Experimental Psychology: Scientific Methodology - PSYCH204 PDF

PSYCH204: EXPERIMENTAL PSYCHOLOGY EXPERIMENTAL PSYCHOLOGY AND THE SCIENTIFIC 5. Self- correction – modern scientists METHOD accept the uncertainty of their own con...

PSYCH204: EXPERIMENTAL PSYCHOLOGY EXPERIMENTAL PSYCHOLOGY AND THE SCIENTIFIC 5. Self- correction – modern scientists METHOD accept the uncertainty of their own conclusions THE NEED FOR SCIENTIFIC METHODOLOGY Falsification - to challenge an existing The word science comes from the Latin explanation or theory by testing a Word scientia, which means knowledge. hypothesis that follows logically from it and demonstrating that this hypothesis is false. Methodology: the scientific technique used to 6. Publicizing results – scientists meet collect and evaluate psychological data. frequently through professional and Commonsense psychology: everyday special interest groups and attend nonscientific collection of psychological data professional conferences to exchange used to understand the social world and guide information about their current work. our behavior. 7. Replication – the process of repeating Scientific Method: steps scientists take to gather research procedures to verify that the and verify information, answer questions, explain outcome will be the same as before; a relationships and communicate feelings. principal tool of the scientific method. THE CHARACTERISTICS OF MODERN SCIENCE THE TOOLS OF PSYCHOLOGICAL SCIENCE 1. Scientific mentality – the psychologist’s 1. Observation – they systematic noting and goal of prediction rests the assumption recording of events; a principal tool of the that behavior must follow a natural order; scientific method therefore it can be predicted. 2. Measurement – the scientific estimation of 2. Gathering empirical data – data that are the quantity, size, or quality of an observable or experienced. observable event; a principal tool of the 3. Seeking general principles – modern scientific method. scientists go beyond cataloging 3. Experimentation – the process undertaken observations to proposing general to discover something new or to principles—laws or theories—that will demonstrate that events that have already explain them. occurred will occur again under a specified Laws – general scientific principles that set of conditions; a principal tool of the explain our universe and predict events. scientific method. 4. Good thinking – organized and rational Experimentation is not always possible. To do an thought, characterized by open- experiment, our predictions must be testable. mindedness, objectivity, and parsimony; a Testable – capable of being tested; principal tool of the scientific method. typically used in reference to a hypothesis. Parsimony – an aspect of good thinking, Two requirements must be met in order to stating that the simplest explanation is have a testable hypothesis: procedures for preferred until ruled out by conflicting manipulating the setting must exist, and evidence; also known as Occam’s razor. the predicted outcome must be observable.  PSYCH204: EXPERIMENTAL PSYCHOLOGY the set SCIENTIFIC EXPLANATION TO PSYCHOLOGICAL is inferred to cause the behavior. SCIENCE o The type of cause and effect Identifying antecedent conditions – in a relationship we establish through scientific context, explanation means experiments is called a temporal specifying the antecedent conditions of an relationship. We look at the even or behavior. differences in behavior after o Antecedent conditions – all subjects are exposed to the circumstances that occur or exist treatment—not before. before the event or behavior to be Necessary versus sufficient conditions – explained; also called antecedents. we need to distinguish between necessary Comparing treatment conditions – in and sufficient conditions. psychology, it would be virtually impossible o E.g. Cutting down on fat intake may to identify all the antecedents that affect be a sufficient condition to produce the behavior of research participants (also weight loss. But is it a necessary called subjects) at a particular time. condition? No. We also could lose o Subject – the scientific term for an weight by increasing our activity individual who participates in level, so reducing fat intake is not a research. necessary condition. Thus, we create specific sets of antecedent o A car will not run without fuel. conditions that we call treatments. Therefore fuel is a necessary o Treatment – a specific set of condition for running a car. antecedent conditions created by the experimenter and presented to FORMULATING THE HYPOTHESIS subjects to test its effect on behavior. Hypothesis: The thesis, the main idea, of an The psychology experiment – a controlled experiment procedure in which at least two different : A predictive relationship between at least two treatment conditions are applied to variables. subjects whose behaviors are then : Different research designs have different measured an compared to test a hypothesis statements hypothesis about the effects of the treatments on behavior. NON-EXPERIMENTAL VS EXPERIMENTAL HYPOTHESIS Establishing cause and effect – the Some nonexperimental designs do not greatest value of the psychology typically include a hypothesis (i.e.: experiment is that, within the experiment, phenomenology, case study) we can infer a cause and effect Nonexperimental hypothesis predicts how relationship between the antecedent events, traits, or behaviors might be conditions and the subjects’ behaviors. related, but not about cause