Experimental Psychology: Scientific Method

Experimental Psychology and the Scientific Method

• Antecedent conditions are circumstances before an event or behavior.
• Applied research aims to solve real-world problems.
• Basic research is for testing theories or explaining psychological phenomena.
• Cause-and-effect relationship occurs when a set of antecedents always precedes a behavior, implying the antecedents cause the behavior.
• Commonsense psychology refers to everyday, nonscientific psychological data.
• Data includes facts and figures from research observations, and it is the plural of datum.
• Empirical data is observable, can be experienced, and can be verified or disproved.
• Experimentation demonstrates that particular behavioral events will occur reliably in specific conditions, and it is a tool of the scientific method.
• Good thinking involves organized, rational thought, open-mindedness, objectivity, and parsimony; also a tool of the scientific method.
• Laws are general scientific principles that explain the universe and predict events.
• Measurement is the systematic estimation of the quantity, size, or quality of an observable event, and a tool of the scientific method.
• Methodology refers to the scientific techniques for collecting and evaluating psychological data.
• Observation is the systematic noting and recording of events, which is another tool of the scientific method.
• Parsimony refers to preferring the simplest explanation until conflicting evidence arises; also known as Occam's razor.
• Pseudoscience appears scientific but lacks a true scientific basis, and hasn't been confirmed using the scientific method.
• A psychology experiment applies at least two treatment conditions to subjects and compares behaviors to test a hypothesis.
• Replication verifies research outcomes by repeating procedures; also a principal tool of the scientific method.
• Science is the systematic data gathering to describe events under specific conditions, enabling explanation, prediction, and control.
• The scientific method involves steps to gather and verify information, answer questions, explain relationships, and communicate findings.
• A subject is an individual who participates in research.
• Testable refers to the capability of being tested, usually relating to a hypothesis.
• A theory is a set of general principles to explain and predict behavior or other phenomena.
• Treatment is a specific set of antecedent conditions created by the experimenter to test its effect on behavior.

Research Ethics

• Animal rights states that all sensate species feeling pain have equal value and rights.
• Animal welfare describes the humane care and treatment of animals.
• "At risk" indicates the likelihood of a subject being harmed by the nature of the research.
• Debriefing is the full disclosure at the end of an experiment, explaining the study's nature and purpose.
• Fraud is falsifying or fabricating data, and plagiarism is a form of fraud.
• Informed consent is a subject's voluntary agreement to participate in research after being informed of the study's nature and purpose.
• The Institutional Animal Care and Use Committee (IACUC) reviews research proposals to protect animal subjects.
• The Institutional Review Board (IRB) reviews research proposals, ensuring the safety and rights of human participants.
• Minimal risk describes a condition where subjects' odds of being harmed are not increased.
• Plagiarism is representing others' ideas as one's own, which is a breach of ethics that can lead to legal action.
• Risk/benefit analysis is an IRB determination of whether individual risks are outweighed by potential benefits and the importance of gained knowledge.

Alternatives to Experimentation: Non-experimental Designs

• An archival study reexamines existing records for a new purpose through a descriptive method.
• A case study is a descriptive record of an individual's experiences or behaviors by an outside observer.
• Deviant case analysis compares deviant individuals with non-deviant ones to find significant differences in a type of case study.
• Empirical phenomenology relies on researcher experiences and participant data in a qualitative approach.
• External validity measures how well experiment findings generalize to other people and settings.
• A field study is a nonexperimental method in real-life settings, employing naturalistic observation and unobtrusive survey measures.
• A focus group is a type of group interview led by a trained facilitator.
• Internal validity measures the certainty that observed behavioral changes are caused by experimental treatments.
• Naturalistic observation is a descriptive, nonexperimental method of observing behaviors as they naturally occur.
• A paradigm is the set of attitudes, values, beliefs, methods, and procedures in a discipline at a certain time.
• A participant-observer study is a field observation where the researcher joins the group being studied.
• Phenomenology gathers data non-experimentally by attending to and describing one's immediate experience.
• Qualitative research focuses on words rather than numbers, emphasizing self-reports and expressions.
• Reactivity is the tendency of subjects to alter behavior when aware of being observed.
• Retrospective data is collected in the present based on recollections of past events.
• Systematic observation uses specific guidelines to record observations and ensure objectivity.
• An unobtrusive measure assesses subjects' behaviors without their knowledge, improving objectivity.

