is Best Individual differences anticipate a large individual difference amp worry about them as confounds Use within subjects completely controls for them Time related effects or carryover effects Do you expect one treatment to have large or long lasting effect on participants that may affect future conditions 0 Use between subjects participants only in one condition Fewer participants fining it difficult to participate Use within subjects requires less as each person is in all conditions Mixed Design More than 1 IV At least one of the IV s is a between subjects and at least 1 IV is within subjects Different participants get on of the levels of the 1St IV SAME participants then experience both levels of the 2rld IV Example Does mood affect memory Do you remember good things when happy amp bad things when sad Manipulate mood between subjects one group is happy amp the other is sad Each participant reads BOTH positive amp negative words within subjects Mood Positive Words Negative Words Happy Iulie amp Iared Iulie amp Iared Sad Cici amp Cedrick Cici amp Cedrick Factorial Design What Happens When You Combine IV s Manipulate 2 IV s in the same experiment Factor an IV Level the of conditions for than factor IV Eg A 2 x 2 design 2 IV s 2 levels Caffeine Level Sleep Deprived Well Rested Normal Coffee Decaf Coffee Empty Boxed DV is anxiety Terminology 3 x 2 2 IV s one with 3 levels 6 conditions 2 x 2 x 4 3 IV s 2 with 2 levels one with 4 levels 16 conditions Example 2 x 2 x 2 x 4 design Gender malefemale x handedness rightleft x wears glasses yesno x college year freshman sophomore junior senior DV reading comprehension Very powerful tests any combination of levels of each IV but 0 All possible options leaves a lot of room for statistical Example ukes In a 2 x 2 x 2 x 4 design 32 different conditions Does smoking increase your risk for heart disease between subjects design answer is yes 0 Yes 25 N0 9 risk Does birth control increase the risk for heart disease between subjects design answer is no Benefits Allows you to look for a main effect amp interactions between variables Main effects effect of 1 variable Interactions effect of 2 variables TOGETHER 0 One factor modifies the effects of the 2nd factor Example interaction of smoking amp taking birth control Smokes Birth Control Yes No Yes 30 8 No 20 10 Principles for Ethics in Research Methods Autonomy participant must have he unbiased right To know what they re participating in To decide whether or not to participate Informed consent ensures autonomy Violation of Autonomy Iewish Chronic Disease Hospital Trust Hypothesis debilitated patients will reject foreign cells 22 chronically ill patients who didn t have cancer were injected with live human cancer cells Physicians didn t inform the patients as to what they were doing because 0 They didn t want to scare patients 0 They thought the cells would be rejected Ensure a relationship of trust between participants amp researchers Confidentiality your identity is linked to data but we keep that link secret eg assign participants specific number Anonymity your identity isn t linked to your data Deception use only when absolutely necessary Debriefing educating participants about the design amp purpose of the experiment eg At the end of experiment explain what participants completed ask if they have any questions Iustice Burdens amp rewards are distributed equally often to demographics Eg College student participates in 2hour study possible burden reward is 10 Nonmaleficence amp Beneficence Minimizes risks amp maximize benefits Always costs or research Always benefits Benefit must outweigh costs Eg College student participates in 1hour study possible burden rewardbenefit is 15 Fidelity amp Scientific Integrity Researchers must be honest Misconduct amp fraud 0 Making up data 0 Leaving out relevant data Violations Eg Stapel social psychologist following an investigation he admitted he never ran experiments for some of his papers fabricated data for other papers life to colleagues and students Oversight Who determines if research follows guidelines People IRB IACUC Institutional Animal Care amp Use Committees Oversight Process Composed of unaffiliated member prisoner advocate child advocate religious affiliate faculty vet as appropriate 0 Submit detailed proposal yearly 0 Government may visit amp inspect animal labs Animals in Research APA Guidelines Summarized Qualified amp trained individuals Research must be justified Must minimize harm amp discomfort 3 R s of Animal Research 1 Reduce of animals used 2 Refine to cause least stress 3 Replace animals with other models Violations Tuskegee Airmen Study US Public Health Service studied syphilis in low income AA Video link on BB
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'