15 Scaling and Measurement

C. Raghava Reddy

epgp books

 

1. Introduction

 

By this time you must be familiar that arriving at explanations or generalizable statements about social phenomena is a difficult and complex proposition. Quite often we come across general universal laws in physics, chemistry and biology (for example, laws of motion, law of gravity, laws on chemical bonding, law of inheritance, respectively) which are taught unequivocally to the younger generations. However, when we look at social science literature the results of attempts at understanding complex web of social relations are stated indeterminately (of course all scientific knowledge is indeterminate and problematic and stands valid until it is disproved). Notwithstanding the inherent limitations imposed by the social world, social scientists have been able to arrive at certain procedures which stood the test of time in reporting social phenomena in a scientific manner.

 

In the previous modules of this paper, you have studied different methods of data collection. The survey method, as an important method of quantitative research, employs specific instruments of data collection such as questionnaire and interview schedule. Data in survey research pertain to a number (whose number ranges from a minimum of 30 to any number beyond 100) of units (units could be individuals, households, villages, districts and organizations). Based on the data collected using a questionnaire or interview schedule, survey researchers arrive at aggregate observations. Aggregate observations are made in statistical terms or in simple numerical forms. For example, percentage of male and female population in a village, average age of respondents, etc. Social researchers also often talk about attitudes, opinions, perceptions of a set of respondents in aggregate in terms of agreement or disagreement, favourableness or unfavourableness. The intensity of attitudes or opinions can also be expressed in the form of responses like strongly agree or strongly disagree. Beyond the level of textual expression of intensity (on a continuum of strongly agree to strongly disagree) it is difficult to quantify.

 

However, scales of measurement help researchers to express the intensity or magnitude of opinion or attitude in quantitative terms. The intensity of attitude can be measured at individual level as well as at aggregate level. For example, attitude of young voters on corruption in public offices can be expressed in certain numeric forms for the individual respondent as well as for a group of respondents. When the individual responses are summated for a set of respondents, the intensity of response could be stated in aggregate form. For example, if young urban voters’ (between 18-20 years of age) attitude towards corruption in public offices is found to be 5.7 (on a continuum of 1 to 7 where 1 being strongly favourable and 7 being strongly unfavourable towards corruption in public offices) it indicates a moderate to strong unfavourableness towards corruption in public offices.

 

In this module the importance of scaling, meaning of scales, types of scaling, and different levels of measurement are discussed. At the end of this module you will be familiarized with the relevance of scaling in social research and also the different levels of measurement of data.

 

2. Importance of Scaling

 

In our daily life we rely on different types of standardized tools of measurement. For example, a petrol dispensing machine in a petrol bunk measures the volume of petrol we ask for. A shop keeper uses a weighing machine to measure the required quantity. In the case of measuring temperature, a thermometer is used which is believed to be a standardized instrument. Similarly height, distance and a whole lot of other things which are part and parcel of our daily life are measured using standardized instruments.

 

Unlike the ones mentioned above, measuring social behaviour (attitudes, perceptions, preferences, or opinion) is difficult for its complexity. It is complex because social behaviour cannot be precisely explained or reduced to certain causal factors. For example, preferences of voters towards candidates in elections are conditioned by several factors like voters’ awareness levels, socio-economic background of the voter, profile of contesting candidates, parties and so on. Hence, it is difficult to single out a factor for preference of a particular candidate by a specific voter.

 

However, through an intensive and scientific exercise, researcher may be able to explain the attitude of voters in a particular constituency towards contesting candidates. This is possible through a set of statements which is carefully worded and ordered in a scientific manner. Scaling, therefore, helps in assessing the attitudes or opinions of respondents with a higher degree of accuracy.

 

3. Meaning of Scaling

 

The process of arriving at a set of statements to measure attitude, opinion or perception is known as scaling. Scaling may be defined as the ‘arrangement of objects to numbers according to a rule’. Here the objects refer to textual statements concerning attitudes or opinion. The reason for the use of a set of statements is simple. For example, if a single statement like ‘do you want a women contestant to be elected as sarpanch of your village?’ the respondent may say yes, no, or prefer not to respond. At the most, the responses can be categorized further on agree-disagree continuum as strongly agree, agree, disagree, strongly disagree, cannot say and assess the intensity of opinion. However, this single statement may not reflect the opinion of the respondents with a greater accuracy. In order to know the intensity of attitudes of the respondents, not just about a woman sarpanch, as in the example, but his/her opinion in general about women as elected representatives, we may use a set of carefully selected and ordered statements. Scale also helps researchers to express the intensity or magnitude of attitudes or opinions in numerical form. In other words, scaling refers to assignment of objects to numbers according to a rule.

