Understanding Research Design

• Research design is a detailed plan or blueprint for testing hypotheses and analyzing data.
• It ensures relevant data is collected for objective analysis of research problems.
• It helps researchers draw valid conclusions about the relationship between variables.
• The selection of a research design should be based on the purpose, variables, and conditions of the research, not personal whims.
• The purpose of any research design is to maximize relevant information while minimizing costs.
• Research designs answer research questions objectively, validly, and economically.
• They guide data collection, variable control, and statistical analysis.
• They act as a control mechanism for unwanted variances: experimental, extraneous, and error variance.

Experimental Variance

• Produced by experimenter's manipulation of independent variables on the dependent variable to achieve valid and objectives
• Investigator designs experiment maximize experimental variance
• For example, an experiment studying the influence of differential rewards on learning involves randomly placing subjects into groups of high, medium, and low reward.
• Maximize variance by creating diverse experimental conditions

Extraneous Variance

• Extraneous variance (or control variance), is produced by extraneous variables.
• Researchers control relevant variables to eliminate variances.
• Controlling extraneous variables is achieved through methods discussed elsewhere (Chapter 20).

Error Variance

• Error variance refers to variability in measures that occur due to uncontrollable factors.
• These factors relate to individual differences (attitude, motivation, ability) or measurement errors (trial differences, emotional state, etc.).
• Error variance is self-compensating, with positive and negative variability.
• Error variance tends to cancel out in repeated measurements, leading to a zero mean.
• Error variance is unpredictable due to its basis in random errors which differs from systematic variance is predictable.
• Minimizing error variance improves the reliability of measures, strengthening external validity.
• It also allows systematic variance to show significance if it exists.
• A large error variance can overshadow the impact of systematic variance.
• Controlling experimental conditions minimizes error variance.
• Uncontrolled conditions lead to accumulated error variance.
• A good research design avoids 'confounding', where multiple variables change concurrently, preventing attribution of effects to a single variable.
• Confounding introduces vagueness in cause-effect relationships

Criteria of Research Design

• Behavioral researchers use design to ensure data discipline in research.
• Strong research designs can be distinguished from weak ones based criteria
• Key criteria include: capability to answer research questions, variable control, and generalizability.
• A good research design properly answers research questions.
• Selecting an inappropriate design for the research question leads to inadequate hypothesis testing.
• Researchers need to ensure the design aligns with the question being investigated.
• Matching subjects without a relevant relationship between matching variables and the dependent variable results in a weak design.
• For example, factorial designs are appropriate for testing interaction hypotheses.

Control of Variables

• Good research design should control all extraneous variables
• Extraneous variables are like to independent variables that can influence dependent variables.
• Uncontrolled extraneous variables are unwanted in research and produces a weak research
• Randomization is considered a good way to control extraneous variables
• Includes random subject selection, random subject assignment to groups, and random treatment assignment.
• Randomization controls extraneous variables, enhancing internal validity.

Generalizability

• Generalizability refers to the extent to which research results can be applied to other subjects, groups and conditions not within the research.
• Good designs allow results to be generalized to larger groups.
• Generalizability is complex, involving technical aspects of research and broader problems of basic and applied research.
• In applied research, generalizability is a key concern.

Basic Principles of Experimental Design

• Analysis of objective data to determine cause-effect between an independent and dependent variable.

• There are three principles of an experimental design: replication, randomization and local control

• There are steps to follow for experimental designs to be effective

Replication

• Replication refers to repetition of the experiment, using a nearly identical procedure with a different set of subjects, in a different setting and at a different time.
• Replication revalidates previous studies and experimental error.

Randomization

• Randomization ensures tests are valid.
• Statistical tests depends on observations being dependent
• Independence is achieved through random assignments of subjects to treatments.
• Randomization controls extraneous variables automatically.

Local Control

• Local control balances, blocks and groups subjects in an experimental design.
• Grouping = Assigning homogenous subjects into a group
• Blocking = Assigning experimental units to blocks
• Balancing = Assigning experimental units to treatment to be balanced
• Designs must possess property of local control.