Psych 101 notes - Operationalization, Experimental Design, and Ethics in Psychology

Chapter 1: Introduction

  • Central problem in designing psychology experiments: many interest areas are conceptual (e.g., a happy life) and must be operationalized into measurable variables.

  • Operationalization is turning a conceptual variable into a measurable one (an operational variable).

    • Operationalization is often imperfect and may require multiple measurements.

    • Example: measuring blood requires multiple physiological indicators, not just one (e.g., heart rate alone is not enough).

  • Key variables introduced: Independent Variable (IV) and Dependent Variable (DV).

    • IV: the variable that is deliberately varied across conditions (e.g., different doses of a drug).

    • DV: the measured outcome (e.g., behavior, heart rate, performance on a task).

  • Situational variables (testing context) should be matched to what makes sense for the measurement (e.g., testing coffee effects should reflect when people normally drink coffee).

  • Participant variables: humans bring lots of variability (age, gender, education, prior exposure to drugs, etc.).

    • Animals are useful because they are more uniform (e.g., same age cohort, same diet, same light/dark cycle).

    • In humans, aim to spread participant variables across IV levels using random assignment; noise should be roughly equally distributed across groups.

  • Sample size matters: larger samples reduce the impact of participant variables and random noise; small samples can let age or other factors drive results.

  • Ethical constraints: some questions cannot be answered with true experiments in humans (e.g., exposing participants to heroin ethically).

    • When true experimental manipulation isn’t possible, researchers use quasi-experimental or nonexperimental methods (correlational designs).

  • Correlation basics (nonexperimental): two variables move together but correlation does not imply causation.

    • Positive correlation: as X increases, Y tends to increase. Negative correlation: as X increases, Y tends to decrease.

