Research Design and Analysis

Goals of Behavioral Sciences

Description, Prediction, Determining Cause, Explanation

Description

Describing how events are related

Types of Research

Temporal, covariation, elimination

Types of Research

Rules for medical research developed in response to Nazi Cruelty

Nuremberg Code

Rules for medical research developed in response to Nazi Cruelty

Declaration of Helsinki

Required journals to enforce the Nuremberg Code

Belmont Report

Principal of Benefice, Principal of Respect for a Person's Autonomy, Principal of Justice

Principal of Benefice

Research should weigh benefits against (minimal) risk

Principal of Respect for a Person’s Autonomy

Informed consent and treated allowed to withdraw

Principal of Justice

Benefits and risks should be allocated fairly

APA Ethics Principal A

Benefice

APA Ethics Principal B

Fidelity and Responsibilty

APA Ethics Principal C

Integrity

APA Ethics Prinicpal D

Justice

APA Ethics Principal E

Respect for People’s Righte and Dignity

Informed Consent Includes

Purpose of research, proceduree, risks/benefits, compensation, confidentiality, voluntary participation/permission to withdraw, contact info for questions

Types of Journal Articles

Literature, theory, empirical

Literature Articles

Review of past research

Theory Articles

Integrates past data to form theories

Empirical Articles (definition)

New research

Empirical Artical (Sections)

Abstract, literature/intro, methods, results, discussion

Abstract

Summary

Literature

Describes the problem as well as past studies

Methods

Describes all aspects/procedures/equipment used

Results

Results given in narrative and statistical form, often including graphs and charts

Discussion

Results and study analyzed

Validity

Extent that a conclusion is reasonably accurate or true

Four types of validity

Construct, internal, external, statistical

Construct Validity

How accurately measurements and manipulations represent a variable

Internal Validity

How accurately a study provides a true cause-and-effect explanation

External Validity

How accurately a study can be generalized to the population

Statistical Validity

The accuracy of statistical conclusions

Operational Definitions

Used to make variables measurable

Reliablity

Consistency or stability over time

Test-Retest Reliabilty

Different form of the same test repeated

Internal Consistency of Reliability

Assessments of reliability at only one point in time

Split-Half Reliability

Correlation of the total score of the first half of a test with the second half

Cronbach's Alpha

Average of all possible split-half reliability coefficients

Item-Total Correlation

Correlation of each item with the total score

Types of Construct Validity

Concurrent, convergent, discriminant

Concurrent Validity

How well a test measure aligns with a criterion measured at the same time

Convergent Validity

Scores of the measure are related to scores of the same or similar concepts

Discriminant Validity

Scores are not correlated to unrelated measures

Response Set

Answering questions out of habit or general attitude rather than thinking through the question

Social Desirability Response Set

Answering questions based on what feels most socially acceptable or desirable

Three Catergories of Survery Questions

Facts/demographic, behaviors, beliefs and attitudes

Focus Group

Interview with a small number of participants together at the same time

Probability Sampling

Each member has a specific chance of getting chosen

Single Random Sampling

Equal probability of getting chosen

Stratified Random Sampling

Population is divided into subgroups (srata), random sampling from each strata

Cluster Sampling

Existing groups or geographic areas are identified and sampled from

Non-Probability Sampling

Non-random sampling with unknown probability of members being chosen

Convenience Sampling

Taking participants as you find them

Purposive Sampling

Sampling with predetermined criteria

Snowball Sampling

Participants recruit others

Quota Sampling

Sample reflects the numerical composition of the population

Solomon-Four Group Design

Only half of participants are given the pretest

Cons of Repeated Measures Design

Order effect, practice effect, fatigue effect, carryover effect

Latin Square

Counterbalance order effect, each treatment appears at each ordinal position and each condition precedes and follows each condition

Probability Sampling (reason)

Sampling to accurately describe the population

Non-Probability Sampling (Reason)

To test hypotheses about behaviors

Measures of the Dependent Variable

Self reports, behavioral measures (observations), physiological measures

Floor effect

Task is so difficult, no one can perform well regardless of IV

Ceiling Effect

Task is so easy, everyone can perform well regardless of IV

Pilot Study

Trial run of the study with a small number of participants

Manipulation Check

Attempt to directly measure if the IV manipulation has the intended effect

Preregistration

Articulating research question, hypothesis, and analysis of the plan before collecting data

Factorial Design

Experiments with multiple IV's

2 × 2

2 IV's with 2 levels of each IV

Main Effect

Direct effect of one IV on the DV

Interaction

When the effect of one IV depends on a particular level of the other IV

Single-Case Experiment Design

Extensive, detailed study of one or a small group of participants

Baseline Period

Time without the treatment that is studied as the control

Reversal

Removing the IV to see if the IV is truly causing the reaction

Quasi-Experimental

Approximates the control

Propensity Score Matching

Subjects paired based on scores separated into the control or experiment group

Cross-Sectional

Developmental experiment, multiple age groups studied at one point in time

Longitudinal

Developmental experiment, follows individuals/groups as they age

Sequential

Combination of cross-sectional and longitudinal

Cohort Effect

Differences in age groups that can be attributed to social, economic, or political conditions and not age