The ultimate secondary research guide
Upgrade your secondary research with agile market research guide.
SHARE THE ARTICLE ON
Data coming from various sources can be either qualitative or quantitative. Qualitative data refers to the data that is descriptive and long in nature. Whereas quantitative data is numeric in nature. This quantitative data can be analysed through statistics. But how do you measure the data in statistics?
Well, we have a concept called “variable”, which is nothing but a property or characteristic that changes throughout the population. Example: when you are collecting data from students of a school, variables for that population can be name, age, address, contact, etc. these variables will be different for different students and they can be measured through qualitative or quantitative means.
Let’s say you gather their test scores too, you can calculate the ranks of the students in the school and assign each to a relevant programme. Similarly, “level of measurement” in statistics is referred to the method by which a particular variable is being measured.
Conducting exploratory research seems tricky but an effective guide can help.
The nominal scale is the first level of measurement. It is identified as named variables. On a nominal scale, the variables are given a descriptive name or label to represent their value. The catch here is that the variables are not arranged in the options as per their importance or value. The calculations on these variables are based on the understanding of the researcher regarding the topic. There is no statistical calculation possible on nominal data as there is no numeric value attached to the variables.
You may see cases where the labels to the variables are numbers, but that does not mean anything more than just labels for classification or bifurcation. The results from the calculations on these numbers will have no quantitative significance.
Which cuisine do you like the most?
As you see, the numbers against the variables indicate their label and not their numerical value. As the options don’t hold any priorities or higher value over each other, it wouldn’t make any difference which one the respondent chooses as long as it is just indicating which cuisine they like the most. Amongst the statistical measures, the nominal scale is known as the fundamental research scale.
Nominal data can be collected through open-ended questions, like asking open-ended questions to the respondents and letting them describe their answers in brief indicating one variable as an answer. Another way is using multiple-choice questions, like the one in the above example where the respondents will choose the option they prefer. The nominal data is analysed to get one or more than one option as the most preferred amongst the respondents.
The second level of variable measurement is an ordinal scale. The ordinal scale is used when variables are used to measure the extent or limit of something. These scales are used in questions that don’t have any mathematical background. Some of the examples can be measuring satisfaction, happiness, pain, etc.
As the name suggests, the ordinal scale is based on the order of the variables. The variables are presented in a particular order from low to high (of any particular feeling) and the respondents will choose the option they prefer.
How happy are you with our customer service?
As you can see, the variable here is “happy” and the ordinal scale goes ahead to manipulate that variable to a range where it ranges from high to low, with decreasing value. The trick most researchers use here is to deliberately put the positive side of the range first to somewhat influence the answers.
As the ordinal data follows a hierarchy and value-based variables, it is possible for the researchers to put the data in form of graphs and diagrams. These figures will help to visually draw out the responses and preferences.
There are tests that you can perform on ordinal data like the Mann-Whitney U test (compare the variables of one group to the other to see which one is bigger) and Kruskal–Wallis H test (compare two ordinal groups to see if they have the same median).
Our third level of statistical measurement is an interval scale. An interval scale is a numerical scale with the order and difference of variables known. As the name suggests, it is used to measure the intervals between the variables and these variables are computable, constant and familiar.
Just like the ordinal scale, the interval scale does not have a starting point or a true zero. You can call it an extended version of the ordinal scale with added functionality to measure the difference between the variables.
How much is your annual income?
Here, the difference between the variable is quite known and the start point isn’t zero either. With constant variable values, respondents can select their relevant income option.
Along with the nominal and ordinal data analysis methods, interval data has its own data analysis methods like descriptive statistics, correlation regression analysis.
They help you calculate the mean, median and mode and summarize the data in a way that makes sense.
The fourth level of statistical measurement allows you to know the variable difference, variable order as well as start with true zero. The ratio scale assumes that there is a fixed difference between the variables and they are in a particular order, starting with a zero value.
The ratio scale allows researchers to operate on the data by being able to find the mean, median and mode for a central tendency. Due to the true zero, the ratio scale cannot be zero. So when a researcher wishes to use a ratio scale, he must look at the properties of the variable and whether it is having all of it needed.
How much is your child’s age?
Ratio scale and ratio data come under quantitative data, which is why all the quantitative analysis techniques like SWOT, TURF, cross-tabulation will work o ratio data. Using these methods for data analysis can help you draw predictive conclusions regarding your organization and products.