    • Graphically: scatter plots with a line of best fit when relationships are clear; many psychology correlations are modest (around
      r \6{-}1 \,to\, 1, with typical values around \(0.3∗∗or<strong>−0.3</strong>).</p></li><li><p>Nonlinearrelationships:manyrelationshipsarenotlinear(e.g.,alcohol′seffectsonarousalarecurvilinear).</p></li><li><p>Curvilinearexample:inverted−Urelationship(highdosereducesarousalafteraninitialboost).Asimplerepresentationis<br>0.3** or <strong>-0.3</strong>).</p></li><li><p>Nonlinear relationships: many relationships are not linear (e.g., alcohol's effects on arousal are curvilinear).</p></li><li><p>Curvilinear example: inverted-U relationship (high dose reduces arousal after an initial boost). A simple representation is<br>y = -a x^2 + b x + c, \text{with } a>0</p></li></ul></li><li><p>Inpractice,correlationstellusaboutassociationbutnotcausationormechanism;trueexperimentsprovidecausalinference.</p></li><li><p>Chapter2foreshadow:theneedfortrueindependentvariablestoestablishcausality.</p></li></ul><h3id="b10ed47a−e552−4f47−8195−b390f644e827"data−toc−id="b10ed47a−e552−4f47−8195−b390f644e827"collapsed="false"seolevelmigrated="true">Chapter2:TrueIndependentVariable</h3><ul><li><p>Inpsychology,correlationsarecommonbuthavelimitations:theycannotestablishcausality,directionality,orcausalmechanisms.</p></li><li><p>Threecoreadvantagesofatrueexperiment(causalinference):</p><ul><li><p>Covariation:thecauseandeffectcovary;formally,forabinaryIVC,theprobabilityoftheeffectEdifferswithC:</p></li></ul></li><li><p>In practice, correlations tell us about association but not causation or mechanism; true experiments provide causal inference.</p></li><li><p>Chapter 2 foreshadow: the need for true independent variables to establish causality.</p></li></ul><h3 id="b10ed47a-e552-4f47-8195-b390f644e827" data-toc-id="b10ed47a-e552-4f47-8195-b390f644e827" collapsed="false" seolevelmigrated="true">Chapter 2: True Independent Variable</h3><ul><li><p>In psychology, correlations are common but have limitations: they cannot establish causality, directionality, or causal mechanisms.</p></li><li><p>Three core advantages of a true experiment (causal inference):</p><ul><li><p>Covariation: the cause and effect covary; formally, for a binary IV C, the probability of the effect E differs with C:P(E|C=1)
      eq P(E|C=0)orequivalentlyor equivalently ext{Cov}(C,E)
      eq 0.</p></li><li><p>Temporalprecedence:thecauseoccursbeforetheeffect.</p></li><li><p>Eliminationofalternatives:bycontrollingconfoundsandusingrandomassignment,otherplausibleexplanationsaresystematicallyruledout.</p></li></ul></li><li><p>Intrueexperiments,youcandemonstrateacausalrelationship:ifthecauseoccurs,theeffectoccurs;ifthecausedoesn’toccur,theeffectdoesn’toccur(undercontrolledconditions).</p></li><li><p>Temporalordermatters:experimentscanrevealbidirectionalrelationshipsthatcorrelationalworkcannot(e.g.,impulsivityanddruguseinfluenceeachotherinafeedbackloop).</p></li><li><p>Real−worldrealizations:manypublishedcorrelationsaremodest;strong,robustcorrelations(e.g.,near0.6–0.7)arerarebutmeaningful;still,causalinferencerequiresmorethancorrelation.</p></li><li><p>Keytakeaway:correlationalresearchisvaluablebutinsufficientforansweringcausalquestions;trueexperimentsprovideaframeworkforestablishingcause−effectchains.</p></li><li><p>Whentrueexperimentsaren’tpossibleforethicalorpracticalreasons,researcherssometimesrelyonanimalmodelstostudymechanismsandthentranslatefindingstohumans.</p></li><li><p>Recapofwhatatrueexperimentaffords:acausalclaimviacovariation,temporalorder,andrulingoutplausiblealternatives;somethingcorrelationscannotdo.</p></li></ul><h3id="e9a885ed−dd54−4f0a−a2a0−1d941f823f5e"data−toc−id="e9a885ed−dd54−4f0a−a2a0−1d941f823f5e"collapsed="false"seolevelmigrated="true">Chapter3:ConfoundingVariable</h3><ul><li><p>Definition:Aconfoundingvariableisanuncontrolledvariablethatvariesalongwiththeindependentvariableandcanbiasresults.</p></li><li><p>Example1:Dose−volumeconfoundinanimaldruginjections.</p><ul><li><p>Problem:asdrugdoseincreases,injectionvolumeoftenincreases.Theobservedimprovementinperformancecouldbeduetovolume(discomfort,physiologicalstress)ratherthandrugeffect.</p></li><li><p>Fix:keeptheinjectionvolumeconstantacrossdoses;adjustdrugconcentrationsothevolumeisidenticalwhiledrugamountchanges.</p></li></ul></li><li><p>Example2:Brainregioninactivationstudy.