One hypothesis appears in master’s and PhD theses more than any other: “There is a statistically significant relationship between variable (A) and variable (B).” The two variables may be job satisfaction and performance, test anxiety and academic achievement, or service quality and customer loyalty. The tool that tests this hypothesis is the correlation coefficient.
The problem is that SPSS gives you three options in a single window: Pearson, Kendall’s tau-b and Spearman, without telling you which one fits your data. Many supervisor and examination committee comments start right here: “Why did you use Pearson when the data are not normally distributed?” In this guide, we explain the difference between Pearson and Spearman, how to calculate them in SPSS, how to read the output table, and how to write the result in your results chapter.
What Is a Correlation Coefficient?
A correlation coefficient is a number that summarizes the relationship between two variables in a single value ranging from -1 to +1. It tells you two things:
- Direction: A positive value means the two variables increase together. A negative value means that as one increases, the other decreases, such as the relationship between test anxiety and grades.
- Strength: The closer the value is to 1 or -1, the stronger the relationship; the closer it is to zero, the weaker it is.
There is one rule you must remember before any interpretation: correlation does not mean causation. If you find a strong relationship between study hours and grades, this does not prove that studying causes the grades. A third variable may be affecting both. So write “there is a relationship” in your thesis, not “affects”.
The Difference Between Pearson and Spearman Correlation
The Pearson coefficient (r) measures the linear relationship between two quantitative variables, meaning a relationship that can be represented by a straight line. The Spearman coefficient (rho or rs) first converts the values into ranks, then measures the monotonicrelationship, meaning whether one variable increases as the other increases, even if the increase does not follow a straight line.
| Comparison | Pearson | Spearman |
|---|---|---|
| Type of test | Parametric | Nonparametric |
| Type of data | Quantitative (interval or ratio) | Ordinal, or quantitative data that do not meet Pearson’s assumptions |
| Normal distribution | Approximately required | Not required |
| Shape of the relationship | Linear | Monotonic (consistently increasing or decreasing) |
| Sensitivity to outliers | Highly sensitive | Less sensitive |
| Thesis example | Total score on a satisfaction scale and total score on a performance scale | A student’s class rank and the father’s education level |
The practical rule: if your data are quantitative and meet Pearson’s assumptions, use it, because it is more powerful in detecting a relationship. If the assumptions are not met, or the data are ordinal to begin with, use Spearman. If the difference between parametric and nonparametric tests is not clear to you, see The Difference Between Parametric and Nonparametric Tests and How to Choose the Right One.
Assumptions of Pearson Correlation and How to Check Them in SPSS
Before you click Pearson, check four assumptions:
- Both variables are quantitative: scores, ages, income, or a total score on a multi-item scale.
- Normal distribution: Go to Analyze, then Descriptive Statistics, then Explore. Put both variables in the Dependent List, and under Plots enable Normality plots with tests. If the Sig. value of the Shapiro-Wilk or Kolmogorov-Smirnov test is greater than 0.05, the distribution does not differ significantly from normal.
- The relationship is linear: Draw a scatterplot from Graphs, then Legacy Dialogs, then Scatter/Dot, then Simple Scatter. If the points cluster around a straight line, the assumption is met; if they form a curve, Pearson will underestimate the true strength of the relationship.
- No influential outliers: A single distant point in the scatterplot can raise or lower the value of r considerably. You can also see this in the Boxplot from the same Explore window.
If the normality or linearity assumption fails, do not try to “fix” the data by deleting cases without a methodological reason. Switching to Spearman is the safest solution, and state the reason in your methodology chapter.
Steps to Calculate the Correlation Coefficient in SPSS
- Open the data file and choose from the menu: Analyze, then Correlate, then Bivariate.
- Move the two variables (or more) to the box Variables.
- Under Correlation Coefficients choose Pearson or Spearman according to your data. You can select both for comparison, but you will report only one result in your thesis.
- In Test of Significance choose Two-tailed if your hypothesis is “there is a relationship” without specifying a direction, andOne-tailed if you specified the direction in advance (“there is a positive relationship”).
- Keep the option Flag significant correlations enabled, so that SPSS marks significant coefficients with an asterisk.
- Click Okay.
If you enter more than two variables, SPSS will give you a correlation matrix that brings all pairs together in one table, which is useful when you test the relationship of an independent variable with several dimensions at the same time.
Reading the Correlation Output Table in SPSS
With Pearson, a table titled Correlationsappears, and with Spearman, a table titled Nonparametric Correlationsappears. In both cases, each cell in the table contains three numbers:
- Pearson Correlation (or Correlation Coefficient in Spearman): the value of the correlation coefficient itself.
- Sig. (2-tailed): the significance value. If it is less than 0.05, the relationship is statistically significant.
- N: the number of cases included in the calculation. Always check it, because cases with missing values are excluded, and the number may be smaller than your actual sample size.
Illustrative example: In a study of 120 teachers, the correlation between job satisfaction and performance was 0.42, and the Sig. value was 0.000. This means there is a positive, statistically significant relationship of medium size according to the criteria in the next section. (These numbers are for illustration only.)
Important note: SPSS shows the significance as 0.000 when it is very small, but it is not zero. In your thesis, write it as p < .001.
