S + M, Chapter 5, Part 2

Measurement theory involves the assumptions explaining how indicators change values as concepts manifest.
It helps in determining if there is correspondence between concepts/variables and indicators/measures.
The central problem in measurement is whether our indicators reflect actual changes in the concepts they represent.

Measurement Errors: These prevent accurate reflection of true differences in concept manifestation.
- Types of Measurement Errors:
  - Real Differences: Reflect actual differences in the property being measured.
  - Artificial Differences: Caused by the measurement process, not reflecting reality.
Measurement errors must be understood and controlled to avoid misleading results.
Any differences attributed to non-real differences are termed as measurement error.

Differences in Characteristics Among Cases:
- For example, intelligence levels affecting responses to political ideology questions.
Temporary Characteristics Differences:
- A person's mood or state of health can influence questionnaire responses.
Subjects' Interpretation of Measurement Instruments:
- Ambiguous wording may lead to varied interpretations and responses.
Setting of Measurement:
- The environment and interviewer characteristics can impact responses.
Administration of Measuring Instruments:
- Errors by interviewers or issues like poor lighting can distort recorded responses.
Processing and Analysis of Data:
- Coding and data entry errors can misrepresent case differences.
Response Forms of Measuring Instruments:
- Variances in respondents' abilities (reading, writing) may skew measurement results.

Systematic Errors: Consistent errors across all applications leading to invalid outcomes.
Random Errors: Vary by application, affecting measures inconsistently.
Both types can distort research findings, emphasizing the need for careful data handling.

Validity refers to how well a measure corresponds to the concept it intends to represent.
Types of Validity:
- Internal Validity: Accurate measurement of theoretical concepts.
- External Validity: Generalizability of findings to other contexts/situations.
- Construct Validity: Relationship validation through multiple measures of the same concept.
Valid measures are both appropriate and complete in reflecting the concept being studied.

History: Events unrelated to the independent variable can influence outcomes.
Maturation: Natural developmental changes may affect results over time.
Instability: Changes in measures due to inconsistent sampling.
Testing Issues: Effects arising from pretest and posttest scenarios.
Instrumentation Issues: Variability in measurement instruments affecting scores.
Regression Artifacts: Changes due to statistical regression effects.
Selection Effects: Discrepancies in group selection impacting outcomes.

Reliability concerns the stability of values yielded by measurements.
Questions about reliability ask if the same value is produced under similar conditions.
Testing Reliability Methods:
- Test-Retest Method: Applying the same measure over time to ensure consistency.
- Alternative Form Method: Comparing different forms of measurement to the same group.
- Subsample Method: Using multiple subsamples to establish reliability across groups.

Pragmatic Validation: Predictive validity through the application of measures.
Construct Validation: Examining relationships between measures and expected theoretical behaviors.
Discriminant Validation: Distinguishing a measure from other unrelated concepts.
Face Validity: Referring to measures that appear valid on their face value.

Ensuring reliability and validity throughout the research process is fundamental to obtaining meaningful and actionable insights from data.