</p><ul><li><p>Themanipulation(inactivationofabrainregion)mayinducestresssimplyfromhandlingtheanimal,whichcanalterbehavior.</p></li><li><p>Fix:handleallanimalsuniformly,regardlessofcondition,toensurestressdoesnotcovarywiththebrainmanipulation.</p></li></ul></li><li><p>Example3:Drugself−administrationinisolatedrats.</p><ul><li><p>Isolatingratscreatesasocialenvironmentconfound;socialisolationitselfinfluencesdrug−seekingbehavior.</p></li><li><p>Insight:confoundscanleadtoimportantdiscoveries(e.g.,socialisolationasapredictorofdruguse).</p></li></ul></li><li><p>Generalpoint:evenwell−designedexperimentscanhaveconfounds;thegoalistominimizethemand,whenpossible,designstudiesthatisolatethevariableofinterest.</p></li><li><p>Technicalnote:thetermconfoundingvariableissometimesusedloosely;thestrictdefinitioninvolvesavariablethatmoveswiththeIV;otherrelatedtermsincludethird−variableproblemorhiddenvariable.</p></li><li><p>Practicaltakeaway:whenevaluatingastudy,checkforpotentialconfounds(e.g.,volume,handling,ordereffects)andlookforwhethertheresearchershavecontrolledfororrandomizedthesefactors.</p></li></ul><h3id="efe18013−2aae−49fe−bc4e−18fd09773105"data−toc−id="efe18013−2aae−49fe−bc4e−18fd09773105"collapsed="false"seolevelmigrated="true">Chapter4:DoseOfDrug</h3><ul><li><p>Indrugstudies,theIVisthedoseandistypicallyimplementedastreatmentarms(levelsoftheIV):aplacebo(nodrug),lowdose,mediumdose,highdose,etc.</p></li><li><p>Placebovsvehicle:placeboisano−drugconditiongiveninpillform(sugarpill)toensureparticipantscannottelliftheyreceivedthedrug;vehiclereferstothesalineorsaltsolutionusedforinjectionsinanimalstudies.</p></li><li><p>Demandcharacteristics:participants’expectationsaboutthedrugcanaltertheirbehavior;therefore,placebocontrolshelpisolatepharmacologicaleffectsfromexpectancyeffects.</p></li><li><p>Blinding:</p><ul><li><p>Single−blind:theparticipantdoesnotknowwhichdosetheyreceive.</p></li><li><p>Double−blind:neithertheparticipantnortheresearcheradministeringthedoseknowsthedoselevel;codesarehiddenandonlyrevealedafterdatacollection.</p></li><li><p>Purpose:reduceexpectationbiasandobserverbias;preventresearchersfromunintentionallycueingparticipants.</p></li></ul></li><li><p>Open−labelstudies:allpartiesknowthetreatment;usedwhensafetyconcernsorethicsrequiretransparency(e.g.,terminalillnesswhereanexperimentaldrugisinvolved).</p></li><li><p>Ethicalconstraints:completelyethicaltodomultipledrugdosesinhumansonlyundercarefuloversight;insomecases,highdosesmaybeethicallyunacceptable,leadingtolimiteddosinginhumanstudies.</p></li><li><p>Designchoicesbeyonddose:considerotherindependentvariablessuchassleepmanipulation,studyingtime,orsocialisolationastheIV,withdose−likeframing.</p></li><li><p>Experimentaldesignlanguage:IVlevelsarecalledtreatmentarms;eachlevelisaseparateconditiontowhichparticipantsarerandomlyassigned(between−subjects)oralllevelsaregiventothesamesubjects(within−subjects).</p></li><li><p>Practicaltakeaway:startexperimentaldesignbygroundingtheIVinadose−likestructureandthinkaboutblinding,placebo,andethicalconstraintsfromtheoutset.</p></li></ul><h3id="072c7afd−0bc5−4805−ae28−02b3ec30ed28"data−toc−id="072c7afd−0bc5−4805−ae28−02b3ec30ed28"collapsed="false"seolevelmigrated="true">Chapter5:ConsentOfDrug</h3><ul><li><p>Whyanimalresearchisvaluable:therearequestionsyoucannotethicallyaskinhumans(e.g.,heroinexposure,brainmanipulation)butareessentialtounderstandmechanisms.</p></li><li><p>Animalmodelsenableinvasiveprocedures,brainrecordings,andgeneticmanipulationthatwouldbeunethicalinhumans.</p></li><li><p>Ethicalgovernance:pre−approvalisrequired;justificationthatnoothermethodcouldanswerthequestion;usetheminimumnumberofanimals;animalsarecloselymonitoredandcaredforunderestablishedguidelines.</p></li><li><p>Biosafetyandoversight:university−levelanimalcarecommittees(CCAC/ACC),provincialandfederalregulatorybodiesensuredailywelfareandcompliance;researchersfollowinstitutionalcareprotocols.</p></li><li><p>Practicalstatistics:evenwithanimalmodels,thenumberofanimalsusediskepttoaminimum;animalresearchisasmallfractionofoverallanimalusecomparedwithagriculture.</p></li><li><p>Endpointandreplacement:researchersmustconsideralternatives(e.g.,simulations,insilicomodels)andonlyuseanimalsifnecessary.</p></li><li><p>Translationtohumans:animaldataoftentranslatewelltohumanbiology(e.g.,gamblingresearch)butstillrequirecautiousinterpretation.