临床与咨询心理学导论 7 - Research Methods 1

L7 Research Methods 1: Background & Types of Research in Clinical Psychology 

参考文献/图片来源：Pomerantz, A. (2013). Clinical psychology: science, practice, and culture (3rd ed.). Thousand Oaks, CA: SAGE Publications.

7.1 Background on Research Methods

Example: Conflict Between Parents & Children

• Some degree of conflict between caregivers and children is typical. Extreme forms are associated increased risk for mental health problems.

• Why do some parent-child relationships have particularly high degrees of conflict? What kinds of problems do we see in these relationships? How can we reduce this conflict?

- What factors influence conflict?

• Potential factors that might influence parent-child conflict: stress levels, parenting techniques, child behavior, parent mental health, child mental health, communication, child maltreatment, sources of support, etc.

• Research helps us figure out how all these variables relate to one another, which helps us figure out how to address the problem.

- Specifying Hypotheses

• Expect links among: Harsh Parenting, Stress, and Parent-Child Conflict

• Path diagrams: Schematics depicting relationships:

- Double-headed arrow connecting two variables: Correlation

- Single-headed arrow connecting two variables: Causation

• Common variable relationships described in hypotheses:

- Correlational: Variables related in non-causal fashion: “Harsh Parenting, Stress, and Parent-Child Conflict are related”

- Causal: One variable (or more) causes another variable (or more): “Harsh Parenting and Stress both cause Parent-Child Conflict”

- Mediation: One variable accounts for / explains the relationship between the others: “Harsh Parenting mediates link between Stress and Parent-Child Conflict”

• Stress leads to Parent-Child Conflict because it causes Harsh Parenting Techniques.

• Relationship between Stress and Parent-Child Conflict disappears if Harsh Parenting is added to model

- Moderation: A variable influences the strength or direction of the relationship between the others: “Stress moderates link between Harsh Parenting and Parent-Child Conflict”

• Strength of relationship between Harsh Parenting and Parent-Child Conflict varies based on how much Stress is present

• Stress increases likelihood of Parent-Child Conflict only if Harsh Parenting is already present

Independent & Dependent Variables

• Independent Variables - Manipulated by the experimenter

• Dependent Variables - Hypothesized to change as a result of the changes in the independent variables

Internal & External Validity

• Internal Validity of Study - Extent to which changes in the dependent variable can be attributed to changes in the independent variable

• External Validity of Study - Generalizability of findings beyond this specific study. Same finding to be expected for different situations

High in internal validity - usually low in external validity

High in external validity - usually low in internal validity

7.2 Research Designs

• Correlational Studies, |r|<1

- Correlations have both magnitude and direction

• Magnitude is stronger as |r| increases

• Direction of association is positive or negative

- Positive r: as X increases, Y increases

- Negative r: as X increases, Y decreases

- Zero: independent variables

- Cannot be sure about causality

• “Third” variables may explain observed relationships

• Causal arrows may go in different directions (chicken & egg)

• Experimental Studies

- Examine effect of experimentally manipulated independent variable (IV) on dependent variable (DV) across multiple persons

- Random assignment to levels of the IV

- Example: “Does manipulating level of alcohol consumption (IV) influence performance in a driving simulator (DV)?”

- Higher internal validity than correlational studies

- External validity varies across studies (actually driving vs driving simulator)

Between vs Within Approaches:

• Between-Group: Evaluates groups at single time and the comparisons are made between the groups

• Within-Group: Follows same individuals over time and individuals are compared to themselves - May be very brief (i.e., before and after a task in a lab) or longer

• Mixed-Group: Includes features of both between and within-groups design

• Quasi-Experimental Studies

- Used when researchers cannot randomly assign people to a particular group (either for practical or ethical reasons)

- Example: “Do children who experience emotional abuse (IV) differ in their ability to recognize facial emotions (DV) than children who have no abuse histories?”

- Lower internal validity than experimental studies

- Higher internal validity than correlational studies

Time Frames (all designs)

• Cross-Sectional: Evaluates sample(s) at single time

- Much cheaper, less time-intensive, good starting point

• Longitudinal: Follows same persons over time

- Helps understand development, course, and treatment of psychopathology

• Meta-Analysis

- Integrates findings from multiple studies

- Refers to both a method and a statistical technique

- Example: Taylor, Swerdfeger, & Elsick (2014) performed a metaanalysis on studies related to vaccines and Autism Spectrum Disorder (ASD); included over 1.25 million children across many studies; found no connection between vaccines and ASD risk.

- Benefits of meta-analysis

• Greatly enhanced external validity.

• Greatly enhanced ability to detect small effects, that may be missed in a single study.

• Gold-standard approach to literature review and synthesis.

Samples Used Across Designs

Types of Samples

- Community Sample

• Individuals from the community

• Usually relatively low rates of disorders

- Clinical Sample

• Individuals have the disorder (e.g., have alcohol use disorder)

- Analogue Sample

• Individuals whose behavior approximates, related to the disorder of interest (i.e., heavy/binge drinkers - have alcohol use disorder)

7.3 Functions of Research in Clinical & Counseling Psychology

Epidemiology

- Cornerstone of public health

- Identifies factors affecting the health of populations

- Demonstrates the need for prevention/intervention

- Helps determine funding priorities

• Epidemiology research in Clinical Psychology:

- Identify rates of mental disorders

- Identify rates of risk factors

Epidemiology Studies in Clinical Psychology

• The National Comorbidity Survey Replication - NCS-R (Kessler et al., 2004)

• Goal - Uncover trends in mental health disorder prevalence, impairment, and service use; Get more detailed information than previous studies

• Sample - Nationally representative US sample (N = 9,836)

• Method: Face-to-face interviews (lasting ~2.5 hours, ranging from 1.5 to 6 hrs)

- Better accuracy than internet, phone, mail surveys and higher quality data

- Higher response rates (many people agree to participate)

- Ability to manage participant burden/fatigue (e.g., providing rests)

- Payment: $50

• Primary NCS-R Findings

- Comorbidity: 45% of people with 1 mental disorder had ≥ 2 diagnoses

- About 33% received “minimally adequate treatment” in the past year

“minimally adequate treatment” - number of sections received