What is Scaling Techniques? Types, Classifications, Techniques

What is Scaling Techniques?

Scaling techniques, also known as scaling methods or scaling procedures, refer to a set of statistical methods used to assign numerical values to objects or variables in order to represent their relative positions or magnitudes. Scaling techniques are commonly employed in various fields, including social sciences, market research, psychology, and data analysis, to transform qualitative or ordinal data into quantitative data that can be more easily analyzed and compared.

Scaling evolved out of efforts in psychology and education. However, scaling remains one of the most hidden and misunderstood aspects of social research measurement. It aims to perform one of the most complicated research tasks, that is, measuring abstract concepts. Scaling refers to a procedure in which numbers or other symbols are assigned to a property of objects. In simple words, scaling is a method to assign quantitative numbers to abstract concepts. Usually, in a scale, there are highest and lowest points with several intermediate points between these two.

Types of Scaling Technique

A scale refers to a tool or mechanism through which individuals are distinguished as to how they differ from one another on the variables of interest of research. Different researchers use different scales depending on the needs of their study.

The different types of scales are shown in Figure:

Comparative scales

Comparative scales include scales wherein the researchers ask the respondents for their relative preference between two or more objects. For example, “Do you prefer Colgate or Babool?” Examples of comparative scales include paired comparison, rank order, and constant sum scale.

Ranking Scale

Ranking scales are used to make relative judgements. There are two approaches used in ranking scales, which are:

Paired comparison

This approach of ranking scale provides a way to make comparisons among objects. For example, for almost every product, there are numerous brands. In that case, the task of respondents becomes quite difficult. According to the method of paired comparison, if there are ‘n’ numbers of objects, then the respondents need to make (n (n–1))/2 comparisons. If there are 8 products, then the respondents need to make (8(8–1/2)) = 28 comparisons.

If the number of comparisons becomes quite large, then there is a risk because the respondents may show reluctance to take part in the research. In such a case, comparisons can be reduced by applying the law of transitivity. This law says that if A is preferred to B and B is preferred to C, then A would automatically be preferred to C.

Rank order scale

In this approach of ranking scale, the respondents are asked to rank their choices according to their preferences. The following is an example of the rank order method:

In the given example, 6 items are shown. The respondent was asked to rank the items as per his/her preferences. Item C is the most preferred and item B is the least preferred by the respondent.

Constant Sum Scale

In the constant sum scaling, the respondents are asked to rate the different attributes of an object and assign some number of units to each attribute. The respondents have to rate each attribute in such a manner that the total number of units or points equal the total number of units assigned by the researcher or the experimenter. A respondent assign number of units to each attribute based on the importance a respondent assigns to the attribute.

If the attribute holds no importance for an attribute, the respondent can assign zero units to it. For example, an HR professional may create a constant sum scale to know the relative importance of different infrastructural attributes in an organisation such as clean washroom, sports room, gymnasium, canteen, etc.

The respondents must remember that the sum of all the allocated units must be equal to 100. After the response of all respondents has been noted, the numbers of points earned by each attribute are counted. These values can be used to infer results or help in research. Constant sum is an ordinal scale because of its comparative nature.

Non-Comparative Scales

Non-comparative scales are those scales wherein each object is measured independently of the other objects under the same research study. Absolute results are obtained for each object. Examples of non-comparative scales include continuous rating scales, Likert scale, etc. They are generally divided into two categories: continuous rating scale and itemised rating scales.

Continuous Rating Scale

A continuous rating scale is a type of scale wherein the respondents are asked to rate different objects on a continuum according to certain criterion. The rating is given by respondents by marking a point on the continuum.

Itemised Rating Scale (Likert Scale, Semantic Scale, Differential Scale, Stapel)

In itemised rating scale, items are shown in the form of ordered statements in and the respondents are required to select the category that best describes the concerned item. The respondents are asked to select one of the choices according to their preferences or opinions.

The itemised rating scales have a number of brief descriptions associated with each category. Most commonly used itemised rating scales include Likert Scale (Summated scale), Semantic Differential Scale, Thurstone and Guttman scale.

Summated Scale (Likert)

Summated scales are constructed by using the item analysis approach. Such scales consist of a number of statements that express either positive or adverse feelings toward any topic or idea. The summated scale is most frequently used in studying social attitudes. It follows the pattern developed by Likert; thus, the summated scale is also termed as Likert scale. Most commonly, a Likert scale contains five degrees of a statement. Let us know more about the Likert scale with the help of the following example (statement and options).

Statement: The Internet is creating a positive impact on Children.

• Strongly Agree (1)
• Agree (2)
• Neutral (3)
• Disagree (4)
• Strongly Disagree (5)

In the preceding example, there are five degrees of responses for the given statement. The right extreme of the scale shows the strongest approval of the statement, whereas, the left extreme indicates the strongest disapproval of the statement. The middle points are between these two extremes.

