The types of data analysis are things that must be known by a researcher. This stage is important because each study has unique characteristics and should be determined at the beginning of research.
The determination of analysis should be done at an early stage. It will be very complicated if you change it when the types of questions are ready or even the data has been collected.
So, you need to understand all the types. Do not worry, I’ll keep it simple and complete in the explanation below.
Let me show you the types of data analysis in research methodology that you should know before starting your study.
Data Analysis Definition
Data analysis is the next process in which a researcher will process data that has been obtained and collected using certain methods. Through the process of data analysis, a researcher will discover whether the data collection methods used can answer the objectives of the research.
Data analysis in qualitative research will certainly be different from quantitative research. This is due to differences in methodology and types of data used so that the methods used are also different.
However, in broad outline, the analysis data will make the data that has been collected to be structured, easy to understand, easy to process, and easy to interpret to the public.
In this process, you may do some data processing more than once because if you didn’t achieve your research purpose, you may use another data analysis until you get the right method.
Types of Data Analysis
Based on the purpose
1. Descriptive analysis
Descriptive analysis aims to describe the characteristics of the data and provide an overview of data conditions in general.
Descriptive analysis is usually only presented in the form of outputs such as tables, graphs, and some simple test statistics.
Some of the measurements often used in descriptive statistics are centering measures such as mean, median, mode.
Meanwhile, the size of the spread that is often used is variance, quartile, range, etc.
This analysis produces a simple output and merely provides information regarding the general conditions of the data or variables that we use.
2. Inferential analysis
The inferential analysis is an analysis that aims to conclude the population by testing and processing a group set of sample.
Inferential analysis can be said is a more complicated stage than descriptive analysis. This is because this analysis involves a variety of complex calculations and can produce more valid test statistics.
Examples of the use of inferential analysis are hypothesis testing, regression analysis, categorical data analysis, time series analysis, panel data analysis, etc.
It requires a lot of effort and mathematical processes to produce the result that you need. Also, there are many assumptions that you have to fulfill to make precision modeling.
Based on the number of variables
1. Univariate analysis
Univariate analysis is a type of analysis for 1 variable only.
In univariate analysis, we can apply descriptive analysis and inference analysis.
For example, in descriptive analysis, we can use various test statistics such as mean, median, mode, etc. to illustrate data characteristics.
Meanwhile, for inference analysis in one variable, we can use the Z hypothesis test, the Durbin-Watson test, etc.
2. Bivariate Analysis
Bivariate analysis is an analysis conducted to determine the relationship between 2 variables.
Because we have involved two variables, this test will feel more varied, diverse, and produce a variety of interesting conclusions.
Examples of bivariate analyses that are often used are simple linear regression analysis, correlation analysis, average difference test of two populations, etc.
3. Multivariate Analysis
Multivariate analysis is an analysis conducted to determine the relationship of more than 2 variables.
The number of variables used will provide a lot of choices that we can use according to the type of data we have.
Examples of multivariate analysis are multiple linear regression analysis, panel data analysis, etc.
Based on the estimation results
1. Datapoint analysis
Point estimation is a data analysis conducted at one point in time.
By using this analysis, a researcher can predict how to estimate an event in the future by using pre-existing data.
Many types of point analysis can be used by a researcher.
An example often used in point estimation is how to predict a particular variable using a trend.
2. Interval data analysis
Interval estimation is a data analysis carried out on a variable where the results use a confidence interval.
With this confidence interval, we can make predictions more accurately. This is because we can determine the level of error that is the tolerance limit in conducting statistical testing.
In many test points, researchers usually do interval estimates to get more varied results.
Based on data time
1. Analysis of time-series data
Time-series data analysis is an analysis of variables ordered by time, for example, weekly, monthly, yearly, etc.
Using this analysis, we will find out how a variable continues to change over time. We can also forecast the future and the past by using various more scientific formulas,
Usually, this approach is used when the data we use has patterns such as random, seasonal, cyclical, and trending.
Analysis of time series data is often used in solving various economic models.
2. Cross-section data analysis
Cross-section data analysis is an analysis that is used on various variables in one particular point of time. By using cross-section data, we can obtain information on how a variable affects other variables without the influence of time changes.
In the cross-section, usually we use more than one variable. The more variable and sample we use, the better model we will generate.
3. Panel data analysis
Panel data analysis is the analysis used with variables that are a combination of cross-section data and time series. Simply put, we will use the analysis of various variables together in a certain period simultaneously.
By using panel data, we will get the results of research that considers the effect of time and interrelationships between variables.
Based on the data form
1. Analysis of numerical data
Numerical data analysis is an analysis performed when the data used has a numeric type.
This analysis certainly makes it easier for a researcher because there is no need to carry out the classification and coding stages of the data used.
With numerical data, almost all types of statistical analysis can be done.
2. Categorical data analysis
Categorical data analysis is an analysis carried out on categorical or qualitative types of data.
With this analysis, a researcher who uses qualitative data can produce research using a quantitative approach.
Before conducting the analysis, a researcher must classify and coding the categorical data into numerical form so that it can be transformed into various formulas.
For example, you want to analyze the relationship between the level of education of the head of the family and the yearly income.
The level of education certainly varies from elementary school, junior high school, general high school, to university.
