7. history

Foundations of Quantitative Measurement

• Variables in clinical psychology research:
  1. Cognitive
  2. Affective
  3. Behavioural
  4. Biological
  5. Social
• Construct: Underlying psychological concept.
• Measure: Indicator or way of observing a construct.
• Operationalization: Process of going from construct to measure.
• Construct is latent and inferred from measurement operations.
• Converging operations: Using multiple indicators to measure constructs.
• Reactivity of measurement: Participants' behavior changes due to being measured.

Measurement Sources and Approaches

• Self-report: Attitude questionnaires, symptom checklists.
• Observation: Behavioural observation, brain scan.
• Qualitative: Qualitative interviews, diaries, journals, participant observation, projective tests.

Advantages of Quantitative Measurement

1. Greater precision in measurement.
2. Well-developed theory of reliability and validity.
3. Established statistical methods for data analysis.
4. Data can be easily summarized.
5. Facilitates comparison across individuals.
6. Fits well with hypothetico-deductive approaches.
7. Sampling theory can be used to estimate generalizability.

Positivism

• Focuses on observable facts.
• Applies methods used in physical sciences to social sciences.
• Emphasizes objectivity and value-free science.
• Logical positivism restricts philosophical discourse to sensory experience.

Methodological Behaviourism

• Focuses on observable behaviour.
• Excludes inner factors like cognitions and emotions.

Criticisms of Positivism

• Excludes psychological constructs related to human experience.
• Linked to capitalism, potentially reducing everything to numbers.

Chronometric Methods

• Reaction time (RT) increases with task complexity.
• Hick-Hyman law: RT is linearly related to information extracted from the stimulus.

Donders’ Method of Subtraction

• Reaction times (RT) are crucial for studying the organization of mental processes.
• Tasks:
  • Task A: Simple detection task
  • Task B: Discrimination task
  • Task C: Go/no-go task
• Assumptions: consecutive cognitive processes are strictly serial and independent of each other.

Additive Factors Logic

• Mental processing stages and their organization can be determined from systematic statistical interactions obtained with a single task.
• Additive effects indicate different processes; interactions suggest variables affect the same process.

Interpreting Reaction Time Differences

• Observed RT effect might be driven by unknown factors.
• Control conditions help exclude alternative explanations.

Descriptive Designs

• Study phenomena without manipulating variables.
• Uses descriptive statistics (percentages, means).
• May lead to correlational designs.

Correlational Designs

• Find out how variables are related.
• Measure many variables for each participant to study relations between them.
• Cross-sectional or longitudinal.
• Methods: Correlation coefficients, multiple regression, factor analysis, structural equation modelling.

Correlation and Causation

• Correlation does not equal causation.
• Conditions for inferring causality:
  • Covariation
  • Precedence
  • Exclusion of Alternative Explanations
  • Logical Mechanism
• Conceptualizing causality: Path analysis.
• Issues in drawing causal conclusions:
  • Direct causality
  • Reverse causality
  • Spurious relationships
  • Mediating factors
  • Moderating factors

Experimental Designs

• Manipulate independent variable; use control group.
• Establish cause-effect relationships.
• Forms: Factorial, repeated-measures.
• Statistical techniques: ANOVA, t-tests.
• Assess validity to determine the strength of experimental design.

Cook and Campbell’s Validity Analysis

• Internal validity: Differences in the dependent variable are caused by the independent variable.
• External validity: Results can be generalized to other tests.
• Framework encompasses statistical conclusion and construct validity.
• Potential weaknesses in research study are flaws and limitations.

Validity Type & Defining Question

• Statistical conclusion: Is there an effect?
• Internal: Is it causal?
• Construct: What does it mean?
• External: Does it generalize?

Nonrandomized Designs

• One-group Posttest-Only Design (XO): Lacks sufficient evidence for causal inferences.
• One-group Pretest-Posttest Design (O1 X O2): Risky for inferring causation.
• Nonequivalent Groups Posttest-Only Design (NR X O NR O): Major threat is uncontrolled selection.
• Nonequivalent Groups Pretest-Posttest Design (NR O X O NR O (Y) O): Mitigate uncontrolled selection with statistical methods.
• Interrupted Time-Series Design (O1 O2 … O20 X O21 … O40): Requires careful monitoring for interfering events.

Randomized Designs

• Random assignment eliminates selection bias.
• Enables manipulation of a single variable and fulfillment of causality conditions.
• Control and comparison groups isolate effects of key variables.
• Internal validity is often prioritized over external validity.

Design Variations

• Basic Pretest-Posttest Design: ROXO R O (Y) O.
• Multiple levels: Adding more experimental or control groups.
• Multi-factorial Designs: Incorporating more than one between-groups factor.
• Repeated-measures Designs: Assessing the same individuals at multiple points in time.
• Blocking Factors: Grouping participants based on individual differences before randomization.
• Analysis of Covariance: Individual differences have a linear relationship with the outcome variable.
• Ethical considerations and scientific value determine the choice of control group.

Limitations of Randomized Designs

• Ethical constraints prevent randomization in studying negative experiences.
• RCTs may be unnecessary or inappropriate for obvious interventions.
• Design requirements can make RCTs unrepresentative of normal clinical practice.
• RCTs do not account for patient choice and may be influenced by researcher allegiance.
• Practical issues: Ensuring group equivalence, dealing with attrition, preventing condition leakage, obtaining staff cooperation.
• RCTs are costly and time-consuming.