 

The following figure illustrates the logical scientific ordering of statements.

 

1.      Do you favour women contesting elections?

2.      Do you vote for a woman candidate in Sarpanch elections?

3.      Do you accept your wife/daughter contesting elections for the position of Sarpanch of your village?

 

These statements are ordered in such a way that a respondent who agrees with the third statement is inferred to be in agreement with the rest two statements preceding it. Thus, the respondent’s score on a three point scale is 3 out of 3, indicating a strong favourableness towards woman participation in elections. When the same scale is administered to 100 respondents the attitude of the particular group of respondents may also be expressed in numerical figure. For example, if out of 100 respondents 40 score 3 points each, 30 score 2 points each and 30 score 1 point each, in all the score for the group is 210 (40 X 3 = 120; 30 X 2 = 60; 30 X 1 = 30). Hence the average value is 2.1 indicating a moderate favourableness towards women’s participation in politics. Similarly opinions or attitudes on complex social issues may be assessed using a scale.

 

The preceding discussion clearly suggests that a scale contains a set of questions (also called as statements or items). As mentioned earlier, attitudes or opinions are complex and measuring them using a single question or statement poses difficulty in assessment. If the objective is to measure the class status of respondents and a statement is made as ‘state your annual income’, it would yield responses which may help the researcher to make an assessment of economic status of the respondents and place the respondent in the appropriate class category. However, if the researcher wants to know the class, surely, income alone would not be sufficient to make the assessment. Hence a set of statements ranging from annual income, education, occupation to household expenditure may yield near sufficient information to place the respondent in the appropriate class category.

 

This leads us to important aspect of scaling, i.e. dimensionality. The first statement refers to measuring a respondent’s economic status on a single dimension, and the latter example refers to measurement on multiple dimensions. Attitudes and opinions are the outcomes of a complex web of social factors which cannot be assessed through single statements. This calls for employing a scale containing multiple statements. Scale also helps to test a hypothesis. For example, the hypothesis on the relationship between class and attitude towards women in politics can be tested using a scale developed to elicit the attitudes across class categories. Here you may find an interesting quotation on scientific social surveys which gives you a glimpse of the importance of scales in social science research. Referring to advancement of scales in social science research, McNemar (1946: 300)

 

…‘(T)he early efforts at attitude or opinion measuring usually involved a questionnaire or battery of questions which were selected on an a priori basis. Numerical values were assigned arbitrarily to the items or question responses and these values were summed to secure scores which were then interpreted as indicating the attitude of the respondents. There was nothing about the procedure to guarantee that any one item tapped the same attitudes as the other items’.

 

Later on, the attitude scales were evolved with the contributions from social scientists as well as psychologists. Since scaling was first applied in 1930s in the west, sociologists and social psychologists have developed hundreds of scales to measure attitudes and to assess the latent dimensions of human personality (Corbetta 2003).

 

A scale consists of a set of statements (statements are also referred to as items) which are, in most cases expressed in the form of questions. The items in a scale are arranged in an order decided by the researcher following the established procedure of scaling. In other words, it suggests that the items in a scale are not placed arbitrarily.

 

3.1. Response scale

 

The responses on each item are collected using a response scale. A response scale is different from scale for the fact that while the response scale is used to help the respondent to choose one among many possible responses on a particular question, a scale measures the attitude or opinions on the issue as a whole.

 

The response scale refers to the alternatives provided for a given question in the questionnaire or schedule. The response categories in the response scale do not by themselves measure the attitude or opinion of respondent. But it is used to allow the respondent to choose one among the given alternatives based on the degree or intensity of his/her response.

 

A closed-ended question has a set of response alternatives presented to the respondent in an order. The order of response alternatives could be either in increasing or decreasing order. The response alternatives could be given in three ways.

 

1.  The response alternatives are autonomous in their meanings. They indicate complete meaning without overlapping into the next. The response alternatives are mutually exclusive. For example, response alternatives on residential status could be urban, semi-urban or rural.

 

2.  The response alternatives could also be partially autonomous. For example very much, somewhat, a little, not at all; or agree strongly, agree somewhat, neither agree nor disagree, disagree somewhat, disagree strongly.

