OdinSchool OdinSchool
Login

Data Measurement

line-below

They say 'What gets measured gets managed'.

How do you measure things, though? It's like Apples vs Oranges. You don't measure milk and cheese the same way, right?

Also, just because you have two numbers, you can't always treat them with arithmetic. For example, consider two numbers - 15 and 30. In one case these two numbers represent the count of fruits - apples and oranges. In another case they represent the serial numbers assigned to each state of a country, like US for example, with 15 representing Ohio and 30 representing Idaho.

It's fair to say in case one that oranges outnumber apples by 100%, but a similar arithmetic makes no sense in the case of Ohio and Idaho. Let's explore these nuances in this section.

There are four types of Data Measurement.

  • Nominal
  • Ordinal
  • Interval
  • Ratio

Let's talk about each in detail.

Nominal data is used to Categorize or Classify data. Our earlier example of states being represented with numbers fits this type.

Examples:

  • Gender
  • Religion
  • Country

This is the lowest level of measurement. As numbers don't mean much in this context, you can't quantify this data and compare one data point to another. Going back to our example on states, there is no meaning in comparing Ohio and Idaho based on the serial numbers.

Ordinal data is used to Rank, Order data. It is a higher level measurement compared to the Nominal data type.

Examples:

Likert Scale is an example of this. See the options for the question below.

How happy are you with your car service?

  • Very happy

  • Happy

  • As expected

  • Unhappy

  • Very unhappy

Another example would be the rankings of states based on the cost of living. As per Patriot, Ohio ranks 2nd and Idaho ranks 32nd.

In this context, numbers mean something. As in you could say Idaho is more expensive to live in compared to Ohio.

However, you can't go to an extent of saying it costs 16 times to live in Idaho compared to Ohio (by dividing 32 with 2). That would be ridiculous.

Given that both Nominal and Ordinal data types have little to offer from a quantitative measurement perspective, they are called Nonmetric or Qualitative data.

Interval data is a level above Ordinal data in that the data here is presented in numbers and they have a meaning.

Examples:

  • Temperatures expressed in degrees Celsius or Fahrenheit
  • Growth in GDP of a country
  • Return on an investment

It is interesting to note here that any of these examples can take on a value of Zero. However, a zero value does not represent the lowest possible value.

The distances between successive points is same, and they can also be ranked. However, you can't compare a value with other and draw conclusions like 30 degree Celsius is twice as hot as 15 degree Celsius.

Ratio data is the highest level of measurement. The big difference between Ratio and Interval data is the aspect of zero. In ratio data, zero has a significance. It does mean the absence of a phenomenon.

Examples:

  • Height
  • Weight
  • Volume
  • Temperature in Kelvin

In the case of ratio data, you can perform all arithmetic operations. You can rank data, and you can even draw conclusions by comparing different data points. It is fair to say that a height of 40 meters is double that of a height of 20 meters.