What is the relationship between emotional intelligence and job performance? Is there a link between years of experience and burnout? Questions that begin with “what is the relationship” are answered with a correlation coefficient, and the two coefficients that appear in almost every thesis are the Pearson correlation (r) and Spearman’s rho.
This article is part of our guide How to Write a Master’s Thesis by the Mastermind PhD team and relates to our Statistical Analysis service. Sources last checked: November 2026.
Computing them is easy; the difficulty lies in choosing the right one and interpreting the value: is .32 weak or moderate, does a significant result mean a strong relationship, and may you say that one variable “affects” the other? This guide answers in order: the choice between the two coefficients, the SPSS steps, reading the matrix, interpreting strength and direction, the APA sentence and table, and the mistakes examiners comment on.
What is a correlation coefficient? A correlation coefficient is a number between −1 and +1 that measures the strength and direction of the association between two variables. Pearson’s r measures the linear relationship between two continuous variables. Spearman’s rho (rs) measures the monotonic relationship between the ranks of two variables and is used with ordinal data or when the assumptions of Pearson’s r are not met. The sign gives the direction, the absolute value the strength, and the p value the probability of observing such a value if the population correlation were zero.
Pearson correlation or Spearman? The selection criteria
| Criterion | Pearson’s r | Spearman’s rho |
|---|---|---|
| Type of variables | Both continuous (interval or ratio), including averaged scale scores | At least one ordinal variable, or two continuous variables that fail Pearson’s assumptions |
| Shape of relationship | Linear | Monotonic (consistently increasing or decreasing, not necessarily in a straight line) |
| Distribution | Prefers approximate bivariate normality; strongly affected by outliers | No distributional assumption; robust to outliers |
| Sample size | Stable from about 30 cases | Works with small samples |
| What is reported | r, N or df, p, r² | rs, N or df, p |
A quick rule: averaged Likert scale scores with a reasonable sample, an acceptable distribution and a scatterplot showing a linear trend call for Pearson correlation. A single Likert item, a ranked variable, a skewed distribution or outliers you cannot remove call for Spearman. Treating multi-item scale scores as interval follows Norman (2010); a single item stays ordinal.
How do you run a Pearson correlation (or Spearman) in SPSS?
- Look at the scatterplot first: Graphs, then Chart Builder, then Scatter/Dot. You are checking whether the relationship is roughly linear and whether any outliers stand apart from the cloud.
- Analyze, then Correlate, then Bivariate.
- Move all the variables of interest into Variables; SPSS produces a matrix of every pair.
- Tick Pearson or Spearman (both, if you are still checking assumptions), keep Two-tailed, and keep Flag significant correlations.
- Under Options, tick Means and standard deviations if you want descriptives in the same output, and choose how to handle missing data (Exclude cases pairwise is the default). In version 27 and later, a confidence interval for a Pearson correlation can be requested; the Bootstrap button is an alternative in older versions.
- Click OK. The Correlations table shows, for every pair, the coefficient, the p value and N.
To correlate two variables while controlling for a third (experience and performance, controlling for age), use Analyze, then Correlate, then Partial.
📱 Ask a statistician about your data — Send a short description of your instrument and sample size on WhatsApp and get a suitable analysis plan. Every analysis comes with the raw SPSS file and a results report.
How do you interpret the strength?
There is no single binding scale. What matters is to name the one you adopt, cite it and apply it consistently.
| Absolute value of r | Cohen (1988) | Evans (1996) | Schober, Boer and Schwarte (2018) |
|---|---|---|---|
| .00 to .09 | Not classified | Very weak (up to .19) | Negligible |
| .10 to .29 | Small | Weak (.20 to .39) | Weak (.10 to .39) |
| .30 to .49 | Medium | Moderate (.40 to .59) | Moderate (.40 to .69) |
| .50 to .69 | Large | Strong (.60 to .79) | Strong (.70 to .89) |
| .70 and above | Large | Very strong (.80 to 1.00) | Very strong (.90 to 1.00) |
Cohen’s benchmarks are the most common reference in the social and educational sciences; Schober and colleagues are widely used in the health sciences. A Pearson correlation of .32 between emotional intelligence and performance is “medium” under Cohen and “weak” under Evans, so stating the source is not a formality.
Direction, significance and the coefficient of determination
- Direction. A positive sign means the variables rise together; a negative sign means one rises as the other falls. A negative correlation between experience and burnout (r = −.41) is not weaker than a positive one of the same size.
- Significance. The p value only tells you whether the population correlation is plausibly zero. With 500 cases, r = .09 is significant at .05 and practically meaningless; with 25 cases, r = .35 may not be significant although it is medium-sized. Interpret strength and significance together, and report a confidence interval when your SPSS version provides one.
- Coefficient of determination (r²). The proportion of variance the two variables share. A Pearson correlation of .50 means that 25% of the variance in one variable is associated with variance in the other. Reporting it prevents medium correlations from being oversold.
- Correlation is not causation. The coefficient does not say which variable drives the other and does not rule out a third variable. Use “is associated with” and “is related to” rather than “affects” or “leads to”, unless the design is experimental or the claim rests on a theoretically justified regression model.