 

3.  The third kind of response alternative could be used to measure the intensity of response on a continuum moving from left to right or right to left. The intensity could be quantified in numerical terms (for example, 1 to 7 or 1 to 10, etc.).

 

Self Check Exercise – 1

 

a.       What is the significance of scaling in social research?

 

Unlike physical properties which can be measured without much difficulty individual’s opinions and attitudes are difficult to measure. We measure the properties of objects (for example, weight, distance, etc.) using appropriate instruments and express the property in numbers to indicate the extent or intensity of the property. However, individual’s attitudes or opinions are difficult to be measured and expressed in numbers. It is believed that expression of property in numbers has greater validity when compared to description in words. Social scientists have been attempting at describing the attitudes and opinions in quantitative terms using scales. By expressing the attitudes and opinions in quantitative terms it is possible to know the intensity and make comparisons.

 

b.      What is scaling?

 

Scaling may be defined as the arrangement of a set of textual statements to numbers according to a rule.

 

4. Types of Scaling

 

Over a period of time, different researchers have developed a variety of scales of which the following three have been popular for their validity and reliability.

 

1.      Thurstone scaling

2.      Likert scaling

3.      Guttman scaling

4.1. Thurstone Scaling:

 

It is the most widely used method of scaling. Its procedure is used to develop scale items to measure attitudes. L. L. Thurstone developed the scale in 1920s. It is also known as equal-appearing scale, and judgment scale. It is a uni-dimensional scale for it measures aggregate attitude towards a specific issue.

 

‘Thurstone attempted to device a method that would represent the attitudes of a group on specific issue in the form of a frequency distribution, the baseline indicating the whole range of attitude gradation from the most favourable at one end to the least favourable at the other with a neutral zero in between’ (Thurstone L. L. and Chave E. J. 1929 cited in Young 1979: 352).

 

 

Procedure in constructing Thurstone scale

 

1.      Focusing on the objectives of the research, a number of statements are prepared. These statements are reflective of attitudes about the issue that is studied. Statements cover an entire range of attitudes including positive, negative and neutral. They may range from extremely favourable attitude towards an issue to extremely unfavourable attitude. Neutral statements also form part of the set of statements. The researchers are encouraged to develop as many statements as possible (more than 100). Important point to be considered is that the statements should be brief, unambiguous and relevant.

 

2.      After a careful editing, the statements are numbered and cut into uniform slips of paper. A large number of judges who are quite familiar with the issue being researched are selected. The judges are asked to sort the complete set of statements into piles. There could be 7 or 11 or whatever the number researcher devices. These piles, for example ranging from 1 to 7 represent attitude ranging from extremely favourable to extremely unfavourable. Each judge based on his/her discretion places each slip of paper containing a statement into one of the piles. For instance, a statement which is reflective of a favourable attitude is placed on 1, 2, or 3 piles (1 being extremely favourable). The judges are not supposed to express their opinion but arrange the statements as objectively as possible on the scale which is nothing but a continuum from extremely favourable to extremely unfavourable. This is considered as the most important step in the construction of Thurstone scale.

 

3.      After sorting the statements by judges, the researcher evolves a table to determine the number of time each statement is included in several piles.

 

4.      Then the scale value for each statement is determined by computing the Median and Interquartile Range (Median is the value above and below which 50 percent of the statements in a specific pile fall).

 

5.      The next step concerns with the selection of certain statements for the final scale. One statement for each Median category with low Interquartile range is selected. On a scale of 11 as many statements are selected and arranged arbitrarily. Thus, researcher arrives at the final formal scale to measure the attitude of respondents ranging from extremely favourable attitude to extremely unfavourable attitude towards an issue (see, http://www. socialresearchmethods.net/kb/scalthur.php for an example on attitude of respondents towards persons infected with HIV/AIDS).

 

6.      Finally, the scale is administered to the respondents. Based on their response, i.e. in agreement or disagreement with the statements, a value for each respondent is obtained. For instance, a value of 8.5 indicates unfavourable attitude of the respondent towards persons with HIV/AIDS.

 

4.2. The Likert Scaling

 

Likert scaling was developed by Rensis Likert in 1932. This is also referred to as the ‘method of summated ratings’ because a respondent’s score is computed by summing the response values. It is like marks obtained by a student in a test. It is a uni-dimensional scaling method. It is the most widely used scaling technique to assess attitudes. Likert scale needs a minimum of two categories, such as agree and disagree. To know the degree of intensity it may further be divided into five or seven point response scale.

