New assignment

An Introduction to Levels of

Measurement

Video Title: An Introduction to Levels of Measurement

Originally Published: 2017

Publishing Company: SAGE Publications Ltd.

City: London, United Kingdom

ISBN: 9781473979833

DOI: https://dx.doi.org/10.4135/9781473979833

(c) SAGE Publications Ltd., 2017

This PDF has been generated from SAGE Research Methods.

[MUSIC PLAYING]

DR. ELBERT P. ALMAZAN: Hello. My name is Doctor Elbert P. Almazan, and I am an Associate

Professor of Sociology at Central Michigan University at Mount Pleasant, Michigan, in the United

States. In this presentation, I will discuss the topic of levels of measurement in social statistics. Levels

of measurement can also be referred to as scales of measurement. Levels of measurement influence

what statistics methods will be used in analyzing relationships between variables.

DR. ELBERT P. ALMAZAN [continued]: I will be covering the following points-- coding, levels of

measurement, statistics tests. Social statistics is the study of variation in the social world. Variables

in the social sciences are social characteristics that have variations, such as gender, race, age, and

income. Levels of measurement refers to specific types of variation in the measurement of variables.

DR. ELBERT P. ALMAZAN [continued]: Variables have two or more categories. For example, if we're

using a binary definition of gender, we have two categories of gender-- female and male. Coding is

assigning numbers to categories to variables, which acknowledges that variation exists in variables.

All variables are coded. If we have age as a variable, each age has a value and a category.

DR. ELBERT P. ALMAZAN [continued]: Notice that the value and category are the same. If variables

represent variation in responses to survey questions from survey participants, survey responses

would be coded and would be displayed as numbers in a database. For example, in a database of

responses to survey questions on gender and age, Person 1 is male, because the value 1 refers to

the male category and the gender variable.

DR. ELBERT P. ALMAZAN [continued]: According to the age variable, Person 1 is 19 years old.

Person 2 is a female, because the value 2 refers to the female category and the gender variable.

According to the age variable, Person 2 is 20 years old. Person 3 is female, because the value 2

refers to the female category and the gender variable. According to the age variable, person 3 is 21

years old.

DR. ELBERT P. ALMAZAN [continued]: The database of numbers could be uploaded in a statistics

program such as SPSS, Stata, and SAS. And, with database, statistics functions would be performed.

Four specific types of levels of measurement are used in social statistics. Coding in the variables are

used to determine the levels of measurement.

DR. ELBERT P. ALMAZAN [continued]: Two levels of measurement have variable codings in which

numbers are meaningless. They are nominal and ordinal. Two levels of measurement have variable

coding in which numbers are meaningful. They are interval and ratio. An example of a variable with

a nominal level of measurement would be gender. Nominal variables have variation in categories of

social characteristics in no particular order.

DR. ELBERT P. ALMAZAN [continued]: The numbers in coding are meaningless. An example of a

variable with an ordinal level measurement would be military rank. A Master Sergeant in the US

Air Force would be higher than a Technical Sergeant. Ordinal variables have order in their variation

in categories of social characteristics. The numbers in coding are meaningless. An example of a

SAGE

2017 SAGE Publications, Ltd. All Rights Reserved.

SAGE Research Methods Video

Page 2 of 3 An Introduction to Levels of Measurement

variable with an interval level of measurement would be a political view scale, in which survey

respondents rate themselves from 1 to 7, with 1 as extremely conservative, 5 as moderates, and 7 as

extremely liberal.

DR. ELBERT P. ALMAZAN [continued]: The numbers in coding are meaningful in interval variables.

Yet, interval variables do not have a meaningful starting or ending zero point. An example of a

variable where the ratio level of measurement would be income, in which survey respondents give

their exact income. The numbers in coding are meaningful and ratio variables. The starting or ending

zero points of ratio variables are meaningful.

DR. ELBERT P. ALMAZAN [continued]: Levels of measurement influence what statistics methods

would be used in analyzing relationships. For example, in examining the relationship between two

variables, one variable is the independent variable, which is the cause variable. And one variable

is the dependent variable, which is usually the outcome variable. The levels of measurement in the

independent variable and the variable influence what statistics method is to be used.

DR. ELBERT P. ALMAZAN [continued]: For example, my independent variable would be gender,

and my dependent variable would be income. Gender is a nominal variable and income is a ratio

variable. I'm interested to know whether gender affects income. Because income is a ratio variable,

I can compute the mean of women's income and the mean of men's income and compare those

means.

DR. ELBERT P. ALMAZAN [continued]: With interval and ratio variables, the numbering and coding

is meaningful, so I can compute some statistics, such as means and standard deviations. A t-test

is a statistical method that compares the means between two categories of an interval dependent

variable, or a ratio dependent variable.

DR. ELBERT P. ALMAZAN [continued]: So, I can conduct a t-test on the relationship between

gender and income. In another example of examining the relationship between two variables, the

independent variable is gender and the dependent variable is color preference. Gender is a nominal

variable and color preference is a nominal variable. I am interested in knowing whether gender affects

color preference.

DR. ELBERT P. ALMAZAN [continued]: Because color preference is not an interval or ratio variable,

I cannot compute means or standard deviations, because the numbering and codings in nominal

variables are not meaningful. But, I can compute frequencies in categories of nominal or interval

variables. A chi-square test is a statistical method that compares the frequencies between two or

more categories of a nominal dependent variable or an ordinal dependent variable.

DR. ELBERT P. ALMAZAN [continued]: So, I conduct a chi-square test on the relationship between

gender and income. In summary, levels of measurement refers to specific types of variation in the

measurement of variables. The four levels of measurement are nominal, ordinal, interval, and ratio.

DR. ELBERT P. ALMAZAN [continued]: Levels of measurement influence what statistical methods

would be used in analyzing relationships between variables. [MUSIC PLAYING

SAGE

2017 SAGE Publications, Ltd. All Rights Reserved.

SAGE Research Methods Video

Page 3 of 3 An Introduction to Levels of Measurement