</p></li><li><p>Humansubjectsethics:</p><ul><li><p>Informedconsentisessential;consentisanongoingsocialcontract,andparticipantscanwithdrawatanytime.</p></li><li><p>Deceptionandconfederatesmaybeusedbutmustbeethicallyjustifiedanddebriefedafterthestudy.</p></li><li><p>Debriefingensuresparticipantsunderstandthestudy′spurposeandtheirrole.</p></li></ul></li><li><p>Deceptionandconsent:sometimesdeceptionisused(e.g.,withconfederates)butmustbeethical,non−harmful,anddisclosedindebriefing.</p></li><li><p>Opendiscussionsaboutconsent:participantsshouldbefreetoremoveconsentevenafterdatacollectioniftheyfeeluncomfortable.</p></li><li><p>Thebottomline:ethicaloversightandinformedconsentarenon−negotiable;thewelfareofparticipantsandanimalssupersedesresearchgains.</p></li></ul><h3id="993779d2−c8e8−4d7b−a4c9−28b7f7dd6a0d"data−toc−id="993779d2−c8e8−4d7b−a4c9−28b7f7dd6a0d"collapsed="false"seolevelmigrated="true">Chapter6:TheIndependentVariable</h3><ul><li><p>Drugstudiesinhumansaredifficultandoftenethicallyconstrained;multiple−dosedesignsareoftennotfeasibleinhumansunlessthedrugisalreadylegallyavailableandusedbythepublic.</p></li><li><p>Screeningconcerns:pre−screeningforhealthconditions(e.g.,heartconditions)isnecessarytoensuresafety;screeningproceduresmayneedtobalancethoroughnesswithethicaldisclosure.</p></li><li><p>Ethicsoutrankdataquality:safetyandethicalconsiderationsoverridetheidealityofthedata;researchersmustadaptdesigntoprotectparticipants.</p></li><li><p>Practicaltakeaway:inmanystudies,you’lldesignaroundthestrongestfeasibleIVconcept(e.g.,dose,sleepamount,orhoursstudied)ratherthanafullidealizeddose−responsedesign.</p></li><li><p>Thekeymessage:berealisticaboutwhatcanethicallybemanipulatedinhumans;useanimalmodelsorexistingclinicians’medicationswhenappropriatetoanswerquestionsthatcannotbestudieddirectlyinhumans.</p></li></ul><h3id="4862598a−4157−4901−9832−04d5effeba63"data−toc−id="4862598a−4157−4901−9832−04d5effeba63"collapsed="false"seolevelmigrated="true">Chapter7:DifferentIndependentVariable</h3><ul><li><p>DesigningexamsandpapersrequiresclearoperationalizationandjustificationforthechosenIV.</p></li><li><p>Theexamstressesthreescoringcomponents:content(≈35</p></li><li><p>Temporal precedence: the cause occurs before the effect.</p></li><li><p>Elimination of alternatives: by controlling confounds and using random assignment, other plausible explanations are systematically ruled out.</p></li></ul></li><li><p>In true experiments, you can demonstrate a causal relationship: if the cause occurs, the effect occurs; if the cause doesn’t occur, the effect doesn’t occur (under controlled conditions).</p></li><li><p>Temporal order matters: experiments can reveal bidirectional relationships that correlational work cannot (e.g., impulsivity and drug use influence each other in a feedback loop).</p></li><li><p>Real-world realizations: many published correlations are modest; strong, robust correlations (e.g., near 0.6–0.7) are rare but meaningful; still, causal inference requires more than correlation.</p></li><li><p>Key takeaway: correlational research is valuable but insufficient for answering causal questions; true experiments provide a framework for establishing cause-effect chains.</p></li><li><p>When true experiments aren’t possible for ethical or practical reasons, researchers sometimes rely on animal models to study mechanisms and then translate findings to humans.</p></li><li><p>Recap of what a true experiment affords: a causal claim via covariation, temporal order, and ruling out plausible alternatives; something correlations cannot do.</p></li></ul><h3 id="e9a885ed-dd54-4f0a-a2a0-1d941f823f5e" data-toc-id="e9a885ed-dd54-4f0a-a2a0-1d941f823f5e" collapsed="false" seolevelmigrated="true">Chapter 3: Confounding Variable</h3><ul><li><p>Definition: A confounding variable is an uncontrolled variable that varies along with the independent variable and can bias results.</p></li><li><p>Example 1: Dose-volume confound in animal drug injections.</p><ul><li><p>Problem: as drug dose increases, injection volume often increases. The observed improvement in performance could be due to volume (discomfort, physiological stress) rather than drug effect.</p></li><li><p>Fix: keep the injection volume constant across doses; adjust drug concentration so the volume is identical while drug amount changes.</p></li></ul></li><li><p>Example 2: Brain region inactivation study.