effect. o Cause and effect relationship – the Experimental hypothesis predicts cause relationship between a particular and effect relationship behavior and a set of antecedents that always precedes it—whereas EXAMPLES OF NONEXPERIMENTAL HYPOTHESES other antecedents do not—so that  PSYCH204: EXPERIMENTAL PSYCHOLOGY evidence that the hypothesis is indeed false. “If you read this book carefully enough, then you will be able to design a good experiment.” PARSIMONIOUS STATEMENTS A simple hypothesis is preferred over one that requires many supporting assumptions. FRUITFUL STATEMENTS It leads to new studies CHARACTERISTICS OF AN EXPERIMENTAL HYPOTHESIS HOW ARE HYPOTHESIS FORMED? Every experiment has at least one hypothesis The Inductive Model A tentative explanation of an event or Reasoning from specific cases to more behavior; one that seems plausible general principles Out of a number of possible causes, the list Examining individual instances, and must be narrowed down constructing an overall explanatory Hypothesis must be synthetic, testable, scheme falsifiable, parsimonious, and fruitful The Deductive Model Reasoning from general principles to make SYNTHETIC STATEMENTS predictions about specific instances Can either be true or false Provides a test of the value of a theory Avoid o Analytic statements – always true COMBINING INDUCTION AND DEDUCTION o Contradicting statements – always In practice, both approaches are not so false neatly separated Can be stated in “If…then” form Expresses potential relationship BUILDING ON PRIOR RESEARCH TESTABLE STATEMENTS The most useful way of finding hypotheses The means for manipulating antecedent is by working from research that has conditions and measuring the resulting already been done. behavior must exist Nonexperimental designs may suggest cause and effect explanations FALSIFIABLE STATEMENTS Prior research is useful in focusing your Research hypothesis must be disprovable thinking on important issues; what by the research finding. researches might have missed, or possible Worded in a way that failure to find the new applications predicted effects must be considered  PSYCH204: EXPERIMENTAL PSYCHOLOGY SERENDIPITY AND THE WINDFALL EXPERIMENT INDEPENDENT AND DEPENDENT VARIABLES A discovery may be made where none was Independent Variable (IV) - the dimension that is intended – serendipity intentionally manipulated (the antecedent) One must be open to possibilities Aspects of the environment Not just a matter of luck; it is also a matter Aspects of a given task of knowing enough to use an opportunity. Different psychological states INTUITION Effectiveness of psychological Knowing without reasoning interventions The more we know about a topic, the better Must be given at least two possible values out intuitive hypotheses are likely to be – Levels of IV – 2 treatment conditions WHEN ALL ELSE FAILS Used in quasi-experiment too Pick a psychology journal and read, In a true experiment subjects must no be something might interest you different on a preexisting characteristic Check out nonexperimental studies Random assignment to groups Observe in a public place Confounding – difference in both subject Turn your attention to a real-world variables and manipulated IV problem Dependent Variable (DV) – the behavior we SEARCHING THE RESEARCH LITERATURE expect to change Important part of conducting research Assumed dependent on the value of the IV. Work done to test your hypothesis or one that is closely related. SAMPLE HYPOTHESIS Can help to develop procedures If an individual is anxious, then he will not Tips for measuring your observations perform well in an unfamiliar task. A journal article on your topic can provide other resources in the reference section WHY DO WE NEED OPERATIONAL DEFINITIONS? Books Many concepts have more than one Overview of a topic area meaning Metaanalysis Meanings are sometime ambiguous Skip popular books and other source from Hypothetical constructs can not be directly pop media observed; are manifested differently. o When is a person anxious? THE BASICS OF EXPERIMENTATION Non-construct variables have many Psychology Experiment possible dimensions. Manipulation of antecedent condition o How bright is light? Preparing at least 2 treatment conditions o How dark is dark? Measure their effect on behavior o How much crying is crying? Compare behaviors using statistics Causal inferences Experiments should be carefully designed and carefully controlled  PSYCH204: EXPERIMENTAL PSYCHOLOGY OPERATIONAL DEFINITIONS Consider levels of measurement Many concepts have more than one o Nominal meaning o Ordinal Conceptual definition – how a word is used o Interval in everyday language o Ratio Operational definition – precise meaning o Interval gives more information than of a variable within an experiment. nominal o Observable operations EVALUATING OPERATIONAL DEFINITIONS o Procedures Reliability o Measurements Validity Other researchers will understand exactly what was done Reliability Consistency and dependability IV: EXPERIMENTAL OPERATIONAL DEFINITION o Interrater reliability – two observers What was done exactly to create the take measurements and agree various treatment conditions o Test-Retest reliability – taking the o Color refers to the variations of red test again after a reasonable or blue that were used as interval background for the presentation of o Interitem reliability – different parts the illustrations during the of the questionnaire attain perception task. consistent results Validity DV: MEASURED