Scale Types of Data Measurement
We started the Data Science Adventure in September with the 3rd Data Science Bootcamp organised by VBO! The first subject of this process was Data Literacy.
To summarize the information highlighted in the first week application;
Data Science: It is all of the processes and techniques related to the evaluation and interpretation of the examined data (revealing the useful and significant information!) for a purpose.
Data Scientist: The person who controls and directs the data science process. Accordingly, that's for sure, this person must also have some technical and personal skills in order to manage the process effectively.
So, “How can I become a data scientist?”. The answer to the question is hidden in the “Nasıl Veri Bilimci Olunur?” essay posted by Data Scientist Mustafa Vahit Keskin :) and “What is Data Science(DS) and How can it be learned?” essay posted by Data Scientist Mehmet A. :)
“Understanding the data, ie data literacy is an important step in revealing significant information from data.”
In this post, I would like to mention about the Scale Types, which is part of data literacy and the first subject of every Introduction to Statistics 101 lesson, with examples.
What is a Scale?
Scale
It is a data measurement method/way used to evaluate the values of a variable and to understand its structure.
The evaluation of the data with the correct scale type is quite important in the terms of executing the analyze method that tests the hypothesis to answer the research question. Hence, it is beneficial to mention about the data measurement Scale Types!
What are the types of Data Measurement Scales?
Scale Types of Quantitative and Qualitative Data
Nominal scale, categorize the properties of units that cannot be directly measured or sorted.
Ordinal scale, categorize the properties of units that cannot be directly measured but sorted sequentially.
Interval scale, where there is order, the difference between the two points is meaningful but no true zero point exists.
Ratio scale, where there is order, the difference between the two points is meaningful and a true zero point exists.
Nominal and ordinal scale structures are considered within the scope of qualitative data because they evaluate descriptive characteristics. Although it is examined in the qualitative data title, it is more correct to express the title of nominal scale and ordinal scale examined as Categorical Data.
Classically, “gender” is an example given for the nominal scale and the “education level” is an example given for the ordinal scale. These examples are ideal to describe the difference between the two scale types. Before explaining the difference of the two scale types through examples, we can say that their common point is that they are data structures that cannot be measured with a numerical measurement method naturally.
Considering the gender variable, let’s assume that male and female individuals constitute two separate categories. Is there a sequence relationship between these two groups? No. Can we calculate four operations between research participants’ information of being “Female” or “Male”? No. That is, there is no distance between groups of a variable indicated with nominal scaled data structure, which we can speak about the difference.
Considering the education level variable, there is a sequence relationship between the categories that high school, bachelor, master and PhD. An individual cannot study at bachelor degree without graduating from high school, or cannot bypass the bachelor after finishing high school to start the master degree. So, can four operations be calculated between research participants’ information of studying at “high school” or “PhD”? eg is PhD 3 times multiplied of high school? No. It is obvious that between the groups of this variable there is a distance at which the difference can be mentioned.
The scale types that the data structures can be measured with a numerical measurement method naturally, are interval scale and ratio scale. These are considered within the scope of quantitative data because a distribution feature can be observed. Again, in classically, an “temperature” is an example given for the interval scale and “weight” is an example given for the ratio scale. These examples are ideal to describe the difference between the two scale types. This difference between these scale types is the true zero point concept.
Considering the weight variable, are there 0 kilograms of people or 0 kilograms of apples existed? For these concepts, zero literally indicates the absence-nothingness. Because there is a real zero point on the Ratio scale. In the data structure measured with ratio scale, the actual amount ot the values can be indicated with a unit of measurement like in year for age, metric system for height and kilogram for weight. Accordingly ratio scale allows to calculate four operations and all quantitative attributes.
Considering the temperature variable, temperature is a measurable concept, but it is relative. For this concept, zero literally does not indicate the absence-nothingness. When Celsius scale is 0°, Fahrenheit scale measures 32°. Also 0° of the temperature does not describe nothingness, you can still feel a temperature of the air. Zero in this scale is more likely a a number to describe a point to define the distance between intervals on the scale, instead of a starting point or nothingness. Accordingly interval scale allows to run all quantitative attributes such as rank, count, subtract, or add, and equal intervals separate each point on the scale.
The Scale Types of Data Measurement gains a greater meaning at the data analysis stage because the data you have must show the data structure required by the data analysis method to be used to test hypothesis. Therefore, it is important to design the research process well if the data has not yet been collected.
Check out! the Research Methods Strikes Back story series for a better research plan :)