</p><ul><li><p>The manipulation (inactivation of a brain region) may induce stress simply from handling the animal, which can alter behavior.</p></li><li><p>Fix: handle all animals uniformly, regardless of condition, to ensure stress does not covary with the brain manipulation.</p></li></ul></li><li><p>Example 3: Drug self-administration in isolated rats.</p><ul><li><p>Isolating rats creates a social environment confound; social isolation itself influences drug-seeking behavior.</p></li><li><p>Insight: confounds can lead to important discoveries (e.g., social isolation as a predictor of drug use).</p></li></ul></li><li><p>General point: even well-designed experiments can have confounds; the goal is to minimize them and, when possible, design studies that isolate the variable of interest.</p></li><li><p>Technical note: the term confounding variable is sometimes used loosely; the strict definition involves a variable that moves with the IV; other related terms include third-variable problem or hidden variable.</p></li><li><p>Practical takeaway: when evaluating a study, check for potential confounds (e.g., volume, handling, order effects) and look for whether the researchers have controlled for or randomized these factors.</p></li></ul><h3 id="efe18013-2aae-49fe-bc4e-18fd09773105" data-toc-id="efe18013-2aae-49fe-bc4e-18fd09773105" collapsed="false" seolevelmigrated="true">Chapter 4: Dose Of Drug</h3><ul><li><p>In drug studies, the IV is the dose and is typically implemented as treatment arms (levels of the IV): a placebo (no drug), low dose, medium dose, high dose, etc.</p></li><li><p>Placebo vs vehicle: placebo is a no-drug condition given in pill form (sugar pill) to ensure participants cannot tell if they received the drug; vehicle refers to the saline or salt solution used for injections in animal studies.</p></li><li><p>Demand characteristics: participants’ expectations about the drug can alter their behavior; therefore, placebo controls help isolate pharmacological effects from expectancy effects.</p></li><li><p>Blinding:</p><ul><li><p>Single-blind: the participant does not know which dose they receive.</p></li><li><p>Double-blind: neither the participant nor the researcher administering the dose knows the dose level; codes are hidden and only revealed after data collection.</p></li><li><p>Purpose: reduce expectation bias and observer bias; prevent researchers from unintentionally cueing participants.</p></li></ul></li><li><p>Open-label studies: all parties know the treatment; used when safety concerns or ethics require transparency (e.g., terminal illness where an experimental drug is involved).</p></li><li><p>Ethical constraints: completely ethical to do multiple drug doses in humans only under careful oversight; in some cases, high doses may be ethically unacceptable, leading to limited dosing in human studies.</p></li><li><p>Design choices beyond dose: consider other independent variables such as sleep manipulation, studying time, or social isolation as the IV, with dose-like framing.</p></li><li><p>Experimental design language: IV levels are called treatment arms; each level is a separate condition to which participants are randomly assigned (between-subjects) or all levels are given to the same subjects (within-subjects).</p></li><li><p>Practical takeaway: start experimental design by grounding the IV in a dose-like structure and think about blinding, placebo, and ethical constraints from the outset.</p></li></ul><h3 id="072c7afd-0bc5-4805-ae28-02b3ec30ed28" data-toc-id="072c7afd-0bc5-4805-ae28-02b3ec30ed28" collapsed="false" seolevelmigrated="true">Chapter 5: Consent Of Drug</h3><ul><li><p>Why animal research is valuable: there are questions you cannot ethically ask in humans (e.g., heroin exposure, brain manipulation) but are essential to understand mechanisms.</p></li><li><p>Animal models enable invasive procedures, brain recordings, and genetic manipulation that would be unethical in humans.</p></li><li><p>Ethical governance: pre-approval is required; justification that no other method could answer the question; use the minimum number of animals; animals are closely monitored and cared for under established guidelines.</p></li><li><p>Biosafety and oversight: university-level animal care committees (CCAC/ACC), provincial and federal regulatory bodies ensure daily welfare and compliance; researchers follow institutional care protocols.