OPERATIONAL DEFINIITION Actually studying the variable intended to Description of what was done to be studied measure the variable Manipulation check – providing evidence Exact descriptions of the specific behaviors for the validity of an experimental or responses, and how those responses are procedure scored o Face validity o In this study, intelligence refers to o Content validity – fair sample of the participants’ scores obtained from quality we intend to measure the Purdue Non-Language Test. o Predictive validity – procedures o In this study accuracy refers to the yield information that enables us to frequency of correct canceled out predict future behavior letters in a series of random o Concurrent validity – scores on the characters measuring device correlate with scores obtained from obtained from DEFINING SCALES OF MEASUREMENT another method Precise quantitative measurements o Construct validity – does the test or o Illumination – Luxes, Foot candles procedure actually measures the o Distance – Meters, Feet construct studied o Sound – Decibels  PSYCH204: EXPERIMENTAL PSYCHOLOGY EVALUATING THE EXPERIMENT: INTERNAL VALIDITY 5. Statistical Regression – Extreme score tend Internal Validity – the degree to which a to move closer to the mean when the test researcher is able to state a casual relationship is given again between antecedent conditions and the 6. Selection – without random assignment, subsequent behavior two group may have different preexisting If other explanations are possible, the characteristics experiment is not internally valid 7. Subject Mortality – more subjects dropout in one experimental condition than the EXTRANEOUS VARIABLES other; the ones left might be unique Other things (besides the IV) that may be 8. Selection Interaction – selection can changing throughout the experiment that interact with history, maturation, or can potentially alter the values of the DV mortality o Time of day A researcher should control as many o Fatigue variables as possible o Subject characteristics o Instructions RESEARCH ETHICS Should be controlled CHARACTERISTICS OF SCIENCE TO CONSIDER Science is commonly thought of as amoral CONFOUNDING Science per se does not include values Value of an extraneous variable changes However, researchers bring their own systematically across different conditions values, ethics, morals, and sense of right of the experiment. and wrong to the work they do. Experimental results cannot be interpreted RESEARCH ETHICS with certainty Treating participants ethically and responsibly is of foremost concern for a CLASSICAL THREATS TO INTERNAL VALIDITY researcher Eight kinds of extraneous variables Responsible research is aimed at Inherent to quasi-experimental designs advancing our understanding of human Must always be considered behavior in ways that will benefit humanity, not merely to satisfy idle curiosity. 1. History – outside events that occurred Research that harms participants is before the experiment undesirable even if it may increase 2. Maturation – any internal changes in the wisdom. subjects: fatigue, boredom, development A researcher is legally responsible for what 3. Testing – individuals frequently perform (in ever happens to the participants. a test) differently the second time; Practice effect 4. Instrumentation – features of the measuring instrument changes; errors in recording, apparatus breakdown, different provided space for response  PSYCH204: EXPERIMENTAL PSYCHOLOGY INSTITUTIONAL REVIEW BOARD (IRB) THE AMERICAN PSYCHOLOGICAL ASSOCIATION In the U.S. law requires institutions (APA) GUIDELINES engaging in research to set up a review Applies to psychologists and students who committee to evaluate proposed studies assume the role of psychologists by before they are conducted – Institutional engaging in research Review Board (IRB) They are responsible for maintaining ethical standards in all research Putting Participants at Risk conducted by them or others they Its first task is to determine whether the supervise. research puts participants at risk. ON INFORMED CONSENT A participant at risk is one who is, more Informed consent is required from all likely to be harmed in some way by participants who are at risk because of the participating in the research. nature of the research. Risk/benefit analysis – determining In some cases, participants are only at whether any risk to the individual are minimal risk – defined as risk that is no outweighed by potential benefits or greater in probability and severity that that knowledge to be gained. ordinarily encountered in daily life. Informed Consent i.e.: observations of public behaviors, The participants agree to take part after anonymous questionnaire, some forms of having been fully informed about the archival research nature of the study. In such cases, informed consent is not 1. Consent must be given freely, without use always mandatory of force or coercion. DECEPTION AND FULL DISCLOSURE 2. They must be free to drop out of the Sometimes, the true purpose of the study is experiment at any time. disguised 3. Researchers must give a full explanation of The consensus among researchers seems the procedures and answer any question a to be that the use deception is most often 4. Researcher must make clear the potential justified by the knowledge that is gained risk and benefits of the experiment. Many important psychological problems 5. Researchers must provide assurances that cannot be studied without the use of data will remain private and confidential. deception. 