Pearson Correlation: Formula, Interpretation, Python, R, SPSS and Excel Guide

Quick Answer: Pearson Correlation Result

The main Pearson Correlation analysis uses 16 numeric variables and 649 complete observations. The workbook calculates 120 pairwise Pearson correlations, including correlations with the main outcome variable G3 final grade.

The strongest absolute correlation is between G2 and G3. The Pearson correlation coefficient is r = 0.918548, which is a very strong positive linear association. The shared variance is r² = 0.843730, meaning approximately 84.37% of the variation in one variable is linearly shared with the other in this pairwise relationship.

What Is Pearson Correlation?

Pearson Correlation is a statistical measure that describes the strength and direction of a linear relationship between two quantitative variables. The coefficient is usually written as r. It ranges from -1 to +1. A value close to +1 means a strong positive linear relationship. A value close to -1 means a strong negative linear relationship. A value close to 0 means little or no linear relationship.

A positive Pearson correlation means that higher values of one variable tend to occur with higher values of the other variable. A negative Pearson correlation means that higher values of one variable tend to occur with lower values of the other variable. In this post, the strongest positive example is G2 with G3, where students with higher second-period grades tend to have higher final grades.

Pearson correlation is best understood visually with a scatterplot. If the points form an upward-sloping cloud, the correlation is positive. If the points form a downward-sloping cloud, the correlation is negative. If the points are widely scattered without a clear line-like pattern, Pearson r is near zero. The stronger and tighter the line-like pattern, the larger the absolute value of r.

This guide connects Pearson r with Correlation in Python, Correlation in R, Correlation in SPSS, Correlation in Excel, Correlation Matrix, Correlation Heatmap, Correlation Assumptions, Correlation vs Regression, p-value, effect size and confidence interval.

When Should You Use Pearson Correlation?

Use Pearson Correlation when both variables are numeric and the research question is about linear association. It is especially useful when you want to summarize the relationship between two continuous or quantitative variables with one coefficient.

In the student performance example, the main grade variables G1, G2 and G3 are numeric grade scores. Pearson correlation is therefore useful for studying their linear relationships. However, the workbook also includes outlier and distribution checks because variables such as failures, Dalc, absences and grade scores may contain skewness, repeated values or influential observations.

Pearson Correlation Formula

The Pearson correlation coefficient is calculated by comparing how two variables vary together relative to their individual variation.

In this formula, xᵢ and yᵢ are paired observations, x̄ is the mean of X, and ȳ is the mean of Y. The numerator measures how X and Y move together. The denominator standardizes that movement so the final coefficient is always between -1 and +1.

The significance test for Pearson r uses:

The degrees of freedom are:

For the strongest pair G2–G3, n = 649, r = 0.918548, df = 647, and the p-value is extremely small. In normal reporting, this is written as p < .001.

Null and Alternative Hypotheses

The Pearson correlation hypothesis test checks whether the population correlation is zero. For the strongest pair, the hypotheses are about the relationship between G2 and G3.

Dataset and Variables Used

The analysis uses the student performance dataset. The workbook selected 16 numeric variables for Pearson correlation. These variables include demographic, academic, lifestyle, attendance and grade variables.

Although Pearson correlation is a numeric method, some selected variables are ordinal-style numeric variables. This is common in educational examples, but it should be reported carefully. For purely ordinal interpretation, Spearman or Kendall methods may also be considered.

Descriptive Statistics and Assumption Diagnostics

The Excel workbook includes descriptive statistics and practical assumption diagnostics for each selected variable. These values help readers understand the scale, spread, skewness and possible outlier issues before interpreting Pearson r.

The grade variables are numeric and suitable for Pearson correlation, but the outlier and distribution checks still matter. For example, failures is heavily skewed because many students have zero failures. Absences is also skewed, with some higher absence counts. These are the types of variables where Pearson correlation should be interpreted along with scatterplots and outlier context.

Verified Pearson Correlation Results

The workbook dashboard identifies G2–G3 as the strongest absolute pair. This is the main result used for the strongest pair scatterplot and top correlation interpretation.

The strongest pair is expected because G2 and G3 are consecutive grade outcomes. G2 is the second-period grade and G3 is the final grade, so they measure closely related academic performance. The scatterplot should show a tight upward pattern, which is consistent with r = 0.918548.

Correlations with G3

Because G3 is the final grade, the workbook dashboard summarizes each selected variable’s Pearson correlation with G3. This section is useful for understanding which variables are most closely related to final grade.

