From the spring scale hanging in the produce aisle to the pit-and-girder monsters that weigh train cars and tractor-trailers, modern life depends on weighing devices. While these machines vary in operational principles, they all measure force by flexing a load cell.
This article examines current practices in scale development, including the importance of convergent validity (the extent to which the new scale correlates with theoretically related constructs) and psychometric limitations reported by researchers.
Definition
A scale is a ratio that represents the relationship between a figure or object on a model and its actual size. It is mainly used in preparing blueprints and building construction.
A simple or plain scale consists of a line divided into a number of equal main parts and the first part is sub-divided into smaller parts. Each main part has a certain number of subdivisions and each sub-division is marked with a particular number. The numbers are given from zero to the last main part.
Scales are also useful in plotting graphs and other mathematical representations. They allow us to represent large real-world objects in a small space and to make them easier to interpret.
In addition, scales are commonly used to shrink vast land masses into maps and to help architects, machine-makers, and engineers work with models of complex machinery and structures that would be too large to hold if they were in their true form.
Types
There are four main types of scale: nominal, ordinal, interval, and ratio. Each type of scale has different characteristics that distinguish it from other types of measurement.
The interval scale classifies scales based on their interval pattern (for example, the difference in pitch between two adjacent notes). This method of classification allows for a wide variety of scales to be classified as pentatonic, diatonic, chromatic, major or minor, etc.
Ordinal scale is another category of measurement, which is characterized by its ability to rank and order data without establishing the degree of variation between items. Examples include the ranking of school students based on their grades, the ordering of objects in an experiment, and the comparative scaling technique commonly used by product researchers who ask customers to choose the preferred one of two closely related products.
A decimal balance is a type of weighing scale that uses a long arm for weights and a short arm for supports to measure the weight of an object, producing an accurate reading in tenths of a kilogram. METTLER TOLEDO offers a wide selection of decimal precision scales for a variety of applications.
Usage
In research, scales are used to measure and compare data. They can also be used to determine the relative importance of different attributes. For instance, a scale of preference can be used to measure how important a particular attribute is to a respondent.
Another usage of scale is in architectural modeling and construction. Different scales are used for various projects depending on their size and complexity. A scale model allows architects to accurately portray the overall size of a project and communicate the design with their clients and workers.
Scales are also used in music. In classical music, for example, the chords that are played based on the scale of the key that is being used. The scale is also useful for musicians to understand the structure and progression of a piece of music. For example, the classical piece “Angels We Have Heard on High” has descending runs that are based on the pentatonic scale.
Reliability
A scale of measurement makes it easier to collect consistent data from a group of individuals by using standardized, uniform response options. However, despite their obvious advantages over unreliable subjective measures like interviews and observation, scales must be carefully validated before use in research.
The first step in scale validation is to perform a factor analysis to ensure that each of the items has a clear and meaningful relationship with the construct being measured. Items that fail to show convergent validity should be dropped from the scale.
For longitudinal surveys, the simplex model is a good alternative to Cronbach’s alpha for estimating internal consistency reliability. The key difference is that the simplex method estimates reliability from the aggregated scale scores rather than the individual item responses. This approach means that inter-item correlated errors do not bias the estimate of reliability. However, if either the error variance or the true score variance vary by wave, the simplex estimator will be biased.