6. Subjects may not be asked to release the The deception must not influence a researchers from liability or to waive their subject’s decision to take part in the legal rights in the case of negligence research Consent should be obtained in writing Full disclosure of the true nature and For minors and cognitively impaired, purpose of the research should be consent from parent should be obtain revealed after - debriefing They themselves can refuse to participate Although consent need to be written, researchers need to verbally reinforce information that is important for participants  PSYCH204: EXPERIMENTAL PSYCHOLOGY APA STANDARD 8.07: DECEPTION IN RESEARCH FRAUD AND PLAGIARISM a. Psychologists do not conduct a study involving deception unless they have determined FRAUD that the use of deceptive techniques is justified by We should report our procedures and the study’s significant prospective scientific. findings honestly and accurately. Educational, or applied value and that effective Fraud – Data falsification nondeceptive alternative procedures are not Research article for publication are feasible. reviewed by the editor and by several b. Psychologists do not deceive prospective experts in the field – peer review participants about research that is reasonably Reviewers assess the merit of a research expected to cause physical pain or severe and suggests improvements emotional distress. Replication – researchers attempt to c. Psychologists explain any deception that is replicate the published findings of others, an integral feature of the design and conduct of especially the surprising, novel and an experiment to participants as early as is important feasible, preferably at the conclusion of their The competitive nature of academic participation, but no later than the conclusion of psychology works against fraud the data collection, and permit participants to withdraw their data. PLAGIARISM To plagiarize means to represent someone Whether the effects of deception can ever else’ ideas, words, or written work as your be fully reversed by debriefing remains a own. serious question. It is serious and can result to legal action Regardless of any later explanation, the Paraphrasing without giving credit is participant’s anxiety and discomfort representing someone else’s ideas s your during the experiment were real. own; it is also plagiarism. Even if it is Sometimes, extensive post experimental common knowledge. debriefings that detail the unreality of what Unfortunately, it is easy to plagiarize, even has gone through are necessary. without being aware of it. STANLEY MILGRAM EXPERIMENT TIPS TO AVOID PLAGIARISM 1. Take complete notes, w/c include citation ANONYMITY AND CONFIDENTIALITY of the source: author’s name, title of the To protect the privacy of participants article, journal name, volume number, Data should be collected anonymously or issue number, the title of the book, the identified by code of fictitious names. names of editors, and the publisher’s Use of aggregated or group data name, city and state. Data must be placed in a secure place, 2. Within your own report, identify the source kept confidential, and used only for the of any ideas, words, or information that are purposed explained to the participant not your own. Data should not be a subject of gossip  PSYCH204: EXPERIMENTAL PSYCHOLOGY 3. Identify any direct quotes by quotation marks at the beginning and end of the quotes and indicate where you got the quotes. Include the page number. 4. Be careful with paraphrasing. There is a great temptation to lift the whole phrases or catchy words from another source. Use your own words instead, or use quotes. Again, be sure to give credit to your source. 5. Include a complete list of references at the end of the report. In psychology, references are written in APA style. 6. If in doubt about whether citation is necessary, cite the source anyway. You will do no harm by being especially cautious.  PSYCH204: EXPERIMENTAL PSYCHOLOGY BASIC BETWEEN-SUBJECTS DESIGN How many subjects are needed for between-subjects design? Experimental Design o at least 15 to 20 subject in each General structure of the experiment treatment condition o Number of treatment conditions o More comfortably, 30 per treatment o Same or different subjects in different treatment conditions BETWEEN-SUBJECTS DESIGNS (1 IV, 2 Groups) Determined by How were the subject assigned? o The nature of the hypothesis o Two Independent Groups o Information the researcher is o Matched Groups seeking What are the levels of IV? o Practical problems in running the o Experimental group – Control group experiment Design Questions to consider o Two Experimental group Design o How many independent variables 2 INDEPENDENT GROUPS are there? Grouped using random assignment o How many treatment conditions are What are the possible treatment needed? conditions (level of the IV) for 2 o Will we use same or different independent groups? subjects? o Experimental group-Control group; Experimental condition-Control Between-Subjects Design condition; Treatment-No treatment; Different subjects take part in each however, “no treatment” does not really condition of the experiment mean “no treatment” all the time. Comparison between the behaviors of o Two Experimental group (No Control different groups of subjects group) – gathers more precise Two Groups information; different values of the IV. o Two Independent Groups When can 2 independent groups be used? ▪ Experimental group-Control o If there is only one independent variable