 

Procedure in developing the Likert scale

 

1.      Similar to the procedure you have studied in the generation of Thurstone scale, Likert scale construction also begins with generating a number of statements relevant to the issue to be researched.

 

2.      A group of judges are asked to rate the items, usually on a 1-5 rating scale where 1 reflects strong favourableness to the concept where as 5 reflects strong unfavourableness to the concept.

 

3.      In the next step, through calculation of correlation between the statement and the total score (summated), a set of statements is selected. Those statements having high correlation with the total score are selected. Thus, the Likert scale is developed.

 

4.      The scale is administered to the respondents who are asked to rate their responses on 1-5 scale. 1 indicates strong disagreement and moves further to 5 which indicate a strong agreement.

 

5.      The respondent’s attitude towards the concept of study is assessed by looking at the total score for the scale. The total score for the respondent on the scale is the sum ratings for all the items (statements).

 

Using the Likert scale, Edward O. Laumann (1965: 26-36) attempted to test a) similar status hypothesis and b) higher status hypothesis. He collected data from 450 respondents belonging to high, middle and low class categories located in three towns in the US. Using the 5 point rating scale, respondents were asked to give their preferences for different occupational categories in terms of their agreement or disagreement. All the responses were summated according to the class category and Laumann tested the hypotheses.

 

4.3. Guttman Scaling

 

Guttman scaling, also known as ‘cumulative scaling’ or ‘scalogram analysis’ comprises a set of statements or items that are arranged in an order, in the fashion of a flight of steps. It suggests that an affirmative response to any given statement implies an affirmation of the preceding statements. Corbetta (2003) explains the cumulative nature of the scale items through an illustration of ‘social distance scale’ of Bogardus. Bogardus’ distance scale designed to assess the respondents’ degree of prejudice towards ethnic minorities, utilised a sequence of questions. The questions were; would you be willing to accept a black person as visitor to your country? Would you be willing to have a black person living next door to you? Would you be willing to make friendship with a black person? Would you be willing to marry a black person?’ A respondent who is willing to marry a black person is presumed to be accepting to have a black friend, neighbour and visitor.

 

Guttman’s scalogram collects response categories as yes/no, agree/disagree for each item. The affirmative responses are given 1 and 0 for negative response. The respondent’s score is obtained by summing up the score for all the items (read, for details, Eckhardt, Kenneth W and Davis M. Erman 1977).

 

As in the case of other scaling procedures, the construction of Guttmann’s scaling involves different phases. In the first phase, items in the form of questions are generated based on the concept under study. Each item must be given two options – in affirmative (agree) and negative (disagree). Items must cover the whole range of the underlying attitude continuum. In other words, items must move from simple agreement to overwhelming agreement. A group of judges rate the statements in terms of their favourableness or unfavourableness towards the concept. They then review the items and select final scale items. Corbetta (2003) suggests that ‘in constructing a scale to gauge the progressive-conservative attitude in politics, the researcher will not only have to draw up a series of statements covering various fields but will also have to construct statements that cover the whole spectrum from extreme radical to extreme reactionary’.

 

The next phase is administering phase wherein the scale, consisting of a set of statements, is administered to the respondents. The respondents are asked to check items with which they agree. The analysis phase involves analysis of responses to statements. Each scale item (i.e. statement) is assigned a scale value and the attitude of the respondent is computed by summing up the scale values of each item with which the respondent is in agreement with [read Eckhardt, Kenneth W. and M. David Erman 1977, for a detail account with an example].

 

5. Measurement

 

Along with theory and method, equally important aspect of social research is data collection. Empirical studies involve data collection using either qualitative or quantitative method or sometime both. Data collection exercise using survey method aims at recording the properties of objects with the help of a questionnaire or an interview schedule. Each question in the questionnaire or schedule is aimed at measuring the properties of objects of research. Here objects refer to respondents and properties refer to the characteristics, opinions, attitudes of respondents. For example, age is a property which is measured in terms of years. Similarly sex is a property measured as female, male or others.

 

The properties of objects are measured using variables derived from concepts. Some properties are fixed and some are varying in intensity. For example, property of marital status is either married, widowed, divorced, or unmarried. It can only be one status. However, there are certain properties which can be measured in terms of their intensity. Thus, the term variable is used to refer to object properties conceived as varying in quantity or magnitude. For example, an individual can have more or less income.