</p></li><li><p>Practical statistics: even with animal models, the number of animals used is kept to a minimum; animal research is a small fraction of overall animal use compared with agriculture.</p></li><li><p>Endpoint and replacement: researchers must consider alternatives (e.g., simulations, in silico models) and only use animals if necessary.</p></li><li><p>Translation to humans: animal data often translate well to human biology (e.g., gambling research) but still require cautious interpretation.</p></li><li><p>Human subjects ethics:</p><ul><li><p>Informed consent is essential; consent is an ongoing social contract, and participants can withdraw at any time.</p></li><li><p>Deception and confederates may be used but must be ethically justified and debriefed after the study.</p></li><li><p>Debriefing ensures participants understand the study's purpose and their role.</p></li></ul></li><li><p>Deception and consent: sometimes deception is used (e.g., with confederates) but must be ethical, non-harmful, and disclosed in debriefing.</p></li><li><p>Open discussions about consent: participants should be free to remove consent even after data collection if they feel uncomfortable.</p></li><li><p>The bottom line: ethical oversight and informed consent are non-negotiable; the welfare of participants and animals supersedes research gains.</p></li></ul><h3 id="993779d2-c8e8-4d7b-a4c9-28b7f7dd6a0d" data-toc-id="993779d2-c8e8-4d7b-a4c9-28b7f7dd6a0d" collapsed="false" seolevelmigrated="true">Chapter 6: The Independent Variable</h3><ul><li><p>Drug studies in humans are difficult and often ethically constrained; multiple-dose designs are often not feasible in humans unless the drug is already legally available and used by the public.</p></li><li><p>Screening concerns: pre-screening for health conditions (e.g., heart conditions) is necessary to ensure safety; screening procedures may need to balance thoroughness with ethical disclosure.</p></li><li><p>Ethics outrank data quality: safety and ethical considerations override the ideality of the data; researchers must adapt design to protect participants.</p></li><li><p>Practical takeaway: in many studies, you’ll design around the strongest feasible IV concept (e.g., dose, sleep amount, or hours studied) rather than a full idealized dose-response design.</p></li><li><p>The key message: be realistic about what can ethically be manipulated in humans; use animal models or existing clinicians’ medications when appropriate to answer questions that cannot be studied directly in humans.</p></li></ul><h3 id="4862598a-4157-4901-9832-04d5effeba63" data-toc-id="4862598a-4157-4901-9832-04d5effeba63" collapsed="false" seolevelmigrated="true">Chapter 7: Different Independent Variable</h3><ul><li><p>Designing exams and papers requires clear operationalization and justification for the chosen IV.</p></li><li><p>The exam stresses three scoring components: content (≈35%), critical thinking/synthesis (≈35%), and pizzazz/organization/creativity (≈30%).</p></li><li><p>Central idea: problems with operationalization are common; there is no single best way to operationalize a variable like “which drink is better.”</p></li><li><p>Coke vs Pepsi example to illustrate multiple valid operationalizations:</p><ul><li><p>Independent variable: beverage type (Coke vs Pepsi) or branding manipulation (Coke in a Pepsi can, etc.).</p></li><li><p>Dependent variable: how much participants indicate they like the drink (self-report scale 1–10), or how much they drink, or how much money they spend on each drink in a budget task.</p></li><li><p>Design choices:</p></li><li><p>Between-subjects design: each participant experiences only one beverage condition (Coke or Pepsi).</p></li><li><p>Within-subjects design: each participant experiences both beverage conditions (Coke and Pepsi) to compare directly.</p></li></ul></li><li><p>Additional independent variables: branding itself as a manipulation (e.g., can label color or branding visible) to separate taste from branding effects.</p></li><li><p>Practical reminder: the best studies synthesize multiple operationalizations (e.g., self-report rating, choice frequency, amount consumed, willingness to pay) to robustly capture the construct of “liking” or preference.