Alternatives to Experimentation: Surveys and Interviews

• Cluster sampling is probability sampling where researchers sample entire clusters or natural groups in a population.
• Content analysis quantifies open-ended question responses by categorizing them with objective guidelines.
• Context effects are produced by a question's position, influencing its interpretation based on the question order.
• Continuous dimension views traits, attitudes, and preferences on a spectrum, with individuals falling along it.
• Convenience sampling obtains a sample using convenient groups, considered weak due to the lack of control over representativeness.
• Interval scale is a measurement scale with equal intervals but no true zero point.
• Latent content is the hidden meaning behind a question.
• Level of measurement is the type of scale used to measure a variable.
• Manifest content is the plain meaning of the words or questions on the page.
• Nay-sayers tend to disagree with a question regardless of its manifest content.
• Nominal scale classifies items into distinct categories based on a common feature, and it is the simplest measurement level.
• Nonprobability sampling doesn't choose subjects at random.
• Ordinal scale measures magnitude with values presented as ranks.
• Population includes all people, animals, or objects sharing at least one characteristic.
• Position preference is the tendency to select a certain position when in doubt.
• Probability sampling allows the odds of any subject being chosen to be known or calculated.
• Purposive sampling selects nonrandom samples that reflect a specific study purpose.
• Quota sampling selects samples through quotas reflecting the population's makeup, but without random selection.
• A random number table ensures unbiased selection in sampling.
• Random selection ensures unpredictability by giving each population member an equal chance of being chosen.
• Ratio scale has equal intervals and a true zero point.
• Reliability is the extent to which a survey is consistent and repeatable.
• Representativeness describe the degree to which a sample reflects the overall population.
• Response set is the tendency to answer questions based on their latent content to create a certain impression.
• Response style reflects the tendency for subjects to respond in a specific way, regardless of content.
• The sample of subjects is a selected subset of the population of interest.
• Sampling is the process of deciding who the subjects will be and selecting them.
• Simple random sampling is the most basic probability sampling, selecting a portion of the population without bias.
• Snowball sampling is a nonprobability method where an initial participant refers the researcher to others meeting study criteria.
• Stratified random sampling is a probability sampling method where subgroups are sampled proportionally to their representation in the population.
• Survey research obtains data about opinions, attitudes, preferences, and experiences using questionnaires or interviews.
• Systematic random sampling is a variation of random sampling, selecting every nth person from the population.
• Validity describes the extent to which a survey measures what it is intended to measure.
• Willingness to answer refers to the differences in how respondents approach uncertain questions.
• Yea-sayers agree with questions regardless of their manifest content.

Alternatives to Experimentation: Quasi and Experimental Designs

• Causal modeling creates and tests models that may suggest cause-and-effect relationships among behaviors.
• The coefficient of determination (r²) estimates the amount of variability in scores on one variable explained by another in a correlational study.
• Correlation is the degree of relationship between two traits, behaviors, or events, represented by r.
• A correlational study determines the correlation between two traits, behaviors, or events.
• Cross-lagged panel design measures the same behaviors at two time points, computes correlations, and uses the pattern to infer causal direction.
• Cross-sectional study measures different subject groups at different stages at a single point in time to examine time-related changes.
• An ex post facto study examines the effects of pre-existing subject characteristics by comparing groups based on naturally occurring differences.
• Linear regression analysis is a correlation-based method estimating a score on one behavior from another when they are strongly related.
• Longitudinal design follows and measures the same subject group at different times to track changes.
• Multiple correlation (R) assesses statistical intercorrelations among three or more behaviors.
• Multiple regression analysis uses a regression equation to predict a behavior score from scores on other related behaviors.
• Negative correlation occurs when an increase in one variable is associated with a decrease in another; also called an inverse relationship.
• Nonequivalent groups design compares the effects of different treatment conditions using preexisting groups of participants.
• Partial correlation analyzes a measured variable's statistical influence held constant while computing the correlation between two other variables.
• Path analysis measures subjects on several related behaviors, creating and testing models of possible causal sequences using advanced statistics.
• Positive correlation occurs when an increase in one variable is associated with an increase in the other; also known as direct relationship.
• Pretest/posttest design assesses behavior alteration by an event, comparing scores from measurements before and after the event.
• Quasi-experimental designs resemble true experiments but lack essential elements, like manipulation of antecedents or random assignment.
• Regression line is the line of best fit, representing the equation describing the relationship between two variables in a correlational study.
• A scatterplot graphs pairs of scores from each subject in a correlational study with one variable on the X-axis and the other on the Y-axis.
• Simple correlations describe relationships between pairs of scores from each subject.
• Subject variable is a characteristic of subjects in an experiment or quasi-experiment that determines how subjects are categorized.