 

Properties are measured at four levels, namely, nominal, ordinal, interval and ratio levels. The process of measurement involves recording the values assigned to the property of objects. Values may be numerical, alphanumerical or alphabetical. Those properties which are fixed are referred to as attributes whereas those which vary are called as variables. For example, sex is an attribute while income is a variable. Let us consider the characteristics of different levels of measurement and the guiding factors in choosing a particular level of measurement.

 

Self Check Exercise – 2

 

c.       What is measured in social research?

 

Social scientists describe the social, cultural, economic and political aspects of group or community being studied. The description of the group or community contains details about various characteristics. Using certain concepts researchers collect data from individual respondents and describe the features of the group or community. Concepts like gender, family size, educational and income levels, etc. are used to measure the property of objects (respondents) through a set of questions. Some properties are fixed while some are varying in nature. Fixed properties are called as attributes whereas those which vary are called as variables. For example, property of marital status is either, married, widowed, divorced, or unmarried. It can only be one status. However, there are certain properties which can be measured in terms of their intensity. Thus, the term variable is used to refer to properties conceived as varying in quantity or magnitude. For example, an individual can have more or less income.

 

d.      What are the different levels of measurement?

 

Properties are measured at four levels, namely, nominal, ordinal, interval and ratio levels.

 

6.  Levels of Measurement

 

6.1. Nominal: The nominal level of measurement is the most basic one. In the construction of nominal scales, the objects with similar properties are placed in one class. It classifies objects into categories which are mutually exclusive. It has a minimum of two classes. All the objects must be placed in any one of these classes. If an object cannot be placed in any class, it indicates that the nominal scale is incomplete. There is no order implied. It involves counting of the frequency of the cases and thus mode is the suitable statistical measure.

 

In nominal level of measurement, the values used are merely labels. These labels may be alphabetic, alphanumeric or numeric. For example, rural or urban options for a question on place of residence may be used as labels. Sometimes these two options are assigned numbers as 1 and 2, say rural 1 and urban 2. Although numbers are used in some situations they carry no significance and act merely as labels. The number labels should not be manipulated by adding, subtracting, multiplying or dividing.

 

6.2. Ordinal: In the ordinal level of measurement, the properties of the objects can be rank-ordered. In other words, the responses in terms of preferences, choices are ranked along the continuum of the characteristic being scaled. For example, students’ preferences for universities in the country can be collected on a rank order of 1 to 5, 1 being the more preferred university to 5 being the least preferred university. Here researcher will be able to know the order of preferences but don’t know about the intensity of such preference. The preferences along the continuum of 1 to 5 are apart by 1 but it doesn’t mean that the difference between them is 1. Thus, the numerical 1 has no arithmetic significance. For example, we cannot say that difference between the choice of university X which is ranked as 1 and university Z ranked as 4, though 3, is not 3 in arithmetic sense.

 

Similar to that of nominal scale, objects which have the same quantity are placed in the same class. In the construction of ordinal scale objects are compared to a common property (examples of social status, power, or preference). The rule of asymmetry applies here. In other words, the objects are ranked in terms of their properties. If object A has more of a property than object B, then object B cannot have more of the property than object A. The order of objects cannot be reversed. The rule of transitivity is also applicable in ordinal ranking. It suggests that if object A has more of a property than object B and if object B has more of a property than object C then object A has more of the property than object C. This rank order relationship persists through the scale (Eckhardt and David Erman 1977).

 

Mostly objects are ranked in terms of numbers for their possession of a common property. For example, number 1 represents the object which possesses more of a property than the other objects. Similarly at number 2 and so on. Thus, all the objects are ordered. Although numbers are used to rank the objects in terms of their possession of property, they do not signify any arithmetic meaning. For example, if object A is ranked 1 and object B 2, it doesn’t mean that the difference between these two is exactly 1. It only implies the more or less of a property, but doesn’t explain the intensity or magnitude of the difference. Consider another example here. If a shopping mall seeks opinion from its customers on its services the responses could be rated as 4 = very satisfied, 3 = satisfied, 2 = dissatisfied, and 1 = very dissatisfied. If someone responds very satisfied, coded as 4, indicates that the respondent is more satisfied than someone who responds dissatisfied, coded as 2. However, we cannot surmise that the person responding with a 4 is not twice as satisfied as the person responding with a 2. Thus, we can only count the number of customers across four categories (www. sagepub.com/upm-data/45955_ chapter_4.pdf)

 

Thus, the intensity or magnitude of differences between objects cannot be explained using ordinal scale. Positional statistics such as median, quartile can be determined using ordinal data. And order correlation can be determined with ranked data using Spearman’s Rho.

