PSYB01 Psychological Research Lab - Lecture 5.docx

PSYB01 Psychological Research Lab Lecture 5 Research Methods – Studying Behaviour (Con’t) Research Example ‘Poverty causes mental illness’ What would a researcher have to show to establish this causal relationship, according to Mills? -Temporal precedence (Poverty must come first before mental illness occurs) -Correlation (how poverty interacts with other factors) -Factors from poverty that causes mental illness (Stress, cognitive functions, etc) Nonexperimental versus Experimental Methods Nonexperimental Method: Look at things as they are, goes out and sample the population, determines whether the variables relate to each other (correlation) -Direction of Cause and Effect -The Third-Variable Problem (confounding variable) Quasi-experimental: Used to measure real life situations and sample a targeted population/group (similar, socio-economic status, demographic, etc) Experimental Method: Manipulates variables to collect data and observe from it -Experimental Control -Randomization -Artificiality E.g., Exercise helps lower cholesterol level -Nonexperimental method ->Goes out to sample a targeted group >Give out survey to see how much people exercise, what are their cholesterol level, etc ->Inability to infer causality, reduce validity -Experimental method >Assign people to exercise condition (experimental group) and non-exercise condition (control group) Qualitative analysis: -Ethnography -Philosophy, sociology -Case studies (social work, sociology, etc) -Contains experimental methods -Correlation survey, quasi-experimental -Grounded theory (Interview people and create a theoretical framework based on data collected) Non-experimental Research Description: -Relationships studied by making observations or measuring variables as they exist naturally Examples: -Behaviour observed as it naturally occurs -Asking people to describe behaviour -Directly observing behaviour -Recording physiological responses -Examining public records Advantages: -Allows measure of Covariation between variables -IV can be observed in a natural context -Allows us to study participant variables that cannot be manipulated Disadvantages: -Difficult to infer cause and effect -Direction and third variable problem -Difficult to control many aspects of the situation Experimental Research Description: -Direct manipulation and control of variables, then response or result is observed Examples: -Measure behaviour, then introduce a manipulation, and measure an outcome -Random assignment of participants -Experimental group experiences manipulation, but control group does not; outcome variable is measured Advantages: -Reduces ambiguity in interpretation of results regarding cause and effect -Attempts to eliminate the impact of all possible confounding third variables -Permits greater experimental control -Reduces the possible influence of extraneous variables through randomization Disadvantages: -High control may create an artificial atmosphere -Can be unethical or impractical Nonexperimental versus Experimental Methods The causal possibilities in a non-experimental study Depression causes alcoholism Depression -> Alcoholism -Self medication -Need more drinks one after the other (Demands for more drinks) 2. Alcoholism causes depression Alcoholism -> Depression -Affects relationship with family, friends, etc -Alcohol is a depressant A third variable such as trauma may be associated with both variables, creating an apparent relationship between alcoholism and depression Trauma -> Alcoholism -> Depression Confounding Variable An uncontrolled third variable that is known to affect a relationship between two variables Example: Is massed practice inferior to distributed practice? Monday Tuesday Wednesday Thursday Friday Group 1 3 hours Exam Group 2 3 hours 3 hours Exam Group 3 3 hours 3 hours 3 hours Exam Results - Group 3 > Group 2 > Group 1 Research Example: Is massed practice inferior to distributed practice? Levels of IV EV1 EV2 DV 1 day 3 hrs 3 days Lousy 2 days 6 hrs 2 days Average 3 days 9 hrs 1 day Great Controlling Extraneous Variables Constants: extraneous variable are held constant Monday Tuesday Wednesday Thursday Friday Group 1 3 hrs Exam Group 2 1.5 hrs 1.5 hrs Exam Group 3 1 hr 1 hrs 1 hr Exam Levels of IV EV1 EV2 DV 1 day 3 hrs 1 day 2 days 3 hrs 1 day 3 days 3 hrs 1 day Eliminating confounding variable: -Same population -Same amount of time -Same instructor -Exam is the same and at the same time Randomization -When the variables cannot be held constant -Subject variables -Manipulated variables (vary order) -Randomization may not be so easy… Research Example -Optimal and non-optimal time of day in aging research -Winocur & Hasher (2004) ->Subjects: Young and old tested at random times ->Sample: A white cylinder above platform ->Test: A black cylinder above platform, the white cylinder is placed randomly at another location above the maze ->Randomization: Found out time in day makes a difference >Old do better at morning, young do better at afternoon Choosing a method: Advantages of multiple methods -Artificiality of Experiments -Ethical and Practical Considerations -Participant Variables -Description of Behavior -Successful Predictions of Future Behavior Evaluating research: three validities Validity = truth and accurate representation of information Construct Validity: -Measuring what we believe to be measuring -Adequacy of the operational definition of variables -Are the results replicable? Internal Validity: -Ability to infer causality from the data -> Need comparisons -Ability to draw conclusions about causal relationships from our data Two general threats to internal validity- When control groups are absent When comparisons are made between ‘nonequivalent groups’ External Validity: -Sampled population can be used to explain the general population (researched study can be used to represent other studies, groups, etc) -Extent to which the results can be generalized to other populations and settings ->Consider cohort effects/time of day effects Note: despite their artificial setting, lab experiments often have great value for understanding human behavior. *Ecological Validity: -Whether measure is an accurate representation in real life ->E.g., Elders doing memory test on computer may affect the results on whether or not their memory is good or bad (Elders may not be willing to or not skilled on using computers) Critically Evaluating Research Construct Validity: Evaluating the adequacy of the operational definition. Is the operational definition sufficiently measuring the construct it claims to measure? Internal Validity: Eva