Formulating the Hypothesis

• An analytic statement is always true.
• A contradictory statement is always false.
• The deductive model reasons from general principles to specific instances, useful when testing theories.
• An experimental hypothesis tentatively explains an event or behavior, predicting the effects of specific conditions.
• A falsifiable statement can be disproven.
• A fruitful statement leads to new studies.
• A hypothesis predicts the relationship between at least two variables.
• The inductive model reasons from specific cases to general principles to form a hypothesis.
• A non-experimental hypothesis predicts relationships among variables without implying causation.
• A parsimonious statement is a simple explanation requiring few assumptions.
• A synthetic statement can be true or false, necessary when forming an experimental hypothesis.
• A testable statement can be tested using experimental manipulation and measurement.

METHODS: The Basics of Experimentation

• Concurrent validity is how well a measurement correlates with an established measure of the same variable.
• Construct validity is how well an operational definition represents the intended construct.
• Content validity is how well a measure reflects the content it is supposed to assess.
• Face validity is how well a measure appears to assess what it's supposed to measure.
• Internal validity is the degree to which the independent variable causes the changes in the dependent variable.
• Predictive validity is how well a measure can predict future performance or behavior.
• Reliability is the consistency and dependability of a measurement or experiment.
• Test-retest reliability is the consistency of test results when the same test is administered at different times.
• Confounding occurs when an extraneous variable changes systematically with the independent variable, providing an alternative explanation.
• History threat is an outside event that affects the dependent variable.
• Instrumentation threat occurs when changes in the measurement instrument affect results.
• Maturation threat occurs when internal changes in participants affect results.
• Selection threat occurs when nonrandom assignment results in preexisting group differences.
• Selection interactions occur when selection bias interacts with other threats to internal validity.
• Statistical regression threat occurs when extreme scores move toward the mean upon retesting, unrelated to experimental treatment.
• Subject mortality threat occurs when dropout rates differ between experimental conditions.
• Testing threat occurs when prior exposure to a test influences later results.
• Test-retest reliability measures the consistency between an individual's scores on the same test taken at different times.
• Validity indicates the soundness of an operational definition when studying the intended variables.

Solving Problems: Controlling Extraneous Variables

• Balancing distributes the effects of extraneous variables equally across treatment conditions.
• Constancy of conditions keeps all aspects of the experiment the same except for the independent variable.
• A context variable is an extraneous variable stemming from the research setting.
• A cover story prevents participants from guessing the true hypothesis by giving a false explanation.
• Demand characteristics are aspects of the experiment that influence participants to behave in a certain way.
• A double-blind experiment reduces bias by ensuring neither the experimenter nor the subjects know the treatment condition.
• Elimination controls extraneous variables by removing them from an experiment.
• Experiment bias occurs when the experimenter unintentionally influences the results.
• Personality variables are the personal characteristics that an experimenter or volunteer brings to the experimental setting.
• Physical variables are aspects of testing conditions that need control.
• A placebo effect occurs when participants experience changes simply because they believe they are receiving treatment.
• The Rosenthal effect (Pygmalion effect) happens when experimenters treat subjects differently based on their expectations.
• A single-blind experiment reduces bias because subjects do not know their condition.
• Social variables are the subject and experimenter relationship qualities that influence results.

Basic-Between Subject Designs

• Between-subjects design involves different subjects taking part in each experiment condition.
• Block randomization creates treatment blocks with one random order of conditions, and subjects are then assigned to fill these blocks.
• A control condition involves subjects receiving a zero value of the independent variable.
• A control group are the subjects in a control condition.
• Effect size is a statistical estimate of the treatment effects' magnitude.
• An experimental condition applies a value of an independent variable to subjects and measures the dependent variable.
• Experimental design refers to the general structure of an experiment.
• An experimental group comprises the subjects in an experimental condition.
• Multiple-groups design is a between-subjects design with one independent variable and more than two treatment conditions.
• Multiple-independent-groups design is the most commonly used multiple-groups design with subjects randomly assigned to different conditions.
• A pilot study pretests selected independent variable levels using a few subjects before conducting the actual experiment.
• A placebo group is a control condition that treats subjects exactly as those in the experimental group, except the prototype of a good control group.
• Precision matching creates pairs of subjects with identical scores on the matching variable.
• Random assignment assigns subjects to treatments, giving each an equal chance.
• Range matching creates pairs of subjects whose matching variable scores fall within a pre-specified range.
• Rank-ordered matching creates matched pairs by placing subjects in order by their matching variable scores; subjects with adjacent scores become pairs.
• A two-experimental-groups design exposes two subject groups to different independent variable levels.
• A two-group design is the simplest experimental design and is used when only two treatment conditions are needed.
• A two-independent-groups design is an experimental design where subjects are placed in two conditions through random assignment.
• A two-matched-groups design is an experimental design with two treatment conditions and with subjects matched on a subject variable.