group o If randomization can be assumed ▪ 2 Experimental group successful o Two Matched Groups 2 MATCHED GROUPS ▪ Experimental group-Control Does/must/can not use random group assignment ▪ 2 Experimental group 2 groups of subjects assigned by the Multiple Groups researcher by equating them on a particular characteristic that might affect Selecting and recruiting subjects the DV The more the sample resembles the whole Matching can be done before or after the population, the more likely it is that the experiment behaviors of the sample mirrors that of the How to match? population. o Precision Matching – matched pairs have identical scores  PSYCH204: EXPERIMENTAL PSYCHOLOGY o Range Matching – matched pair fall in As a general rule, select the simplest a specified range design that will make an adequate test of Note: Some subjects have to be discarded your hypothesis. o Rank-order Matching – subject scores are ranked; adjacent scores are paired. PRACTICAL LIMITS When can 2 matched groups be used? More subject are needed o Presence of a strong extraneous Takes more time variable More complicated statistical procedures o Need for more efficient measures of the Use of Pilot study (mini experiment) pretest IV’s effect selected levels of an IV before conducting o When there is a very small number of the actual experiment. subjects BETWEEN-SUBJECTS FACTORIAL DESIGN BETWEEN-SUBJECTS DESIGNS (1 IV, MULTIPLE MORE THAN ONE INDEPENDENT VARIABLE GROUPS) In real life, variable rarely occur alone. Sometimes, it takes more than two Experiment with more than one treatment conditions to make a good test independent variables are efficient and of a hypothesis provide more information than When amount or degree of the IV is experiments with one independent important variable. o Different values of the same variable Factorial designs – designs with two or can produce different effects more independent variables o May give complete understanding Independent variables are called factors of how the variable works Two factor experiment – the simplest Each group is run though a different factorial design condition One treatment condition may be the FACTORIAL DESIGNS control They give us information about the effects Usual multiple independent groups of each independent variable in the Matching is possible experiment – main effects They enable us to answer the question: ASSIGNING SUBJECTS How does the influence of one Table of random number independent variable affect the influence Use of block randomization of the other variable. THE MAIN EFFECTS CHOOSING TREATMENTS The action of a single independent variable How does one choose how many in an experiment. treatments to use? A change in behavior associated with a o Theoretical basis change in the value of a single o Equal or proportional intervals independent variable or factor in the What will I gain by adding these extra experiment. conditions to the experiment?  PSYCH204: EXPERIMENTAL PSYCHOLOGY There are as many main effects as there are factors LOOKING FOR INTERACTIONS Factorial design allows us to test for the relationships between the effects of different independent variables An interaction is present if the effect of one independent variable changes across the levels of another independent variable. Alcohol alone or sleeping pill alone can reduce anxiety, together, they can be fatal DESCRIBING THE DESIGN The effects of one factor will change 2 x 2 (two by two) factorial design depending of the level of the other Other methods of describing design An interaction can tell us that there are variables: limits to the effect of one or more factors 1. Factor-Labeling Method Two independent variable = one 2 x 2 (Type of Name x Length of interaction Name) between-subjects factorial More than two independent variables design result to more complex interactions – 2 (name type) x 2 (name length) higher-order interactions between-subjects factorial design It is possible to have interactions but no 2. Factor and Levels Method main effects 2 x 2 (Type of Name: given, nickname Or significant effect with no interaction x Length of Name: short, long) between-subjects factorial design LAYING OUT A FACTORIAL DESIGN 2 (given name or nick name) x 2 Step 01: Indicate the 2 independent variables (short or long name) between- Factor 1 (type of name) subjects factorial design Factor 2 (length of name) 2 x 3 x 2 factorial design Step 02: Indicate the levels of Factor 1 o There are 3 numbers which means Factor 1 (Type of name) there are 3 factors or independent o Given name variables o Nickname o Factor 1 has two levels Factor 2: (Length of name) o Factor 2 has three levels Step 03: Indicate the levels of Factor 2 o Factor 3 has two levels Factor 1: (Type of name) o There are 12 separate conditions o Given name Measuring interactions in one key reason o Nickname for doing factorial research. Factor 2: (length of name) By looking at two or more variables o Short together, we can assess whether the effect o Long of one variable depends on the level of another variable  PSYCH204: EXPERIMENTAL PSYCHOLOGY CHOOSING A BETWEEN SUBJECTS DESIGN Practical reasons for keeping factorial designs simple: o More treatment condition means more subjects o More treatment condition means more time to run the experiment o More treatment condition means more time to do the statistical analysis o Complicated design are virtually uninterpretable o Four way interactions are practically impossible to conceptualize and explain o 2 x 2 factorial design has 3 possible effects