The most important pattern is that grade variables dominate the G3 correlations. G2 and G1 are far stronger than demographic, lifestyle or attendance variables. Failures is the strongest negative non-grade relationship with G3.

Strongest Pairwise Pearson Correlations

The full correlation matrix contains 120 unique variable pairs. The strongest pairs by absolute value are listed below.

This table shows that Pearson correlation can reveal both academic and lifestyle relationships. Grade-to-grade relationships are the strongest, while parent education and alcohol-use patterns also show meaningful associations.

Python Chart-by-Chart Interpretation

The Python assets include five charts: Pearson correlation matrix, Pearson p-value heatmap, correlations with G3, strongest pair scatterplot and outlier context. These charts explain the relationship structure visually and support the Excel workbook values.

Python Chart 1: Pearson Correlation Matrix

The Pearson correlation matrix heatmap shows all pairwise linear associations in one chart. The strongest positive block appears among G1, G2 and G3. This means the grade variables move together strongly across students.

The matrix also shows positive associations such as Medu–Fedu and Dalc–Walc. Medu–Fedu means mother education and father education are strongly related. Dalc–Walc means weekday and weekend alcohol consumption move together.

Negative cells show inverse linear relationships. The most important negative pattern for final grade is failures–G3, with r = -0.393316. Students with more past failures tend to have lower final grades.

Python Chart 2: Pearson p-value Heatmap

The p-value heatmap shows which correlations are statistically significant. Strong grade relationships have extremely small p-values. Many weaker relationships also become statistically significant because the sample size is large.

This chart should always be interpreted together with the coefficient matrix. The p-value tells whether the result is statistically detectable. The Pearson r value tells the strength and direction of the linear relationship. A small p-value does not automatically mean a strong correlation.

For example, some G3 correlations such as goout or absences may be statistically significant but practically small. The correct report should include both r and p-value.

Python Chart 3: Correlations with G3

This chart focuses on the main outcome variable, G3. It shows that G2 has the strongest positive relationship with G3, followed by G1. These two grade variables dominate the G3 correlation pattern.

The strongest negative relationship with G3 is failures. This makes sense educationally because students with more past failures tend to have lower final grades.

Other variables, such as studytime, Medu, Fedu, Dalc and Walc, have weaker relationships with G3. They may be statistically meaningful, but they are much smaller than the grade-to-grade relationships.

Python Chart 4: Strongest Pair Scatterplot

The strongest pair scatterplot shows the relationship between G2 and G3. The points form a tight upward pattern. This visual pattern matches the Pearson correlation coefficient r = 0.918548.

This chart is the best visual explanation of the main result. It shows that students with higher second-period grades usually have higher final grades. The cloud is not perfectly straight, but it is strongly linear and positive.

When reporting this chart, avoid causal wording. Do not say that G2 causes G3. A better statement is that G2 and G3 have a very strong positive linear association.

Python Chart 5: Outlier Context

The outlier context chart is important because Pearson correlation can be sensitive to influential observations. Extreme points can inflate, weaken or reverse a correlation depending on where they appear in the scatterplot.

The workbook flags several variables for review, including failures, absences, G1, G2 and G3. This does not mean Pearson correlation is invalid. It means the results should be interpreted with scatterplots and diagnostics, not only with numeric r values.

For the strongest G2–G3 relationship, the correlation remains very high, but the outlier chart still supports responsible reporting. For variables such as absences and failures, outlier context is even more important because these distributions are skewed.

R Chart-by-Chart Interpretation

The R assets validate the same analysis with colorful charts. The R report includes a Pearson correlation matrix, correlations with G3, strongest pair scatterplot and outlier context chart.

R Chart 1: Colorful Pearson Correlation Matrix

The R correlation matrix confirms the same overall structure as the Python heatmap. The grade variables form the strongest positive cluster. Medu and Fedu also show a strong positive relationship, and Dalc and Walc show a strong positive lifestyle relationship.

This chart is useful because it validates the matrix pattern in a second software environment. When Python, R and Excel show the same relationship structure, the result is more trustworthy for readers.

R Chart 2: Colorful Correlations with G3

The R G3 chart confirms that G2 and G1 are the strongest positive correlates of final grade. Failures is the strongest negative correlate. Studytime, parent education and alcohol-use variables have weaker but interpretable relationships.

