Nonexperimental Designs Psyc 231 Research Methods Nonexperimental Designs

Non-experimental Designs Psyc 231: Research Methods

Non-experimental Designs l l Surveys Developmental Designs Small N Designs Quasi-experiments

Developmental Designs l l l Used to study development or changes in behavior Describe relationship between age and other variables Three main types l l l Cross-sectional Longitudinal Cohort-sequential

Cross-sectional Designs l l Uses a separate group of participants for each age group being compared Different groups measured once and compared to each other Between subjects design Most commonly used

Cross-sectional Designs l l Study the development of memory over time All three age groups tested at one point in time Age 4 Age 7 Age 11

Cross-sectional Designs l Advantages l l l Short period of time No real commitment Gather all data at one time

Cross-sectional Designs l Disadvantages l l l Cohort/Generation effects Does not tell real development of individual Cannot infer causality

Longitudinal Designs l Same participants are observed over time l l l Assesses stability of traits Individuals are compared to self throughout repeated measures over time Within subjects design

Longitudinal Designs l l Study of the development of memory over time Same participants tested over time Age 4 Age 7 Age 11

Longitudinal Designs l Advantages l l l No generation effects Examine individual differences Can see developmental changes

Longitudinal Designs l Disadvantages l l l Very time consuming and costly Hard to find patient participants – Subject Attrition/Mortality Researchers lose interest Practice effects Cross-generational effects l l Conclusions based on members of one generation may not apply to other generations Cannot determine causality

Longitudinal Designs l Wisconsin Longitudinal Study (WLS) l l l Began in 1957 and is still on-going (50 years) Originally studied plans for college after graduation Now it can be used as a test of aging and maturation

Cohort-sequential Designs l l Measure groups of participants as they age Combines the best features of both longitudinal and cross-sectional designs l l Studies specific age groups over time Both between and within subjects design

Cohort-sequential Designs l l Study of the development of memory over time Test multiple age groups over time Age 4 Age 8 Age 12 Age 16

Cohort-sequential Designs l Advantages l l Saves time Get more information l l l Long-term effects and developmental changes Compare to different ages No generation effects

Cohort-sequential Designs l Disadvantages l l More time consuming than cross-sectional Does not mean causation

Small N Designs l l l Study one or few participants (typically 3 -8 participants) Each individual is analyzed separately Common type of design until 1920’s Still used in some areas of research: clinical settings, phenomenon Different from case studies

Small N Designs l l Typically observe participants Baseline studies l l Observation/testing generally occurs at 3 points l l Effect doesn’t occur before IV (baseline) Show that effect occurs with IV (treatment) Doesn’t occur without IV (reversibility) Before treatment, after reverse treatment Examine level and trend to determine effect

Small N Designs l Level l How frequent or intense is the behavior? Are the data points high or low? Trend l l Does the behavior increase or decrease? Are the data points flat or on a slope?

Small N Designs l ABA design (baseline, treatment, baseline) l l Must be able to reverse effect l Could not have been due to maturation, history, etc. Effectiveness of a drug

Small N Designs l Advantages l l l Focus on individual performance Can see big effects Avoid some ethical problems (nontreatments/controls) Allows to look at unusual (and rare) types of subjects Often used to supplement large N studies, with more observations on fewer subjects

Small N Designs l Disadvantages l l l Generalizability Effects may be small relative to variability of situation Some effects are by definition between subjects Treatment can lead to a lasting change, so you don’t get reversals Ethical issues with reversing treatment

Small N Designs l l l Hermann Ebbinghaus (1885) studied memory of nonsense syllables on himself Discovered the forgetting curve and learning curve Know a lot about memory today because of him

Quasi-experimental Designs l l Almost “true” experiments but lack of control over assignment of participants Independent variable cannot be manipulated (inherent confound) l l l Subject variable Time could be variable (Developmental) Random variable already present

Quasi-experimental Designs l Advantages l l Allows applied research when experiments not possible Threats to internal validity can (sometimes) be assessed Practical and more feasible than true experiments, especially in clinical settings Some generalizability

Quasi-experimental Designs l Disadvantages l l Difficult to make clear cause-and-effect statements Statistical analysis can be difficult l l Most statistical analyses assume randomness Can not randomize assignment to groups

Quasi-experimental Designs l Common types l l Non-equivalent control groups design Time series designs l l Interrupted time series design Control group interrupted time series design

Quasi-experimental Designs l Non-equivalent control groups design l l l Typically used as a pretest-posttest Assignment based on already established variable Between subjects design Non-Random Assignment Dependent Variable Measure Independent Variable Experimental group Dependent Variable Measure participants Measure Control group Measure

Quasi-experimental Designs l Non-equivalent control groups design: Pretest -posttest l Example l l Individuals high on self-esteem and low on selfesteem Pretested on depression levels Intervention given to low self-esteem group Posttested on depression levels

Quasi-experimental Designs l Time series designs l Interrupted times series design l Observe on several occasions before and after the independent variable occurs l Within subjects design obs obs Treatment obs obs l l The pretest observations allow the researcher to look for pre-existing trends The posttest observations allow the researcher to look for changes in the trends

Quasi-experimental Designs l Time series designs l Control group interrupted time series design l l l A variation of the interrupted time series designs Series of observations followed by treatment for experimental condition Compared to a control group obs obs Treatment obs obs obs

Questions?