What is Measurement? Scales, Types, Criteria and Developing Measurement Tools

What is Measurement?

Measurement is the process of assigning numbers or values to quantities or attributes of objects or events according to certain rules or standards. It is a fundamental aspect of science, mathematics, and everyday life, allowing us to quantify and compare various aspects of the physical world.

In simple words, measurement means using a yardstick to determine the characteristics of a physical object. In addition to physical objects, qualitative concepts, such as songs and paintings, or an abstract phenomenon, can also be measured. However, measuring qualitative concepts is a comparatively difficult task because numbers cannot be easily assigned to them. For example, it is easy to state that the weight of an object or a subject is 10 kg.

However, if a person is asked to measure a song for its good composition, then it becomes difficult to say that the song is 10 per cent good or so. Today, there exist standardised tools to measure abstract concepts such as intelligence, unity, honesty, bravery, success and stress. High accuracy and confidence can be expected while measuring quantitative characteristics of an object.

Measurement Scales

A measurement scale refers to a classification that defines the nature of information within the numerals assigned to variables. Measurement scales have been classified into four types as shown in Figure:

Let us now discuss the types of measurement scales in detail.

Nominal Scale

In this scale, the variables are named or labelled in no specific order. This is the measurement scale in which numbers are assigned to things, beings or events to classify or identify or label them. All of the nominal scales are mutually exclusive and bear no numerical significance. For example, the assignment of different numbers (1, 2, 3, …) to cricket players in a team, books in a library, and computers in the Internet café. These numbers cannot be used to perform mathematical operations.

If 11 players of a cricket team are assigned numbers from 1 to 11, finding the average of 1 to 11 does not signify any meaning. In this case, the only use of these numbers is to count the team members. The nominal scale represents the lowest level of measurement. However, it is helpful when there is a need to classify data. For example, for a question “What are your political views?”, we can have the following nominal scale:

Ordinal Scale

This is the scale that only implies greater than or less than but does not answer how much greater or less. Only inequalities can be set up with respect to ordinal scale and other arithmetic operations cannot be performed. The ordinal scale can be used to make only comparisons. In ordinal scale, data is shown in order of magnitude. An example of the ordinal scale is as follows:

For example, the ordering of colour preferences of Mr. A are:

Thus, with the ordinal scale, researcher can use median or mode to determine the central tendency of a set of ordinal data.

Interval Scale

This is the scale in which the interval between successive positions is equal. The positions are separated by equally spaced intervals or basis. For example, a person represents his/her level of happiness along a scale rated from 1 to 10.

With the interval scale, the following conclusions can be made:

• Number 1 represents least happy and number 10 represents most happy.
  
  
• Number 6 represents a higher level of happiness than number 5.
  
  
• The difference in the level of happiness between 6 and 5 is same as the difference in the level of happiness between 8 and 9.
  
  
• However, a major problem with the interval scale is that the ratio of two observations cannot be taken. It cannot be stated that number 4 represents double happiness level as compared to number 2.

The basic limitation of the interval scale is that it does not contain an absolute zero. A simple example of the interval scale is the scale of temperature in which absolute zero is unattainable theoretically. Therefore, the interval scale does not have the provision to measure the absence of any characteristic such as zero happiness (or absence of happiness). The interval scale contains features of nominal and ordinal scales. In addition, it involves the concept of equality of intervals. In the interval scale, more arithmetic operations, such as addition and subtraction can be performed. Mean can be calculated for interval scale.

Ratio Scale

This is the scale that contains absolute or true zero, which implies the absence of any trait. For example, on a centimetre scale, zero implies the absence of length or height. In the ratio scale, it is possible to take ratio of two observations. For example, it can be stated that the weight of Ram is twice that of Shyam. The ratio scale is the most powerful scale of measurement, as almost all statistical operations, which cannot be performed by other scales, can be performed by it.

Developing Measurement Tools

Any tool used to measure or collect data is called measurement tool, which is also known as assessment tool. There are several types of measurement tools, such as observations, scales, questionnaires, surveys, interviews and indexes (or indices). There are four stages of developing measurement tools, which are explained as follows:

Concept Development

This is the first stage in the process of development of measuring tools. At this stage, the researcher develops a good understanding of the topic he/she wants related to his research study. For example, research on the pros and cons of the multiparty political system requires a proper understanding of the concept behind this system.

Without referring to the theories related to the multi-party system, a good understanding about this system cannot be developed. However, if the research is being done on a concept such as stress (which has already been researched extensively), no exclusive concept building is required.

Specification of Concept Dimension

After developing a concept at the first stage, the researcher is required to clearly identify the dimensions of the concept. For example, when a researcher wants to conduct a study related to image of a company, he/she may relate image of the company to dimensions or factors such as customer service level, customer treatment, product quality, employee treatment, social responsibility, corporate leadership, etc.

Selection of Indicators

At this stage, indicators for the research subject are selected. Indicators help in measuring the elements of a concept such as knowledge, opinion, choices, expectations and feelings of respondents. Examples of indicators are variables.

For example, the effectiveness of a medicine (concept) used for treating a chronic disease is related to indicators such as changes in the mortality rate, recurrence of that disease, etc. The researcher may convert these indicators into variables that can be measured. For instance, number of deaths caused by that disease, number of patients who were again affected by that disease, can be the indicators of the said concept.

Formation of Index

After determining multiple elements of a particular concept and selecting suitable indicators for the research, the researcher needs to combine all the indicators into a summated scale because separate indicators cannot give a certain measurement of the concept. For example, a price index is based on the weighted sum of prices.

Criteria of Good Measurement Tool

What should be the characteristics of a good measurement tool? The answer to this question is that the tool should clearly and accurately indicate what the researcher intends to measure. In addition, the good measurement tool should be easy to use and should give reliable results.

The two major characteristics of a good measurement tool are explained as follows:

Reliability

Reliability of a good measurement tool refers to the degree of confidence with which the measurement tool can be used to derive consistent results upon repeated application. A reliable instrument is not necessarily a valid instrument. However, a valid instrument must be reliable.

For example, a scale is used to measure the weight of objects. The scale consistently shows all objects to be overweight by 2 kg. In that case, the scale can said to be reliable because it is consistent, but it is not valid at all.

The reliability of the measurement tool can be affected because of factors such as subject-related factors, observer/interviewer reliability, instrument reliability and situational reliability. Reliability of physical instruments can be tested by using calibration. However, in case of non-physical instruments such as questionnaires, the reliability of instruments is based on their stability and internal consistency. Test-retest method is used to measure reliability of the instruments.

Validity

Validity refers to the degree to which a measurement tool succeeds in measuring what is expected to be measured. Reliability and validity are interdependent concepts. There can be reliability without validity; however, there can be no validity without reliability. There can be three types of validity based on which the validity of a measuring instrument is assessed.

These are content validity, criterion-related validity and construct validity. Content validity refers to the judgement of one or more subject matter experts regarding the measurement tool. At times, researchers use a well-established measurement procedure [for example, an established survey (say Survey A) for measuring stress level] that measures a variable of interest (say, V) as a base for developing another measurement procedure [for example, newly created survey (say Survey B) for measuring stress level] that measures the same variable of interest V.

Construct validity indicates how well the measurement tool measures different constructs. Validity is assessed by one of three methods: content validation, criterion-related validation and construct validation.