This chart is ideal for students writing a report because it narrows the full matrix down to the final-grade outcome. It shows which predictors or background variables are most associated with G3.

R Chart 3: Colorful Strongest Pair Scatterplot

The R scatterplot again shows the tight upward relationship between G2 and G3. The pattern supports the very strong positive Pearson coefficient.

Scatterplots are essential for Pearson correlation because they show whether the relationship is approximately linear. A high r should be supported by a line-like cloud, and the G2–G3 chart provides that visual support.

R Chart 4: Colorful Outlier Context

The R outlier context chart reinforces the same assumption check as the Python chart. Some variables contain skewness or possible influential values. This is expected in real educational data.

The correct conclusion is not to reject Pearson correlation automatically. The correct conclusion is to combine Pearson r with visual checks. If a relationship is driven by a small number of extreme observations, Spearman correlation, robust correlation or a regression diagnostic check may be useful.

SPSS Output and Workflow Interpretation

The SPSS output PDF is included in the resources. SPSS is commonly used for Pearson correlation because it produces a clean correlation table with coefficients, significance values and sample sizes.

Open the SPSS Pearson Correlation Output PDF

In SPSS, Pearson correlation is usually run through Analyze → Correlate → Bivariate. Select the numeric variables, choose Pearson, select two-tailed significance, and request means and standard deviations if needed.

For the strongest pair, SPSS should be interpreted as: G2 and G3 have a very strong positive Pearson correlation, r ≈ .919, p < .001, N = 649. When reporting SPSS output, always include whether the test is two-tailed and whether missing values were handled pairwise or listwise.

Excel Worked File Explanation

The Excel workbook is a fully worked Pearson Correlation analysis file. It contains raw data, selected variables, descriptive statistics, a Pearson correlation matrix, pairwise N, p-value matrix, confidence interval sheets, pairwise results, assumptions, formulas and a dashboard.

Download the Pearson Correlation Full Excel Analysis File

Excel Formula Method

The workbook uses the following Excel formula logic:

The Excel dashboard is especially useful because it provides the final story quickly: 649 dataset rows, 16 numeric variables, 120 variable pairs, main outcome G3, strongest pair G2–G3 and the interpretation rule for coefficient strength.

Python, R, SPSS and Excel Workflows

Pearson Correlation can be calculated in all four tools. The same statistical result should appear if the same variables, missing-value rules and numeric coding are used.

For a complete Salar Cafe workflow, use Excel to verify the formulas, Python to generate the main blog charts, R to validate the outputs, and SPSS to provide the official-style statistical output PDF.

Code Blocks and Excel Formulas

Python Code for Pearson Correlation

import pandas as pd
from scipy import stats

df = pd.read_csv("dataset.csv")

x_var = "G2"
y_var = "G3"

work = df[[x_var, y_var]].dropna().copy()
work[x_var] = pd.to_numeric(work[x_var], errors="coerce")
work[y_var] = pd.to_numeric(work[y_var], errors="coerce")
work = work.dropna()

r, p_value = stats.pearsonr(work[x_var], work[y_var])
n = len(work)
dfree = n - 2
t_value = r * ((dfree / (1 - r**2)) ** 0.5)

print("Variable X:", x_var)
print("Variable Y:", y_var)
print("N:", n)
print("Pearson r:", r)
print("R squared:", r**2)
print("t statistic:", t_value)
print("df:", dfree)
print("p-value:", p_value)

Python Code for Pearson Correlation Matrix

import pandas as pd
from scipy import stats

df = pd.read_csv("dataset.csv")

vars_used = [
    "age", "Medu", "Fedu", "traveltime", "studytime", "failures",
    "famrel", "freetime", "goout", "Dalc", "Walc", "health",
    "absences", "G1", "G2", "G3"
]

data = df[vars_used].apply(pd.to_numeric, errors="coerce")

corr_matrix = data.corr(method="pearson")

p_matrix = pd.DataFrame(index=vars_used, columns=vars_used, dtype=float)
n_matrix = pd.DataFrame(index=vars_used, columns=vars_used, dtype=float)

for x in vars_used:
    for y in vars_used:
        pair = data[[x, y]].dropna()
        n_matrix.loc[x, y] = len(pair)
        if x == y:
            p_matrix.loc[x, y] = None
        else:
            r, p = stats.pearsonr(pair[x], pair[y])
            p_matrix.loc[x, y] = p

print(corr_matrix)
print(p_matrix)