How do you write the result in APA style?
In the text: “Emotional intelligence was positively and moderately correlated with job performance, r(298) = .42, p < .001; emotional intelligence accounted for about 18% of the variance in performance (r² = .18).” For Spearman: “rs(118) = −.36, p < .001.” The number in parentheses is the degrees of freedom, N − 2; some faculties prefer N itself, in which case write “r = .42, N = 300, p < .001″. No leading zero before r or p, two decimals for r, and exact p values unless p < .001.
In a table: a triangular correlation matrix with the coefficients, one asterisk for p < .05 and two for p < .01, and a note beneath giving N and the symbols. Do not paste the SPSS matrix with its three lines per cell.
Worked example: a matrix of three variables
A study of 300 employees measured emotional intelligence and job performance (two scale scores on a 5-point format) and years of experience (four ordered bands). The scatterplots show linear trends with no outliers and the two scale scores are acceptably distributed, so Pearson correlation is used between them; experience is ordinal, so Spearman is used for any pair involving it.
Results: emotional intelligence with performance, r = .42, p < .001 (medium-to-large under Cohen; r² = .18, or 18% shared variance); experience with performance, rs = .36, p < .001 (medium); emotional intelligence with experience, rs = .09, p = .12 (not significant).
The thesis table presents a three-by-three triangular matrix with two asterisks on the two significant coefficients and the note: “N = 300. Coefficients between the two scale scores are Pearson’s r; coefficients involving experience are Spearman’s rho.” In the text: “Emotional intelligence was moderately and positively correlated with job performance, r(298) = .42, p < .001, but not with years of experience, rs(298) = .09, p = .120.” This wording pre-empts the examiner’s question about which coefficient was used and why.
Which mistakes do examiners comment on?
- Computing Pearson correlation between single Likert items instead of scale scores.
- Skipping the scatterplot, so that a strong curved relationship appears weak, or a single outlier manufactures a correlation.
- Using Pearson correlation with a dichotomous variable such as gender; the point-biserial correlation or a t-test is appropriate and gives the same conclusion.
- Describing r = .12 in a large sample as “a strong relationship” because it is significant.
- Reporting r without N or p, or writing p = 0.000.
- Interpreting the correlation causally in the discussion chapter.
- Correlating within a restricted range (only top performers) and generalizing the weak result to everyone.
📱 Ask on WhatsApp now — 5 years of experience across all disciplines and 14 academic services. Two-stage payment (80/20); revisions are free until supervisor approval, within 30% of the chapter.
Frequently asked questions
Can I use Pearson’s r with Likert-scale data?
Yes, for scores averaged from several items, which are treated as interval in research practice (Norman, 2010), provided you check the scatterplot and the distribution. Single items are ordinal and call for Spearman.
What is the difference between Spearman’s rho and Kendall’s tau?
Both measure monotonic association between ranks. Kendall’s tau-b is preferable with small samples and many tied ranks; Spearman is more common in theses and easier to interpret because it is on the same scale as *r*. Either is acceptable with a stated reason.
How do I handle an outlier that changes the correlation?
First check whether it is a data-entry error. If it is genuine, report the coefficient with and without it, or use Spearman, which is less affected, and document the decision. Deleting it without justification is selective reporting.
How large a sample do I need for a correlation?
There is no hard minimum, but below 30 cases the coefficient is unstable and its confidence interval very wide. To detect a medium Pearson correlation of .30 with 80% power at alpha .05 you need about 85 cases according to Cohen’s tables, which G*Power reproduces.
Should I answer a “relationship” question with correlation or regression?
Correlation is sufficient for a question about the existence, direction and strength of a relationship. If the question asks about effect or prediction, or you have several predictors, use regression, which builds on the same correlation matrix but adds a prediction equation and the proportion of variance explained.
Sources
- Cohen, J. (1988). Statistical Power Analysis for the Behavioral Sciences (2nd ed.). Lawrence Erlbaum.
- Evans, J. D. (1996). Straightforward Statistics for the Behavioral Sciences. Brooks/Cole.
- Schober, P., Boer, C., & Schwarte, L. A. (2018). Correlation coefficients: Appropriate use and interpretation. Anesthesia & Analgesia, 126(5), 1763–1768. https://doi.org/10.1213/ANE.0000000000002864
- Norman, G. (2010). Likert scales, levels of measurement and the “laws” of statistics. Advances in Health Sciences Education, 15(5), 625–632.
- Field, A. (2018). Discovering Statistics Using IBM SPSS Statistics (5th ed.). SAGE.
- Laerd Statistics. Pearson’s Product-Moment Correlation using SPSS Statistics. https://statistics.laerd.com/spss-tutorials/pearsons-product-moment-correlation-using-spss-statistics.php
- IBM. IBM SPSS Statistics documentation: Bivariate Correlations. https://www.ibm.com/docs/en/spss-statistics
Sources last checked: November 2026. This article is general educational guidance and does not replace your university regulations or your supervisor’s instructions. How we prepare our content.