Univariate analysis is an analysis used on one variable with the aim of finding out and identifying the characteristics of the variable. This analysis is the most basic analysis technique that is often used in various types of research.

Because only one variable is analyzed, the results of the univariate analysis cannot and should not be concluded with other variables.

This analysis is often equated with descriptive analysis because it only gives a description of one variable without the intervention of other variables.

In general, the purpose of univariate analysis:

1. Understanding the characteristics of the data

If we talk about data characteristics, we can see whether the data we use at a glance is normally distributed, left-winged, right-handed, there are outliers, etc.

2. Knowing the size of concentration, size of distribution, and other descriptive statistics from a data set.

The size of concentration, presentation, etc. is an initial identification to carry out further analysis such as analysis of variants, regression, etc.

3. Generate a frequency distribution of data

By grouping data based on distribution, you will get various interesting information such as how many children have height more than 160 cm, less than 170 cm, etc.

4. Doing some hypothesis testing

Even you are using only one variable, you can use inferential statistics on your data set. There are several tests you could use and make your data prove some valid information. But, there is quite restriction of the test.

In research, before we carry out various tests, modeling, estimations, etc., it’s good we do an analysis of each data or variable that we use.

This will later affect how the quality of your own research results.

Contents

## What needs to be understood in univariate

1. Understand the type of data first

There are many types of data and measurement scales. This will influence the direction of using the univariate analysis itself. It’s good, this type of data is identified in advance to facilitate the analysis process later.

2. In general, the univariate analysis produces only descriptive statistics

Yes, results from univariate analysis cannot be used to draw conclusions from populations such as inferential statistical analysis.

Most likely, the output you produce is frequency distribution, centering size, spread size, percentage of each variable, classification of variables based on certain criteria, etc.

3. Use the normality test, t-test, and chi-square test when you need it

Sometimes, the three tests above are needed when analyzing data. You can use normality test such as Kolmogorov-Smirnov test to make sure that the data is normally distributed, t-test and chi-square test for hypothesis testing.

## Example of Univariate Analysis

Suppose you have a group of height data from 30 people. Let’s look at a univariate analysis of what we might do with that data

### Example of Univariate Analysis with SPSS

For univariate analysis, I am more likely to use SPSS. This is because of the many features of the analysis and the very easy to use process without the need to know formulas or various types of syntax. If you use SPSS, here are the steps in this analysis:

1. Prepare your data set

2. Choose **Analyze > Descriptive Statistics > Frequencies**

3. Click **statistics** and choose what do you want to analyze, and click **continue**

4. Click **chart**

5. Choose the chart that you want to show, and click continue

6. Click **ok** to finish your analysis

7. See and interpret your output

### Example of Univariate Analysis with Microsoft Excel

1. Prepare your data set

2. Activate Analysis ToolPak Addin, choose **File >> Options >> Add-Ins >> Analysis Toolpak**

3. Analysis toolpak is activated in **data** menu

5. Click the icon, and choose **descriptive statistics**

6. Click the analysis that you want to see

7. Here is the output

Based on the results of our univariate analysis, the following information can be obtained:

1. The average height of the 30 samples is 169.86

2. The standard deviation of the height of the 30 sample samples is 5.87

3. The mode or height value that appears most in the data set is 174

4. The range is 20

5. Based on the histogram, the data does not follow the normal distribution. This can be used as an initial identification that the data we use is not normal. However, this certainly cannot be used as an absolute conclusion. Statistical testing is needed in order to get more valid conclusions.

6. The value of kurtosis from the data is -1,224. This means that the data distribution shows left-winged.

7. The skewness value of the data is -0.42. This means that the distribution of data shows a flat distribution (platycurtic).

Closing

Univariate analysis might look like a simple analysis using only one variable. However, univariate analysis is early detection for the use of further analysis in your research.