PSY 2174 - Oct 10 - Notes on Measurements and Construct Validity

When conducting a study, it is essential to define what is being measured using operational definitions.
Variables should be clearly conceptualized so that they can be measured effectively.
Measurements involve determining the relationship between attributes of variables.

Nominal Level
- Definition: A nominal level of measurement uses categories and names without any quantitative value.
- Example Attributes:
  - Political affiliation (e.g., Republican, Democrat, Independent).
  - Religion, mode of transportation, college major.
- Key Point: The numbers assigned (e.g., 1, 2, 3) for categories are simply labels with no intrinsic order or significance.
Ordinal Level
- Definition: Ordinal measurement applies to variables where there is a clear order or ranking among attributes.
- Example:
  - Likert scales (e.g., rating satisfaction from very good to very bad).
  - Educational levels (e.g., high school, bachelor's degree).
- Key Point: While there is an order, the distance between values is not uniform or meaningful. You cannot quantify the exact differences.
Interval Level
- Definition: In interval measurement, the distance between measurements is meaningful and consistent.
- Example Attributes:
  - Temperature in Celsius or Fahrenheit; GPA (e.g., 3.0 vs. 3.2).
- Key Point: There is no absolute zero in interval measurements. For instance, 0 degrees Celsius does not signify the absence of heat.
Ratio Level
- Definition: Ratio level measures have all the properties of interval measurements, but also a meaningful absolute zero.
- Example Attributes:
  - Age, income, weight, number of students in a class.
- Key Point: You can calculate meaningful ratios with these values (e.g., a salary of $50,000 is twice that of $25,000).

Numbers on the shirts of a soccer team:
- Level of Measurement: Nominal
- Reason: Just categories without significance (number 16 is not quantitatively better than number 20).
Hair color of students:
- Level of Measurement: Nominal
- Reason: Hair colors are merely labels; no inherent order.
Ages of students in a statistics class:
- Level of Measurement: Ratio
- Reason: Age can be ordered, has meaningful distances, and contains an absolute zero.
Ratings of movies (e.g., poor to excellent):
- Level of Measurement: Ordinal
- Reason: Ratings are ordered but distances between them aren't mathematically meaningful.
Final grades in the form of letters (A, B, C, D):
- Level of Measurement: Ordinal
- Reason: Grades can show order but cannot quantify distance accurately.
Annual salaries for teachers in Ottawa:
- Level of Measurement: Ratio
- Reason: Monetary value includes a meaningful absolute zero, allowing for comparisons and ratios.

Definition: Construct validity addresses how well a test or measure captures the concept it intends to measure.
Importance: Ensures that what is being observed aligns with theoretical concepts.

Translation Validity: Examines if the conceptual definition translates well into operationalization.
- Face Validity: Does the measure seem valid at face value?
- Content Validity: Does the measure cover all dimensions of the construct?
Criterion-Related Validity: Examines the performance of the measure in relation to other measures.
- Predictive Validity: Can the measure accurately predict outcomes?
- Concurrent Validity: Does the measure distinguish between groups that should differ?
Convergent Validity: Measures of the same construct yield similar results.
Discriminant Validity: Measures of different constructs yield different results.

Reliability: Concerns consistency of a measure across time and scenarios.
Validity: Concerns whether the measure accurately captures what it intends to measure.

Operational definitions and correct levels of measurement are crucial for data collection.
Understanding validity types helps ensure accurate conclusions in research.
Selecting validated measures from databases can enhance research quality.