Study Notes on Levels of Measurement
Chapter 2: Levels of Measurement and Their Importance
- Source: Salkind, Tests and Measurement 3e. SAGE Publishing (2018) unless otherwise noted.
Introduction to Variables and Measurement
- Key Questions: What is a variable? What is measurement?
- Source: Salkind, Tests and Measurement 3e. SAGE Publishing, 2018.
Origin of Measurement Levels
- S.S. Stevens (1946): The foundational work by Stevens focused on how different types of variables are measured.
- Key Concerns:
- How does the precision of measurement help in classifying variables?
- Is it possible to create a system to categorize outcomes based on the characteristics of variables?
- Source: Salkind, Tests and Measurement 3e. SAGE Publishing, 2018.
Importance of Measurement Methods
- Core Concept: The methodology used to measure a variable significantly impacts the level of measurement.
- Example: Height Measurement
- Group classification based on height can simply rank individuals or provide detailed numeric data which is more informative.
- Source: Salkind, Tests and Measurement 3e. SAGE Publishing, 2018.
Levels of Measurement
- Definition: The levels of measurement indicate how much information is conveyed by the outcome measure.
- There are four levels of measurement: Nominal, Ordinal, Interval, and Ratio.
- Source: Salkind, Tests and Measurement 3e. SAGE Publishing, 2018.
Nominal Level of Measurement
- Characteristics:
- Focus on differences in quality rather than quantity.
- Categorical or discrete in nature.
- Qualitative - involves labels without inherent order.
- Mutually Exclusive: A variable can only belong to one category at a time.
- Example:
- Categorization such as nurse vs. non-nurse (Verhaeghe et al., 2003).
- Room assignments such as blue room vs. red room.
- Source: Salkind, Tests and Measurement 3e. SAGE Publishing, 2018.
Ordinal Level of Measurement
- Characteristics:
- Allows for variables to be ordered or ranked.
- Groups are placed in categories with an implied order, indicating “better than” or “worse than” relationships.
- Limitations: Does not indicate how much “better” one category is over another.
- Example: Ranking common childhood fears (Burkhardt, Loxton, & Muris, 2003).
- Applications:
- Pop music charts, college football standings.
- Note: There exists some quantitative difference, but the exact magnitude between ranks is unknown.
- Source: Salkind, Tests and Measurement 3e. SAGE Publishing, 2018.
Interval Level of Measurement
- Characteristics:
- Assigns values based on an underlying continuum with equal intervals.
- Provides a significant amount of information, revealing what the score truly means.
- Facilitates judgment about performance or levels (how much “better” or “worse”).
- Example: Scores on the Wechsler Intelligence Scale for Children with low birth weight participating in early intervention programs (Hill, Gunn, Waldfogel, 2003).
- Equal Intervals: Each incremental increase reflects an equal amount of the variable being measured. For example, the interval between scores of 1 and 4 is equivalent to that between 6 and 9.
- Source: Salkind, Tests and Measurement 3e. SAGE Publishing, 2018.
Ratio Level of Measurement
- Characteristics:
- Assumes the existence of an absolute zero, denoting a complete absence of the trait being measured.
- Commonly utilized in physical and biological sciences.
- Examples in behavioral or social sciences are rare.
- Interpretations of Zero:
- A zero score on an intelligence test does not indicate a lack of intelligence, while a zero on a finger-tapping task represents absence of a specific ability.
- Examples:
- Measurement categories such as rainfall, height, weight, and temperature.
- Nutritional Values:
- The difference in calorie content between food items exemplifies true zero – the absence of that nutrient is meaningfully represented.
- Source: Salkind, Tests and Measurement 3e. SAGE Publishing, 2018.
Summary of Characteristics Defining Levels of Measurement
- Table 2.1: Summary of Levels of Measurement and Definitions:
- Nominal: Least information, categorical.
- Ordinal: More information, order can be assigned, but lacks quantitative details.
- Interval: Most information available, allows ranking and can measure quantitative differences.
- Ratio: Most precise, allows for absolute zero, can measure true quantities.
- Source: Salkind, Tests and Measurement 3e. SAGE Publishing, 2018.
- Lesson: When selecting a technique for measuring a variable, opt for the highest level of measurement.
- Most Preferred: Interval level is often sought for access to the most detailed information.
- Example Application: Assessing strength in senior citizens can differ when describing them simply as 'weak or strong', ranking them, or specifying weights they can lift.
- Source: Salkind, Tests and Measurement 3e. SAGE Publishing, 2018.
Final Insights on Measurement
- The level of measurement influences accuracy and precision in research findings.
- Important Note: Accurate measurement processes are linked to higher precision levels of measurement.
- Caution: Researchers must be careful when treating ordinal and interval data as these are inherently different. Misinterpretations can arise.
- Example of Misinterpretation: Treating ordinal scores such as personality tests as if they were interval-level observations can lead to incorrect conclusions.
- Complexity in Instruments: Differences in intelligence test scores (e.g., 100 vs. 105 versus 125 vs. 130) imply varying implications based on their interval nature.
- Source: Salkind, Tests and Measurement 3e. SAGE Publishing, 2018.