121-123

Levels of Measurement

Understanding Measurement Scales

  • Degrees, such as in temperature scales, do not imply a complete absence of heat. For example, -30 degrees does not equal 30 degrees less than no heat.

  • The Celsius scale operates similarly, while the Kelvin scale is different as it is based on absolute zero which signifies a complete lack of heat.

Types of Measures

Ratio Measures

  • Ratio measures exhibit characteristics that meet all requirements of interval measures and have a true zero point.

  • Common examples in criminal justice research include:

    • Age

    • Dollar value of property loss (e.g., burglary losses)

    • Number of prior arrests

    • Blood alcohol content

    • Length of incarceration

  • Example involving judges:

    • Judges can categorize themselves based on years of experience, allowing comparison of levels of experience.

    • Years of experience can be ordered (ordinal), measured by equal intervals (interval), and includes a true zero point, enabling ratio comparison.

Implications of Levels of Measurement

  • Table 5.1 illustrates different crime seriousness levels based on a survey conducted on over 200 different crimes:

    • Nominal Measure: Identifies the type of victim (home, person, etc.).

    • Seriousness Scores: Are interval measures ranging from 0.6 (trespassing) to 35.7 (murder).

    • Rankings: Provide a qualitative measure of seriousness without specifying the distance between them.

    • Property Loss: Treated as a ratio measure with an absolute zero point (e.g., burglaries can be summed).

Statistical Techniques and Measurement Levels

  • Different statistical methods necessitate specific levels of measurement:

    • Average property loss is computable as an interval measure.

    • Modal victim type is reportable for nominal variables (most common type in Table 5.1).

  • Researchers may apply diverse levels of measurement to a variable:

    • Example: Age (ratio by default) may be treated as ordinal when examining its relationship with delinquency involvement.

Criteria for Measurement Quality

Measuring Precision

  • Measurement precision refers to how detailed the distinctions between variable attributes are:

    • Example: "43 years old" is more precise than "in her forties."

    • A felony sentence of "18 months" is more specific than "over one year."

  • Generally, precise measurements are superior, but precision may not always yield useful results depending on the research objective.