Contents
Introduction
What is partial correlation?
Application of partial correlation in reality.
Null Hypothesis for partial correlation
How to carry out partial correlation
Analysis
Interpretation of the output of partial correlation.
Conclusion
Introduction
When two or more variables are linked together in a process, such is referred to as Correlation or co- occurrence. But there are some variables that have no mutual relationship in real life but can be influenced by a certain factor otherwise known as control variable which aims at linking them together. This brings us to the reason why partial correlation must be studied.
Are you a statistical analyst, you wish to learn more or comment suggestion as regards this post, you are welcome as this article will cover for the following sub topics. They are; definition of partial correlation, application of partial correlation in reality, how to carry out partial correlation analysis in SPSS, interpretation of the output of partial correlation and lastly the conclusive aspect.
What is partial correlation?
Partial correlation is a relationship between two variables that in the real sence cannot be linked together but it is been related by the introduction of external variable to influence the relationship.
Partial correlation makes it possible to relate variable that has no relationship in real life such variables does not occur in same process but can be linked when a factor is considered.
Application of partial correlation in reality.
Under this subtopic, examples of cases where partial correlation can be applied shall be considered. As this will give a better understanding of the subject matter.
Does men bring jobs?
in this question, there is no relationship between men and jobs available in a country because women also goes to work. But a third factor can be considered in the analysis which is availability of natural resources which aids employment and create the job. If there are natural resources present in abundance, there will be available jobs.
Are the number of women and children in the society related?
There is a bit of relationship between women and children but I decided to bring this example under partial correlation for you to also have a vivid idea of what partial correlation is all about.
Women on thier own cannot have babies which turn into children but when you consider the number of men as well it influences the question if women and children are correlated because it is the men the have intercourse with the women for babies to be form which later becomes children.
I own a fish farm, and I want to find out if the amount of sale is related with the amount of expenses made.
Do you agree with me that sales and expenses are two different things?
I can only determine the answer when I consider advertisment. How have I been advertising my product to the market? And how has it affected my rate in expense deduction?
The third variable in control is advertisment of product. When I have high advertisment rate of my fishes, people tends to buy more by the grace of Jesus Christ and more income will be generated to by business which may or may not influence my expenses.
When trying to link things together and considering a third variable known as control variable brings about the idea of partial correlation.
Null Hypothesis for Partial Correlation
The null hypothesis for partial correlation is the same with the null hypothesis for correlation which states that there is no relationship between the variables in consideration.
You will determine if you are to accept or reject the null hypothesis when considering the level of significance.
How to carry out partial correlation
Analysis in SPSS
To carry out partial correlation Analysis in SPSS follow the procedures listed below.
Ensure you view the directionality of the relationship using scatter dot diagram.
To view directionality;
Go-to graph
Click on chart builder and
Select scatter dot
Drag the variables to be visualized unto the x and y axis. SPSS will create a scattered diagram of the variables in consideration.
You can click add fit line to view direction better.
Interpretation of the line of best fit
If the line is slanted to the y axis, there is a positive correlation meaning that increase in the variable of the x axis will lead to increase in the variable of the y axis.
If the line is slanted to the x axis, there is a negative relationship between both variables. That is to say a decrease in the variable of x axis will lead to an increase in the variable of y axis.
To carryout partial correlation Analysis test in SPSS:
Go to analyze
Click on correlate
Select partial
A dialogue box will be opened for you to add the variable that may not be related. It is either you drag the variables or use the arrow button on the dialogue box after selecting the variable.
Add the controlling factor.
Click on options
In order to compute additional statistics such as mean and standard deviation but if you have not done a correlation Analysis of the variables, tick the zero-order- Correlation option given.
Click on ok to display the output.
Interpretation of the output of partial correlation.
The output of the analysis is shown immediately after the ok button is clicked, the result is splited into two parts.
The first half shows intercorrelation between all the variables in consideration including the controlling variable.
The second part presents the result when the third variable is considered otherwise known as controlling factor.
Now use the correlation value and interpret the direction of Correlation aswell as read the significance level to determine if you have to accept or reject the null hypothesis.
Conclusion
This article has been able to explain the term 'partial correlation' to the best of one's understanding. It is important to run correlation test of variables that are not related inorder to view directionality of relationship. However, if there are questions, suggestions or comments as regards should be indicated in the comment section of this guide.