</p></li><li><p>The core skill: articulate a narrative of the experimental design that a TA could follow, including the narrative flow, boundaries, and justification for choices.</p></li></ul><h3 id="9f9a6676-a4d8-4616-a90b-644b0e61f133" data-toc-id="9f9a6676-a4d8-4616-a90b-644b0e61f133" collapsed="false" seolevelmigrated="true">Chapter 8: Conclusion</h3><ul><li><p>Experimental design considerations for Coke vs Pepsi include controlling for non-ideal factors like time of day and prior meals; temperature can affect taste perception, so standardize serving temperature.</p></li><li><p>Palate cleansing between beverages helps reduce taste carryover, and ordering effects should be mitigated by randomizing presentation order (or counterbalancing).</p></li><li><p>Ordering effects and palatability: the first drink is often perceived as more enjoyable; to avoid bias, counterbalance the order of presentation and consider preliminary baselines.</p></li><li><p>Potential confounding variables to anticipate:</p><ul><li><p>Palate effects: first drink bias; mitigate with palate cleansers and counterbalancing.</p></li><li><p>Temperature effects: ensure consistent serving temperature as temperature alters flavor perception.</p></li><li><p>Prior taste preferences: preexisting brand opinions; mitigate via full counterbalancing including branding manipulations (e.g., using mismatched branding).</p></li></ul></li><li><p>Practical strategies for robust design:</p><ul><li><p>Randomize order of drinks across participants (or use a Latin square design).</p></li><li><p>Use a palate cleanser between tastings; provide water to rinse.</p></li><li><p>Consider multiple dependent measures (self-report ratings, choice frequency, and amount of beverage consumed) to capture the construct of “preference.”</p></li><li><p>Explicitly address the IV design (within-subjects vs between-subjects) and justify the choice in terms of statistical power and control of individual differences.</p></li></ul></li><li><p>The overarching goal: present a narrative that makes the experimental design transparent to the TA, showing a coherent flow from operationalization to measurement, control of confounds, ethical considerations, and interpretation.</p></li><li><p>Final takeaway for exam prep: when designing any experiment (not just taste tests), start with the IV (the dose or amount of the manipulation), define clear DVs, anticipate potential confounds, decide on blinding and control conditions, and articulate your narrative so others can follow your logic from question to answer.</p></li></ul><p><strong>Key formulas and concepts to remember</strong></p><ul><li><p>Operationalization: Concept → Measured variable (often multiple measurements needed).</p></li><li><p>Independent Variable (IV) vs. Dependent Variable (DV): IV changes across conditions; DV is the outcome measured.</p></li><li><p>Covariation concept in true experiments:<br>P(E|C=1)
      eq P(E|C=0) \text{or} \ \text{Cov}(C,E)
      eq 0.</p></li><li><p>Temporalprecedence:thecauseoccursbeforetheeffect.</p></li><li><p>Eliminationofalternatives:controlforconfoundingvariablestoruleoutthird−variableexplanations.</p></li><li><p>Correlationcoefficient:</p></li><li><p>Temporal precedence: the cause occurs before the effect.</p></li><li><p>Elimination of alternatives: control for confounding variables to rule out third-variable explanations.</p></li><li><p>Correlation coefficient:r \in [-1,1], with typical psychology values around \(0.3\) or \(-0.3\).* Strong correlations are closer to ±1.

    • Nonlinear relationships can occur (e.g., inverted-U): y = -a x^2 + b x + c, \text{with } a>0.

    • Blinding types: single-blind (participants unaware), double-blind (participants and researchers unaware), open-label (everyone knows).

    • Placebo vs vehicle: placebo is the no-drug condition in humans; vehicle is the inert solution used in animal studies.

    • Between-subjects vs within-subjects designs: design choices that affect randomization, statistical power, and control of individual differences.

    • Ethical principles: informed consent, debriefing, ongoing consent, minimum use of animals, justification for animal use, and adherence to institutional/federal guidelines.

    If you want, I can convert this into a condensed study guide with a quick-reference checklist for designing a psychology experiment.