Understanding Data Measurement Scales

Why Data Levels Matter

Before we dive into the definitions, it's important to understand why this classification matters. Imagine trying to calculate the "average" gender in a study group. The result would be meaningless. This is because gender is a categorical variable, not a numerical one you can perform arithmetic on. Recognizing the level of measurement prevents such errors and guides you toward valid analysis.

The video below provides a great overview of the main data types and why it's so important to understand them before you start analyzing your data.

As the video explained, we can broadly group data into two main types: categorical (representing qualitative characteristics) and numerical (representing quantitative measurements). Within these two groups lie our four levels of measurement. Let's look at them one by one, from the simplest to the most sophisticated.

1. Nominal Scale: The "Naming" Level

The nominal scale is the most basic level of measurement. Think of it as simply assigning names or labels to categories. The key feature is that there is no natural order or ranking among the categories.

In pharmacy and clinical research, common examples of nominal data include:

• Patient's blood type: (A, B, AB, O)
• Smoking status: (Smoker, Non-smoker)
• Presence of a disease: (Yes, No)
• Type of medication: (Tablet, Capsule, Injection)

In each case, the categories are distinct and mutually exclusive, but you cannot logically say that "Blood Type A" is higher or lower than "Blood Type B." They are simply different labels.

2. Ordinal Scale: The "Ordering" Level

The ordinal scale takes it one step up from the nominal scale. It also involves categories, but with a crucial difference: these categories have a meaningful, natural order or rank. However, the exact difference or interval between the ranks is not uniform or measurable.

This data type is very common in healthcare for measuring subjective states. Excellent pharmacy-related examples can be found in the "Pharmacy Joe" and ACCP articles.

As the article highlights, examples of ordinal data include:

• Patient satisfaction: (Very Unsatisfied, Unsatisfied, Neutral, Satisfied, Very Satisfied)
• Pain scale: (A rating from 1 to 10, where 10 is the worst pain)
• New York Heart Association (NYHA) Functional Classification: (Classes I, II, III, IV for heart failure severity)

For all of these, you know that NYHA Class IV is more severe than Class II, and a pain score of 8 is worse than 4. However, you can't say that the difference in severity between Class I and II is the same as between Class III and IV.

3. Interval Scale: The "Equal Interval" Level

The interval scale is the first of our numerical data types. It has the properties of ordinal data (a meaningful order) but adds one more: the intervals between values are equal and measurable. The one thing it lacks is a true zero point. A "true zero" means the complete absence of the thing being measured. On an interval scale, zero is just another point on the scale; it's arbitrary.

The classic example is temperature in Celsius or Fahrenheit. The difference between 10°C and 20°C is the same 10-degree difference as between 50°C and 60°C. However, 0°C does not mean there is "no temperature" or "no heat." It's simply the freezing point of water. Because there's no true zero, you can't make ratio comparisons. For example, you can't say that 20°C is "twice as hot" as 10°C.

4. Ratio Scale: The "Absolute Zero" Level

The ratio scale is the highest and most informative level of measurement. It has all the properties of the interval scale (order, equal intervals) but also includes a meaningful, absolute zero.

Most of the clinical measurements you will encounter in pharmacy are on the ratio scale. The ACCP article provides a clear distinction.

Examples of ratio data include:

• Blood pressure (in mmHg): 0 mmHg means a complete absence of pressure.
• Heart rate (in beats per minute): 0 bpm means the heart is not beating.
• Drug concentration in plasma (in mcg/mL): 0 mcg/mL means there is no drug present.
• Height and weight: A weight of 0 kg means a complete absence of weight.

Because there is a true zero, you can make meaningful ratio comparisons. For example, a patient weighing 100 kg is twice as heavy as a patient weighing 50 kg.

Summary of the Four Levels

The image below provides a fantastic visual summary of how these four levels build upon each other. Nominal is just naming. Ordinal adds order. Interval adds equal spacing. And Ratio adds a true zero.

Let's Practice

To check your understanding, try to classify the following variables collected in a pharmacy study. What is the highest level of measurement that can be applied to each?

1. A patient's primary language (e.g., English, Marathi, Hindi).
2. The stage of a patient's cancer (e.g., Stage I, Stage II, Stage III, Stage IV).
3. The year a new drug was approved by the FDA (e.g., 2018, 2020, 2022).
4. A patient's exact white blood cell count (cells/µL).

Think about your answers, then check them against the solutions below.

Solutions:

1. Nominal: The languages are just categories with no inherent order.
2. Ordinal: There is a clear order (Stage IV is more advanced than Stage I), but the difference in severity between stages is not necessarily equal.
3. Interval: The years are ordered with equal intervals (1 year), but there is no "true zero" year. The year 0 does not mean the absence of time.
4. Ratio: There is a clear order, equal intervals (1 cell/µL), and a true zero (0 cells/µL means no white blood cells).

Conclusion

In this lesson, you've learned to classify data into the four fundamental levels of measurement. This is a crucial first step in any statistical analysis.

Key Takeaways:

• Nominal: Categorical data with no order (e.g., blood type).
• Ordinal: Categorical data with a meaningful order but unequal intervals (e.g., pain scale, patient satisfaction).
• Interval: Numerical data with order and equal intervals, but no true zero (e.g., temperature in Celsius).
• Ratio: Numerical data with order, equal intervals, and an absolute zero (e.g., blood pressure, weight, drug concentration).

Understanding these levels prevents you from using inappropriate statistical methods and allows you to correctly interpret research findings.

In our next lesson, we will look at another important way to classify data: distinguishing between discrete and continuous variables, which builds directly on what we've learned about numerical data today.