Quantitative Research: Experimentation, Scales & Measurement Errors

Quantitative Research

3. Experimentation

What it is: It is the controlled and systematic observation of the effects that occur when you manipulate variables — one or more characteristics of the experimental situation — with the aim of establishing causal relationships between variables.

How

How: Construction and application of the independent variable, regulated upon whom, when and where.

Features

• Manipulation of the independent variable (the experimental treatment).
• Control of context.
• Subjects, equalization and experimental groups.

Problems

• Difficulty of manipulating variables.
• The experimental situation is artificial; because control must be built, it is difficult to generalize.
• Experimenter effects may influence the results.
• Ethical handling.

4. Scales

"There are procedures developed by social research to measure, to operationalize complex concepts we cannot directly observe, measuring them in a systematic and formalized way."

What it is

What it is: A coherent set of elements that are considered indicators of a general concept. It operationalizes a complex concept as measured.

Most common use

Most common use: Attitudes. The unit of analysis is the individual; the concept is to measure attitude and opinion items.

Types

Types: Likert, Guttman (scalogram), etc.

Errors in Measurement and Data Collection

At that time the error occurs:

• The selection of indicators.
• When we are collecting data.
• Coverage errors.
• Sampling error: The error of the sample selected in accordance with mathematical procedures; it is the only thing we can control.
• Non-response error: When an individual of the sample refuses to respond.
• At the time of collection: observation, interview.
• In the questionnaire developed (interview, etc.).
• How it is administered (mail, telephone, etc.).
• In the data transaction.
• The development of the analysis of certain procedures.

Summary of error types

ERROR CORBETT

ERROR DURING Operationalization: Occurs in various stages of research.

SAMPLING ERROR: It is the only one that is controlled.

Non-response errors.

SYSTEMATIC ERRORS IN PHASE produced by the researcher or by the subject we interviewed.

ERROR BY INSTRUMENT: On the instrument we use.

Validity

Content validity: Indicators cover the specific content and meaning of the concept.

Criterion validity: Correspondence with a concept and an external criterion used (behavior already used; comparing results with other external indicators used in other studies).

Predictive: To relate the data with another indicator measured at a different time.

Simultaneous: Related to another indicator at the same moment in time.

Construct validity: How the new indicator is related to the level of construct.