Before doing the test you can do the grouping as follows
- primary school: 1
- junior high school: 2
- public high school: 3
- universities: 4
By categorizing the data with the classification above, we can predict how the relationship between the level of education (which is categorical data) and the yearly income.
3. Text analysis
Text analysis data analysis is a breakthrough in the world of statistics and technology.
Text analysis allows one to analyze various unstructured information in the form of text that is spread on the internet into a more useful form of information.
With this analysis, a researcher can find out the point of view of internet users on certain things.
One analysis that is often used in text analysis is Twitter data processing. By processing the text data from Twitter with certain keywords, we can classify and classify various things according to the research objectives.
You will often find this text analysis especially when elections are held in an area.
If you want to use simple text analysis, you can use the Google Trend feature that has been provided by Google.
4. Spatial analysis
Spatial analysis is a data analysis that considers regional aspects such as topology, geography, etc.
By using spatial analysis, we can find out how the condition of a problem is not only reviewed based on existing data conditions but also considers the geographical conditions of the region.
An example of using spatial analysis is the mapping of population movement patterns based on the economic level of an area. Usually, the greater the economy of an area, the population will tend to move to the area to try to get a better life.
Based on statistical types
1. Parametric statistical analysis
Parametric statistical analysis is an analysis that aims to conduct inference and estimation of population parameters.
In parametric statistics, the data used must be able to meet various assumptions such as normality assumptions.
This assumption is very important so that the estimation and terms of the use of statistical tests can be met.
The type of data used in the parametric analysis is usually the interval or ratio scale.
Examples of parametric analysis: multiple linear regression analysis, panel data test, Z-test.
2. Nonparametric Statistical Analysis
Non-parametric statistical analysis is an analysis that does not aim at conducting inference and estimation of population parameters.
Typically, data on non-parametric statistical analyzes do not have a normal or unknown distribution.
The type of data used in the non-parametric analysis is usually nominal or ordinal periodic.
Examples of using non-parametric statistical tests
– One sample test: Chi-Square test, Binomial test, Kolmogorov Smirnov test, Run test
– Testing of two samples: the Mann-Whitney test, the Kolmogorov-Smirnoff-Z test,
– Testing of two non-free samples: Sign Test, McNemar Test, Wilcoxon Test
Best software to do data analysis
Well, there is a lot of software that you could use to do data analysis. Every software has its characteristic and we have to know which one the most suitable package to analyze the research.
Based on my perspective, here is the most favorite and popular software you could use for the statistical research method.
R is a programming language that is designed so that it is very friendly in statistical analysis and data visualization. With R, lots of data analysis can be used without even making a cent.
In the past 10 years, R has begun to develop among statisticians as a language that must be mastered to help solve various statistical models.
There are so many types of analysis that can be done with R. Almost all types of analysis that I mentioned above you can do, including making maps.
Python is a popular programming language used in the world of information technology. In the last 5 years, Python is growing and being used for various statistical calculations.
In comparison, Python and R have some similar things. Both require users to do simple coding in using various statistical functions.
3. Statistical Package Social Science (SPSS)
SPSS is a very popular statistical processing program. I am sure, almost all people who have come into contact with research will know this program.
SPSS is a software developed by IBM that has features to process, analyze, and present various data as needed.
The SPSS display which is very comfortable and user friendly makes many researchers love this software.
As the name implies, SPSS is suitable for use in analyzing social-themed research. If you hope to produce interesting data visualizations using this device, you should forget it
Eviews is a statistical software specifically used in various quantitative modeling.
Because it is designed for quantitative research, Eviews is less suitable for qualitative research.
Eviews are often used in econometric modelings such as time series and panel data. The advantage of eviews is the availability of various tests that are very complete, especially the conditions included in the economic model.
Pro tips choosing the right data analysis
1. Understanding the types of data
There are several types of data in the statistical process. You have to understand which the correct type your data is. Is it nominal, ordinal, interval, or ratio? Is it qualitative or quantitative?
Every type of data has its unique treatment. The nature of your data types will determine the right analysis.
2. Understanding the types of research
In general, there are two types of research. There is qualitative research and quantitative research. But, you have to be specific about your types of research.
Example, if you are choosing the qualitative study,
You must determine whether your research belongs to oral studies, focus group discussions, etc. If you choose quantitative research, you must ascertain whether your research will lead to the cohort, longitudinal studies, panel data, etc.
3. Knowing the goal of your research
At the beginning of your research, please decide what do you want to achieve? Are you only want to have a descriptive explanation? Are you want an inferential and numerical result? Or you want to prove that your hypothesis is correct?
Let me make a complete summary for you
Data analysis based on purpose: descriptive analysis and inferential analysis
Data analysis based on numbers of variables: univariate analysis, bivariate analysis, and multivariate analysis
Data analysis is based on the estimation results: data point analysis and interval data analysis
Types of data analysis based on data form: categorical data analysis, numerical data analysis, text analysis, spatial data analysis
Types of data analysis based on statistical types: parametric statistical analysis and non-parametric statistical analysis
The most fundamental thing that we have to remember is every type of data analysis has a unique purpose. Never compare it one to another.
Understanding your data, know your resources, and connect it with the research purpose, that’s how you will get the suitable data types of analysis.