Psych 217 PDF

Psych 217 4 goals of psychological research - “ I wonder about the relationship between Covid-related isolation and anxietyˮ Description- What is the current status of isolation requirements? anxiety prevalence? Prediction- What...

Psych 217 4 goals of psychological research - “ I wonder about the relationship between Covid-related isolation and anxietyˮ Description- What is the current status of isolation requirements? anxiety prevalence? Prediction- What is the relationship? Who is most at risk? Does there appear to be a relationship between isolation and anxiety, generally? Is there predictable increase in anxiety that tracks where restrictions are implemented? Finding Causes- Does complying with isolation rules cause increase in anxiety? Explanation Basic research Focus: theories of behaviour. Often not immediately useful in everyday life. Theories inspired by real-world psychological functioning Applied Research Focus: maximising human happiness and psychological functioning. More immediately useful in everyday life. Borrows from theories developed through basic research. Terminology Psych 217 1  Idea- Unorganized principles or thoughts about social behaviour, cognition, emotion. Theory- Organised set of principles, which state how social behaviour operates. Comprehensive and systematic. Hypothesis- Statement of a relationship between two things (variables) Derived or taken from theories. Model- One meaning is the metaphors that we use to understand the human experience (or part of the human experience) Freuds Iceberg metaphor Idea- False memories are common Theory- Fuzzy trace theory 61 page tome, verbatim memory- relatively accurate. Gist memory- relatively inaccurate. Hypothesis  Time pressure increases use of gist memory and interferes with verbatim memory. Prediction- Assertions about what will occur in a specific study 3 Predictions  Nondirectional prediction “There is a relationship between schizophrenia diagnosis and viral loadˮ Low risk, prediction is falsified only if no relationship  Directional prediction “When viral load increases, schizophrenia will also increaseˮ Falsified if no relationship, or if relationship is negative  Point prediction: exactly how much one variable is related to another “When viral load increases by 1mg/cc, schizophrenia risk will doubleˮ Risky, very easy to falsify Psych 217 2  Parts of a research article  Abstract Brief 100200 word summary of this article What did you find? Why is it so important?  Introduction Where did you get your idea? What do we already know? What theories are useful? What hypotheses and predictions are you making?  Method Section Who was in your sample? What did you do? Design, procedure  Results section Key statistical tests Other interesting data Tables, figures Brief, objective description and interpretation  Discussion Summarize results and what you think they mean Psych 217 3  How we should think differently about the idea or theory? What should we do next?  Citations APA citation Loftus, E. F., & Palmer, J. C. 1974. Reconstruction of automobile destruction: An example of the interaction between language and memory. Journal of Verbal Learning and Verbal Behavior, 13, 585589 Variables  Started with an idea or theory  Derived a hypothesis that stated a general relationship between two concepts  Made a prediction about how two concepts interact specifically in our study Hypothesis→Prediction General,Abstract  Specific, Practical Conceptual→Operational Variable  An event, situation, behaviour or characteristic that takes on more that one value Conceptual  Concept or meaning of a variable Operational- A definition of the variable in terms of the techniques used to measure or manipulate it. Relationships between variables Positive Correlation coefficient = r Negative Psych 217 4  No Relationship Curvilinear Values of a variable Nominal Groups with no numeric properties Alive or dead, distraction-no distraction Quantitative Amounts or Quantities Numerical Ways of studying relationships  Non-experimental method Advantages Natural behaviour, no interference Example, littering behaviour Some variables cannot be manipulated Height/Sex Great for prediction Because we can closely replicate real-world Disadvantages No interference = no control No causal interference Psych 217 5  Problems Unknown direction of cause and effect Does exercise cause happiness? Or does happiness cause exercise? Unknown number of cause and effect relationships Uncontrolled third variables Maybe wealth can cause both happiness and exercise? If so, exercise and happiness will appear to be causal but not.  Experimental method Manipulate variable 1 d observe the effect on variable 2 Control, or keep constant, all other variable Allows cause and effect inferences Advantages Causal interference High research control Psych 217 6  Can observe small or transient effects milliseconds to identify words Disadvantages Unrealistic Crappy predictions Not always possible to manipulate variables Independent/Dependent variables Independent Variable “Causeˮ Manipulated/situational Variable that we can change Dependent Variable “Effectˮ “Dependsˮ on manipulated variable Aka outcome/response/measured variable Variables that we measure Establishing Causality  Correlation, or relationship, between IV and DV  Temporal precedences Causes must precede effects IV precedes DV  Eliminate alternative explanations, or confounds Similar to third variables in non-experimental) IV is the ONLY variable impacting DV Psych 217 7  Confounds A variable that researchers manipulate that is