Each point on the scale has a numerical value. This example constitutes only one statement, but more than one statement can be used in Likert scale. In the Likert scaling xmethod, each statement is assigned a numerical value. The total score for each respondent is calculated by considering his/her response to each statement.

Semantic Differential Scale

It measures the connotative meaning of objects, events and concepts. The semantic differential scale consists of bipolar adjectives, such as goodbad and valuable-worthless. The respondent is asked to select his/her position between these two adjectives. Let us understand the concept of the semantic differential scale with the help of the following example.

A semantic differential scale analysing candidates for a managerial position is shown in Table:

In Table, two adjectives are shown on two extremes. In between these two extremes, scores (3, 2, 1, 0, –1, –2, and –3) are mentioned to rate candidates according to the level of traits possessed by them. In Table, successful-unsuccessful, progressive–regressive, and true–false represent the evaluative attitude. The potency attitude is represented by the severe–lenient and strong-weak pairs. The rest of the adjectives shown in Table represent the activity factor. The semantic differential scale is widely used to measure the attitude of different people.

Guttman Scale

The cumulative scale consists of a series of statements to which a respondent expresses his/her agreement or disagreement. It is important to note that in the cumulative scale, statements appear in the form of a cumulative series. It means that if there are seven statements and the respondent agrees with statement 4, then he/she would also agree with statements 1, 2 and 3.

Let us understand the concept of cumulative scale with the help of the following example:

• I can do counting
• I can do addition
• I can do subtraction
• I can do multiplication
• I can do division

Bases of Scale Classification

As discussed earlier, scaling is a procedure to assign numbers to abstract concepts. There are some factors on the basis of which numbers are assigned.

These factors are explained as follows:

• Subject orientation: In this basis of scaling, differences in the responses obtained from different people are studied and examined.
  
  
• Response form: It refers to the style according to which responses would be represented on a scale.
  
  
• Degree of subjectivity: It refers to the development of scale either by measuring personal preferences or non-preference judgements. In the first case, respondents may be asked to select the person or solution they favour.
  
  For example:
  
  Which of the following organisations you favour the most?
  
  
  • A
  • B
  • C
  • D
    
    
    In the second case, the respondent may be asked to judge which organisation would be the most profitable. It is clear that in the second case, scope of personal opinion is not there.
    
    
• Scale properties: In this basis of scale classification, scales can be classified as nominal, ordinal, interval and ratio scales. These scales have already been discussed in previous sections
  
  
• Number of dimensions: Under this basis of scale classification, there are two types of scales – one-dimensional scales and multidimensional scales. In the one-dimensional scale, only one characteristic of the respondent or the object is considered. In one-dimensional scales, each dimension is a construct. In the multidimensional scale, an object can be described by using attribute space of ‘n’ dimensions rather than using a single dimension.
  
  
• Scale construction approach: It refers to the scale-classification on the basis of different approaches.

Let us discuss these approaches in the next section.

Techniques of Scale Development

Scales are used in almost all fields of research. However, it is used extensively in studies related to psychology and social sciences. In an attitude scale, different questions are framed and the response of the respondent is noted. In developing the question statements, care must be taken so that the questions elicit the responses that are psychologically related to attitude being measured. In addition, the question statements must be made in such a way that the slightest of individual differences are made visible.

While carrying out the measurement, the researcher must take into account the fact that respondents may conceal their true attitudes and spell out only the socially acceptable opinions. At times, respondents may be faced with a question about some incident or experience which they never have had. After recognising all these issues and limitations, psychologists and sociologists have developed five basic scale construction techniques. The researchers must know these techniques so that they can develop an appropriate scale for their study.

Let us now discuss these techniques as follows:

Judgement Methods

These methods usually require the responses of respondents that are based on their perception of the subject and their judgements. These include the following approaches:

Arbitrary Approach

In this approach, a scale is developed on an ad-hoc basis. An arbitrary scale is developed for a specific purpose; therefore, it cannot be generalised. Arbitrary scales are easy and inexpensive to develop. Arbitrary scales can also be developed in a manner that they are highly specific and adequate about a particular topic if designed by a person skilled in scale designing.

Consensus Approach

In this approach, items included in the scale are decided by a panel of experts who evaluate the items for attesting their relevance to the research topic, and the level of attitude the scale item represents.

Item Analysis Approach or Summated Scale Approach

In item analysis, the respondents’ responses to each individual item (question) in the test are segregated based on the discriminating power of the item. Such segregation helps in assessing the quality of each individual item and of the test as a whole. This approach is used to select items which can be reused in successive test or to select the items that should not be reused in successive tests. An example of a scale developed using item is summated scale.

Factor Analysis Approach

In this approach, a correlation between different items is established on the basis of a principal component. An example of a scale developed using the factor analysis approach is semantic differential scale.