The analysis method is an operational tool, and the result is a list of analyzed data. So, how to view and interpret the results after conducting correlation analysis using SPSS?
1、 Bivariate analysis
1. Data
Figure 1: Data List
The data sample shown in the above figure is the basic situation of employees recruited by a certain company. What we want to explore is the correlation between education level and starting salary.
2. Analysis settings
Figure 2: bivariate analysis setup
Set the two variables as shown in the figure above, selecting Pearson correlation and two tailed test.
3. Result analysis
Figure 3: Analysis Results
There are usually two tables in the results (using only one analysis method), the first table is the frequency statistics and basic calculation results, showing the mean, standard deviation, and number of cases of the data.
The second table represents the correlation of data, and one parameter that we need to focus on is the Pearson correlation coefficient, also known as the p-value. If the absolute value of this value is greater than 0.05, it indicates a significant correlation between variables.
There are two data points marked with asterisks in the table, which are p-values greater than 0.05, indicating a significant positive correlation between education level and starting salary.
2、 Partial correlation analysis
1. Analysis settings
Figure 4: Analysis Settings
Partial correlation analysis is a method used to eliminate interference from relevant variables and analyze the correlation between other variables.
Using the same sample, if we want to explore the correlation between starting salary and education level more accurately, we can exclude all other variables as confounding variables, and the analysis setup is shown in the above figure.
2. Result analysis
Figure 5: Partial correlation results
Due to the large number of controlled variables, this table is also quite complex, but it can be visually observed the direction and strength of the correlation between each two variables.
At the end of the table, there is a correlation analysis between education level and starting salary after controlling for variables. It can be found that after controlling for other variables, the Pearson coefficient significantly decreased, from 0.633 in the bivariate analysis to 0.235. There is still a positive correlation between them, but the degree of correlation is greatly reduced.
It can be seen that for data with a large number of variables and complex relationships, using partial correlation analysis to analyze the correlation between two variables or specific variables will be more reliable.
3、 Distance analysis
1. Non similarity distance analysis of individual cases
Figure 6: Distance Analysis Settings
Select the first five sets of data in this sample as the analysis objects, and calculate the non similarity distance between cases by analyzing variables such as education level, employment category, starting salary, current salary, and experience.
The result of this analysis method is a non similarity matrix, where the smaller the parameter values in the table, the greater the similarity. The table shows that the distance between Case 2 and Case 5 is the smallest, indicating the strongest correlation between them.
2. Analysis of Variable Correlation Distance
Figure 8: Analysis Settings
Change the analysis object to individual cases and non similarity to similarity, and set it as above.
Figure 9: Result analysis
Similar to the dissimilarity matrix, we will obtain a similarity matrix where the parameters represent the similarity distance between variables, and the larger the parameters, the stronger the correlation.
From the parameters displayed in the table, it can be seen that among the three variables analyzed, the distance between starting salary and employment category is the largest and the correlation is the strongest.