not directly related to the hypothesis. Confounds may covary/correlate with IV Confounds must cause change in DV Example Hypothesis- “Listening to relaxing music improves test performancesˮ Method- Randomly assigns participants to either wear headphones plays relaxing music or a control condition IV Relaxing music vs control DV Test performance Psych 217 8  Internal Validity  Is the cause valid? Foe Confound- Manipulated variable that is unrelated to the hypothesis Intended IV music vs no music Accidental IV headphones vs no headphones, ambient noise vs no ambient noise Friend Experimental Control- Only IV changes or varies Everyone wears headphones No ambient noise Planning an experiment How will you control participants variables Randomly assign Ss to ONE condition: between subjects design Pair Ss based on pretest: matched pair design Ss experience ALL conditions: within subjects design Types of experiments Between subjects design: Independent groups Between-groups Post test-only Matched pairs design: Dependent samples Psych 217 9  Pretest-Posttest Within-subjects design Repeated measures Within groups Independent groups Each group is randomly assigned Participants experience only one level of the manipulation Compare results from each group Matched pairs Measure additional variable in a pretest that we wish to control with certainty Perceptual processing speed Experience matched pairs Measure variable that you believe a priori will be important to control for Pretest-Posttest Reasons to use matched pairs Study can only recruit small sample Ensure successful random assignment on key participants fMRI/EEG  pretest necessary so that participants understand tasks Ensure high-quality scanning data High study mortality expected Ensure Ss dropouts don't create unequal conditions Challenges Pretest may change psychological experience of IV Psych 217 10  Practicing colour task with nonsense letters may change behaviour (likely help performance in post test) Solution Solomon four group design: A comprehensive experimental approach that combines both pretested and non-pretested groups to control for potential testing effects. This design includes four groups: 1 pretest, treatment, posttest; 2 pretest, no treatment, posttest; 3 no pretest, treatment, posttest; and 4 no pretest, no treatment, posttest. By comparing these groups, researchers can assess the impact of both the treatment and the pretest itself on the outcome measures. Experiment repeated measures All participants together in a single group All participants see each and every level of the IV Repeated-Measures: Comparisons Repeated-measures is MUCH more powerful 16 Ss for repeated-measures vs 450 for between-subjects Matched pairs somewhere between repeated measures and between- subjects Random assignment of Ss unnecessary for repeated measures Participants variables automatically, and completely controlled Between subjects need 3050 Ss for successful random assignment Repeated-measures subject to order effects, a type of confound Participants develop strategies for conditions Comparing designs Repeated-measures is much more powerful Psych 217 11  16 participants vs 450 for independent groups Matched pairs somewhere between Random assignment unnecessary for repeated measures Participant variables automatically controlled for Independent groups need 3050 participants per condition Repeated measures subject to order effects Participants develop strategies for conditions Carryover effect: Ss becomes aware of the manipulation Manipulation awareness Participants know what is being manipulated Demand effect- Ss behaves in a way they think experimented wants Reactant effects- Ss behaves opposite from what experimenter wants Order effects Performance may depend on order of the conditions, even without manipulation awareness Practice effects- Performance improves on later tasks Fatigue effect- Performance may decline or inattention may develop later in experiment Contrast effect- Participants contrast their response against previous manipulation. Repeated measures Challenge: Order effects Psychological experience is confounded with the order of manipulations Example- taste test for 6 different flavours of seltzer water Psych 217 12  Solution- Counterbalanced design List all possible orders of the IV Ss complete one of these orders Mediator Variables Mediating variable helps explain the relationship between two other variables Mediating variable is changed by the IV, and in turn, causes changes in DV Studying mediating variables seeks to explain the relationship between a cause and effect relationship Which of the four goals is this type of study most related to? Description Prediction Finding Causes Explanation Practical Research Design Participant samples Who will participate in my experiment? Who can I generalise my results to? Psych 217 13  External validity concern How large should my sample be? Random assignment Statistical power Smaller effects require much larger samples If r=.8, then N11 If r=.2, then N266 Independent variable How strong should my manipulation be? Size of the effects we observe is a ratio of manipulation strength to noise Stronger manipulations are easier to see (planet size) Controlled experiments reduce noise (lens clarity) Recommendation: Start w/strong manipulations More likely to determine if your hypothesised relationship is present Follow up with more subtle operational definitions of IV What will my manipulation look like? Straightforward manipulations Simply showing participants research stimuli Survey research, most computer-based research Advantages: Simple; cost-effective; deception not needed Staged manipulations: Elaborate “playˮ to generate psychological situations Advantages: Experimental realism- Ss experience the situation as engaging Psych 217 14  Generalisability beyond lab will usually be high, good external validity Dependent Variables  Self-reports  Behavioural  Physiological Two general classes: Direct “Are you feeling anxious?