SCALE ENGINEERING

a. Introduction

In preparing research instruments, especially in quantitative research, researchers must know the scale techniques used in measurement and the type of scale so that the data obtained is in accordance with what is measured and is reliable for research problems. Another purpose of the scaling technique is to find out the characteristics of a data, which is based on a certain size, so that we can distinguish, classify, and even sort these characteristics.

b. Nature of Scale

Variable is a concept that has a variety of values, and is a changer. A researcher must know the concept of a variable in order to be able to measure or give the right value for the observed data. The level of variable value measurement can be divided into four scale levels, namely Nominal, Ordinal, Interval, and Ratio.

Measurement Scale Image

1. Nominal Scale

The nominal scale is the first and simplest level of measurement. At this scale there are no assumptions about the distance or order between the categories within the measure. This scale is just a differentiator from a category with other categories of a variable. The following is an overview of the nominal scale.

Image Nominal scale overview

The numbers given to objects are labels and it is not assumed that there are levels between one category and another category of one variable. With this nominal size level, researchers can classify respondents into two or more categories. Qualitative variables that are transformed into quantitative data in the form of nominal measurements are also called dummy variables. Data that only has differentiating characteristics, for example gender: male (value = 1) and female (value 2), color type, for example red (value = 1), white (value 2). Blood types such as A, B, AB and O.

2. Ordinal Scale

Ordinal scale is the next scale with a higher level than the nominal scale. This scale aims to distinguish between categories in one variable with the assumption that there is an order or scale level. Ordinal numbers show more ranking order from lower to higher or vice versa. These numbers do not represent an absolute quantity, nor do they provide any indication that the intervals between any two numbers are the same. The following is an illustration of an ordinal scale:

Ordinal scale drawing

For this variable an ordinal measure is usually used; top, middle, and bottom. This measure does not show the average economy class score, and does not provide information on the interval between the lower economy class and the upper economy class. Ordinal levels of measurement are widely used in social and economic research, especially to measure interests, attitudes, or perceptions. Through this measure, researchers can divide respondents into ranking order based on their attitudes towards certain objects or actions. For example, a student wants to measure the level of employee performance at Bank Syariah Mandiri (BSM). So by using an ordinal scale he can divide the sequences as follows: Score I for numbers 100-80, then Value II for numbers 79-75, then number III for values 74-60, and value IV for numbers less than 60. Note that the difference in numbers for grades I, II, III and IV are not the same, but can be sequential. It can also be based on social status, for example number 1 for the Poor category, number 2 for Simple, and number 3 for the Rich category. Other examples include: Measuring class rankings I, II, III, IV and so on; seniority level of employees and others. Note that on an ordinal scale, we cannot definitively say that a rich number (3) is twice as simple as a simple number (2).

3.Interval Scale

Interval scale is a variable scale that only has distinguishing properties, and is tiered and has intervals. Apart from being differentiated and having levels, it is also assumed to have a definite distance between one category and another in one variable. Attitude scales and indices usually produce interval measures. Because of this, this measure is one of the most frequently used measures in social and economic research. Here is an overview of the interval scale:

Image Interval scale overview

One of the popular interval scales is the Likert scale. This scale is divided into Very Dissatisfied (STP) with a weight of 1, Dissatisfied (TP) with a weight of 2, Fair (C) with a weight of 3, Satisfied (P) with a weight of 4, Very Satisfied (SP) with a weight of 5. All assessments are positive or good, for example Satisfied, Agree, Good, High and so on, is rated the highest, namely 4, while unfavorable or negative assessments start from a low value, for example 1.

Parametric and econometric statistical analysis can be applied to this type of data. An example is the age variable. One characteristic of values is the absence of zeros. Another example is the value of the student’s IQ score, units of time such as seconds, minutes, hours, days and so on. College exam: A = 4, B = 3, C = 2, and D = 1. So the interval values are: A with B is 4-3 = 1, and B with D = 3 -2 = 1 and so on. The IP number does not show the number of student numbers, but measures the order of student rankings.

4. Ratio Scale

Ratio scale is a variable scale which apart from being differentiated, has levels, and the distance between one value and another, it is also assumed that each variable value is measured from a state or the same point (has an absolute zero point). The ratio size is obtained in addition to information about the order and interval between respondents. The numbers on the scale indicate the actual magnitude of the property we are measuring. Because there is a zero point, the ratio comparison can be determined. With this absolute zero value, the value on the measuring scale is the actual amount of what is being measured, and therefore all mathematical operations can be applied to ratio sizes. The following is an illustration of the ratio scale:

0 1 2 3 4

Image Ratio scale overview

For example, if there is a ratio scale for achievement, then if a student who gets a score of 8 on that scale has an achievement that is twice as great as another student who scores 4.

Another example is the highest data order. Data that has distinguishing, tiered, and interval characteristics and has absolute zero and equidistant. For example, age with a value of 0, 1, 2, 3, 4 years and so on, height, land area, test scores, amount of Alms Alms Zakat collected, salary payments and so on.

Tabel Perbandingan Sifat Skala