How to Present a Correlation Matrix in Your Thesis
If you tested the relationships among several variables, do not copy the SPSS table as it is. The original table repeats each value twice (above and below the diagonal) and places the significance value and number of cases under each coefficient, which makes it long and crowded. It is better to rebuild it in Word like this:
- Put the variable names in the first row and first column in the same order, and number them (1, 2, 3).
- Write the correlation values in only half of the table (below the diagonal), and leave the diagonal blank or put a dash in it.
- Use asterisks for significance: one asterisk for p < .05 and two for p < .01, and explain this in a note below the table.
- Add two columns for the mean and standard deviation of each variable if your university requires it; this is common in APA style.
After the table, do not repeat all the numbers in the text. Write a paragraph highlighting only the most important relationships that answer your hypotheses, and leave the rest to the table.
Interpreting the Strength of the Correlation Coefficient
The most widely used criterion in research is Cohen’s (Cohen, 1988):
| Correlation coefficient value (absolute value) | Strength of the relationship |
|---|---|
| About 0.10 | Weak |
| About 0.30 | Medium |
| 0.50 or more | Strong |
Some universities and supervisors use other cut-offs, so follow the criterion used in your department and cite its source.
There are two points that many researchers confuse:
- Significance is not strength. With a very large sample, a weak correlation such as 0.15 can be statistically significant. Significance tells you that the relationship is probably not due to chance, while strength tells you whether it matters in practice.
- The squared correlation coefficient (r²) gives you the proportion of shared variance between the two variables. A correlation of 0.42 means that about 18% of the variance in one variable is associated with the other. This number adds depth to your discussion chapter.
How to Write the Correlation Result in the Results Chapter
The following wording follows APA style, and you can adapt it to your study:
Pearson:
To test the first hypothesis, Pearson’s correlation coefficient was calculated between job satisfaction scores and performance scores. The results showed a medium, positive and statistically significant relationship between the two variables (r = .42, n = 120, p < .001), meaning that the higher the teachers’ job satisfaction, the higher their performance.
Spearman:
Because the normality assumption was not met (Shapiro-Wilk test, p < .05), Spearman’s correlation coefficient was used. The results showed a positive and statistically significant relationship between the two variables (rs = .35, n = 85, p = .001).
Note that the second wording justifies the choice of Spearman. This one line alone prevents an expected question from the examination committee. For more details on table layout and wording, see How to write the statistical results chapter in a master's thesis and doctoral dissertation.
Common Mistakes When Using the Correlation Coefficient
- Using Pearson with a single Likert item. A single item on a five-point scale is ordinal data, and Spearman suits it better. A total score on a multi-item scale, however, is usually treated as quantitative data, and Pearson can be used with it if the assumptions are met.
- Writing “affects” instead of “is related to”. Correlation does not prove causation, and committees notice this word.
- Ignoring the scatterplot. A coefficient close to zero does not always mean there is no relationship. The relationship may be curved, and Pearson does not capture it.
- A large correlation matrix without attention to the number of tests. The more pairs you test, the greater the chance of finding a significant relationship by chance. If the matrix is large, discuss using a correction such as Bonferroni with your supervisor.
- Confusing correlation with regression. Correlation tells you whether a relationship exists and how strong it is. But if your goal is to predict one variable from another variable or several variables, you need Linear regression analysis in SPSS and interpreting the results.
Frequently Asked Questions
Can Pearson’s correlation be used with Likert data?
With the total score of a multi-item scale, yes, if the assumptions are met. With a single item, Spearman is more appropriate.
What is the difference between Spearman and Kendall (Kendall’s tau)?
Both are nonparametric and based on ranks. Kendall is often preferred when the sample is small or has many tied ranks.
What does a negative correlation value mean?
It means the relationship is inverse: as one variable increases, the other decreases. A negative sign does not mean the relationship is weak; a correlation of -0.60 is stronger than one of 0.30.
What is partial correlation and when do I need it?
Partial correlation measures the relationship between two variables after removing the effect of a third variable. For example: the relationship between study hours and achievement after controlling for age. You can find it in SPSS under Analyze, then Correlate, then Partial, and you need it when your supervisor suspects that your relationship is due to another variable.
What sample size is appropriate for correlation?
There is no fixed number that fits every study. Small samples give unstable coefficients, and it is better to calculate the sample size in advance with a power analysis using a program such as G*Power.
How do I report the correlation coefficient in my thesis according to APA?
In APA style, the value is written without a zero before the decimal point (r = .42), because a correlation coefficient cannot exceed one. Follow the writing guide adopted by your university.
Not sure which test to use? Try the free wizard
Answer short questions about your study aim and your data, and the wizard suggests a suitable test, where to find it in SPSS and its assumptions, with a ready methods paragraph.
Before You Write the Correlation Result in Your Thesis
Choosing the right test matters more than the calculation itself. If you are not sure whether your data meet Pearson’s assumptions, or how to write the result in a way that convinces the committee, the Science Club Academy team can review your data and hypotheses in a free consultation, then run the analysis and write the interpretation in the language of your thesis.
Do you need help analyzing your research data?
The Science Club Academy team helps you prepare your data, analyze it, and interpret the results for master’s theses, doctoral dissertations, and research papers. Learn more about our Integrated statistical analysis service, and the consultation is free.


No comment