ˮ (direct self-report) “Make as many free throws as you canˮ(direct behavioural) Indirect Measure heart rate (indirect physiological) Covertly observe participant behaviour (indirect behavioural) Different measures impose different psychological constraints Recommendation: Use multiple methods/measures Is the relationship true under different conditions Pilot studies Primary use  Debugging Are there logistical problems you didn't foresee? Probably! Is there data you'll be collecting useful? Ceiling effects or floor effects? Is the psychological experiences as you expect? Manipulation check Psych 217 15  NOT Is the hypothesis correct/'working'? Is the data “rightˮ or “wrongˮ? Demand Characteristics Participant awareness of hypotheses generally leads to compliance Example: Does exercising increase happiness? IV Do not exercise vs exercise 7x this week DV Self-reported happiness Is this a confound? Rules Confound variable varies with level of the independent variables Confound variable is causally related to DV Solutions to demand characteristics  Between-subject designs: Participants now only see one experiment condition Not foolproof, especially when working with highly intelligent and social participants  Filler Additional experimental procedures, items, or delays that obscure the hypothesis  Placebo or expectancy manipulations Create identical psychological experience, with and without the active ingredient.  Blinding: Fully obscuring either hypotheses and/or the experimental manipulation Psych 217 16  Single-blind design: Participants are unaware of their condition(eg placebo vs treatment) Double blind design: Both participants and experimenters are unaware Does the amount of likes a Tiktok has have an impact on the probability of someone liking the Tiktok? Independent Variable The independent variable is the amount of likes the Tiktok has. We are taking the following measures to operationally define the independent variable: making two versions of a google form with the same Tiktok videos the video will be manipulated to show high likes on one form and low on the other (high level likes: 200k+; low level likes: 02000 Dependent Variable The dependent variable is the rating on the probability scale of liking the Tiktok. Rating 14 1  Very Unlikely 2  Unlikely 3  Somewhat Likely 4  Very Likely Hypothesis Participants are more probable to like a video with more likes and less probable to like a video with fewer likes. Psych 217 17  We believe our results will yield a higher probability of liking a video with a high amount of likes because social validation and comforting to popular opinion encourages users to engage with the content. Our study avoids confounds by using our 4-point rating scale thus ensuring a consistent measurement and eliminating a “middle groundˮ. Our variables are able to be measured in the same way, but not manipulated. We are using random assignment (alternating forms) to divide the participants. We can confidently say our study, in its entirety, is minimal risk. Lecture 5 Multiple levels of IV Research idea- Does a happy mood lead to better memory? Procedure- participants read a story, then recall the story 2 IV levels: Read a story that is either happy or sad Results- Happy story is remembered better Why? Does happiness improve memory? Does sadness reduce memory? Multiple levels Experiment with more than 2 levels of IV Can help explain causal relationships (eg effect of happy vs sad mood on memory) Needed to detect non-linear relationships. Research question- Does time permit us to control racial bias in judgement? Psych 217 18  We'll start with simple design, and layer on complexity… Background- Racial stereotypes influence our judgement Weapons priming task Method- quickly determine whether a stimuli is a gun or a tool But each tool or gun target is preceded by a black or white prime face Results Black primes increase at which people misclassify tool as a gun Black primes speed up correct gun identifications Explanation Seeing black vs white faces activates threat related stereotypes When threat is active, we can misperceive benign objects as threatening Modified racial bias task Starting with validated measure of racial bias, but lets simplify it to include only critical psychological ingredients New operational definition of racial bias Original- People judged weapons vs tools. Simplified- People judge faces as threatening or not Conceptual variables are the same: Threat-related racial bias in judgement Psych 217 19  Designing the experiment Research question- Does time permit us to control racial bias? An important theory made a point prediction that 400ms of time was needed for control Theory's hypothesis- Variation in time below 400ms will not affect racial bias Our alternative hypothesis: Longer amount of times, even below 400ms, will reduce racial bias IV time Operational definition- Stimulus onset asynchrony or time between prime and target IV has 2 levels 150ms Psych 217 20  325ms DV Racial bias in judgement Operational definition Percentage differences in “threatˮ judgements comparing Black and white prime trials Greater differences indicate greater racial