 

6.3. Interval: This has the advantage of specifying the degree of difference between objects. The interval level of measurement suggests that the distance between the ranked objects has some meaning. The interval between the ranked preferences is equal. Interval level of measurement not only tells the order of objects but also the distance between them. This is possible because of the utility of numbers used here. The numbers are used in such a manner that they imply the extent of interval between order indicating the degree or magnitude of difference between different object. Thus, the difference between 1 and 2 is considered to be equal to the difference between 2 and 3 and so forth. It is possible to add, subtract a constant to the scale values without affecting the form of the scale. For example, if the scale values for performance in a general knowledge test for students are 4, 8, 12, 16 and 20 (arranged in the progressive order) it may be surmised that the respondent X who scores 16 is said to have scored 8 points more than the respondent Y who scores 8. However, the basic limitation of interval scale is that it has no true zero point. Hence, we cannot infer that the respondent X is twice knowledgeable than Y because of the fact that knowledge has no true zero point. Other example of this type is ranking nations on human development index. In interval level of measurement the zero is an arbitrarily selected point.

 

6.4. Ratio: It is the highest level of measurement. It has a true or absolute zero that is meaningful. This means that we can subject this to arithmetic calculations like multiplication and division. For example, if two cities A and B are located at a distance of 200 and 400 kilometres respectively from the city X we can say that the distance between X and B is twice that of A. This can also be expressed in the form of an equation, B = 2A. We are able to express this in the form of an equation because distance has a true zero. Ratio level of measurement is amenable for many statistical operations.

 

Figure 1 attempts to provide a comparative picture of different levels of measurement and their characteristics:

 

Figure 1: Levels of measurement and their characteristics

 

Level of Characteristics Examples Permissible statistics
measurement
Nominal Classification Male/female; Frequency distribution,
No ordering rural/urban Mode
No true zero
Ordinal Classification Preferences for Frequency distribution,
Order universities; caste Median, Interquartile range,
No interval Chi-square
No true zero
Interval Classification Temperature, Frequency distribution,
Order intelligence Mean, standard deviation,
Interval is known correlation, regression
No true zero
Ratio Classification Distance, age Frequency distribution,
Order Correlation, and many
Known interval other statistical operations
True zero

 

Self Check Exercise – 3

 

e.       What is the importance of understanding different of levels of measurement in social research?

 

Research is a systematic inquiry to understand patterns of occurrence or relationship between various aspects of natural and social phenomena. Causal explanations about phenomena have greater relevance in proper understanding and prediction. Thus researchers aim at measuring the factors contributing to an event or a phenomenon. Using concepts researchers attempt to measure the property of objects (in social science objects refer to respondents or primary units of analysis). All the properties of objects are measurable. However, the levels of measurement vary depending up on the property that is measured. For instance gender is measured by nominal scale whereas income is measured on a ratio scale. Individual preferences or choices can be measured at ordinal level while beauty or intelligence can be measured using interval scale. Researcher has to have an idea of the level at which each property is to be measured before s/he embarks on primary data collection. This is important because once properties are measured using one scale cannot be subjected to a higher level of measurement. For example, if the researcher collects data on income from respondents using ordinal scale (different categories such as high income, middle income, low income) cannot explain the intensity of difference in income levels between the respondents. Researcher can only make a statement like x % of respondents belong to high income y % belong to middle income and z % belong to low income. But explanation like k % of respondents have double the income of l % of respondents cannot be made.

 

7.  Summary

 

Scaling forms an important part of social science research for its role in drawing inferences about any study as the researcher is able to explain the aggregate attitude or perceptions in a scientific manner. Attitudes, perceptions, opinions considered to be difficult for empirical observation and understanding are analyzed scientifically through scaling. The issues discussed in the module sought to provide insights into the importance of scaling and the procedures involved in scaling. Discussion on different types of scaling suggests the nature and appropriate use of these by the researchers. Measurement of responses in terms of attitudes, opinions, perceptions on the issues related to research takes place at four levels. Understanding of these becomes important for researchers as it forms basis for analysis of data. Researcher’s familiarity with different levels of measurement helps him/her in the preparation of questionnaire or interview schedule.

you can view video on Scaling and Measurement
  1. References