bias Psych 217 21  Lesson 6 Control conditions Not every study needs a “controlˮ condition 'Control' usually connotes absence of the treatment Using the term comparisons group may be more useful We compare our treatment against other treatments (sometimes these are controls) Important: The comparison groups we choose will contain what conclusions we can draw from our study This will be especially important in a two-group design! Psych 217 22  Control vs Comparison Research idea: Is my new therapy program useful? Procedure: Participants will be diagnosed anxiety will all complete an intake survey and then are assigned to one of two groups Treatment group: CBT/Mindfulness therapy for 2 months Comparison group: Absence of CBT/Mindfulness therapy Results: The CBT/ Mindfulness therapy produced a reduction in anxiety relative to control. The therapy is efficacious. Or  Waitlist control Results  The CBT/Mindfulness therapy produced a reduction in anxiety relative to a wait-list control. The therapy is efficacious above and beyond the effect of expecting to receive treatment/relief from symptoms Sham therapy (placebo) Results  The CBT/Mindfulness therapy produced a reduction in anxiety relative to a a sham therapy control. The therapy is efficacious above and beyond the known positive effect of meeting with a therapist and feeling that someone cares. Treatment as usual Results  The CBT/Mindfulness therapy produced a reduction in anxiety relative to a wait-list control. The therapy works better than the current therapeutic practices of the clinic. CBT but not mindfulness therapy for 2 months Results- Mindfulness is an important ingredient of the newly developed CBT/Mindfulness therapy. Psych 217 23  Single case baseline design  Estimate a baseline level of DV  Manipulate variables and observe changes in DV Anecdote: A science-minded friend with a mysterious symptom Stomach pains- where are they coming from? Every day, he tracked the levels of his stomach pain An unexpected patter emerged… Suspect:?? Solution: Menthol in Ricola cough medicine ABA reversal design Day 1 No menthol Day 2 Take menthol Day 3 No menthol ABA Frequently add another “Bˮ Multiple Baseline Design Seek to generalise results from single-case designs: Replicate across different participants Does menthol work for others with mystery stomach pain? Replicate across behaviours Does menthol negatively correlate with other types of pains? Replicate across situations Psych 217 24  Does menthol work both when sore throat is present and when it is not? Replicate across operational definitions of IV or DV Does menthol work in cough drop form and toothpaste? Quasi-experiments Resemble experiments, but weak internal validity. Quasi-experiments lack: Random assignment True control conditions Quasi experiments are used when Random assignment poses ethical problems True control conditions pose ethical problems Example: Program teaching children to read Research question: Is our reading program successful in teaching children to read?  One-group post-test only  One-group pretest-posttest  Non-equivalent control group  Non-equivalent control group pretest-posttest  Interrupted time-series  Control series One group posttest only Participants: 45 year old children  IV Reading program Psych 217 25  a One level  Posttest DV Reading comprehension Potential conclusion: “Children in reading program learned how to read!ˮ What are we not saying? What do we want to say? One-group Pretest-posttest  Pretest (baseline): Reading comprehension test  IV Reading assistance program a One level (one group)  Posttest DV Reading comprehension test Potential conclusion: “Children in reading program improves their reading scores!ˮ What are we not saying? Why not? Alternative explanations Maturation: Participants change over time Children become developmentally capable between baseline and DV Mortality: Participants drop out of studies Children who couldn't read dropped out of the study Testing/Practice effect: Pretest can influence posttests Reading pretest, not the program, increased performance on reading posttest History effects: Something happens outside the experiment that influences posttest New TV show about the reading starts Psych 217 26  Regression towards the mean: Extreme group tend to become less extreme because of measurement error Relevant if we selected extreme groups.. Instrument decay: Measure no longer quantifies the same concept at posttest Example: “Modernˮ Racism Scale 1986 Racism still exists, but the instrument no longer captures it. Non-equivalent control group Experiment with a control group, but lacks random assignment Selection differences always possible: Especially when participants choose their condition Example: Parents of children who are struggling with reading request treatment group Treatment group: More struggling readers than expected Control group: Fewer struggling readers than expected. Non-equivalent control group, pretest-posttest. Control group: Yes Random Assignment: No But… Pretest to ensure children do not differ on average with respect to reading ability Most important variable may be controlled Other variables likely differ and may or may not be important Highest internal validity of the designs thus far! Psych 217 27  Interrupted time-series Similar to single-case baseline design Multiple AB cases Example: Passage of opioid control bills States implement different variants at different times  Pretest: Incidence of overdose in the state  IV Time when bill was passed in state  Posttest: Incidence of overdose in the state Psych 217 28