R Code for Pearson Correlation

df <- read.csv("dataset.csv", stringsAsFactors = FALSE)

x_var <- "G2"
y_var <- "G3"

work <- na.omit(df[, c(x_var, y_var)])
work[[x_var]] <- as.numeric(work[[x_var]])
work[[y_var]] <- as.numeric(work[[y_var]])

test <- cor.test(work[[x_var]], work[[y_var]], method = "pearson")

print(test)

cat("N:", nrow(work), "\n")
cat("Pearson r:", unname(test$estimate), "\n")
cat("R squared:", unname(test$estimate)^2, "\n")
cat("p-value:", test$p.value, "\n")

R Code for Pearson Correlation Matrix

df <- read.csv("dataset.csv", stringsAsFactors = FALSE)

vars_used <- c(
  "age", "Medu", "Fedu", "traveltime", "studytime", "failures",
  "famrel", "freetime", "goout", "Dalc", "Walc", "health",
  "absences", "G1", "G2", "G3"
)

data <- df[vars_used]
data[] <- lapply(data, as.numeric)

corr_matrix <- cor(data, use = "pairwise.complete.obs", method = "pearson")

p_matrix <- matrix(NA, nrow = length(vars_used), ncol = length(vars_used))
rownames(p_matrix) <- vars_used
colnames(p_matrix) <- vars_used

for(i in seq_along(vars_used)){
  for(j in seq_along(vars_used)){
    if(i != j){
      pair <- na.omit(data[, c(vars_used[i], vars_used[j])])
      p_matrix[i, j] <- cor.test(pair[[1]], pair[[2]], method = "pearson")$p.value
    }
  }
}

print(corr_matrix)
print(p_matrix)

SPSS Syntax for Pearson Correlation

* Pearson Correlation in SPSS.
OUTPUT CLOSE ALL.
OUTPUT NEW NAME=Pearson_Correlation_Output.

DESCRIPTIVES VARIABLES=age Medu Fedu traveltime studytime failures famrel freetime goout Dalc Walc health absences G1 G2 G3
  /STATISTICS=MEAN STDDEV MIN MAX.

CORRELATIONS
  /VARIABLES=age Medu Fedu traveltime studytime failures famrel freetime goout Dalc Walc health absences G1 G2 G3
  /PRINT=TWOTAIL
  /MISSING=PAIRWISE.

GRAPH
  /SCATTERPLOT(BIVAR)=G2 WITH G3
  /TITLE='Strongest Pearson Correlation Pair: G2 with G3'.

OUTPUT EXPORT
  /CONTENTS EXPORT=VISIBLE
  /PDF DOCUMENTFILE='Pearson-Correlation-SPSS-Output.pdf'.

Excel Formula Patterns for Pearson Correlation

Assume:
X_range = G2 values
Y_range = G3 values

Pearson r:
=CORREL(X_range,Y_range)

R squared:
=r^2

Pairwise N:
=SUMPRODUCT(--ISNUMBER(X_range),--ISNUMBER(Y_range))

Degrees of freedom:
=N-2

t statistic:
=r*SQRT((N-2)/(1-r^2))

Two-tailed p-value:
=T.DIST.2T(ABS(t),N-2)

Fisher z:
=ATANH(r)

Standard error of z:
=1/SQRT(N-3)

95% CI lower:
=TANH(ATANH(r)-NORM.S.INV(0.975)/SQRT(N-3))

95% CI upper:
=TANH(ATANH(r)+NORM.S.INV(0.975)/SQRT(N-3))

Decision:
=IF(p_value<0.05,"Significant correlation","Not significant")

Pearson Correlation Assumptions

Pearson correlation is simple to calculate, but the interpretation depends on assumptions and data checks. The workbook includes an assumptions sheet covering numeric scale, linearity, outliers, approximate normality for inference and independence.

For this dataset, Pearson correlation is appropriate for grade-to-grade relationships, especially G2–G3 and G1–G3. For variables with strong skewness or ordinal categories, Pearson r should be interpreted alongside scatterplots and possibly compared with rank-based measures.

How to Report Pearson Correlation

A complete Pearson correlation report should include the variables, sample size, coefficient, confidence interval if available, p-value, direction, strength and practical interpretation. It should also mention that correlation does not prove causation.

Common Mistakes in Pearson Correlation Interpretation

Downloads and Resources

External References

For additional learning, review documentation and textbooks on Pearson product-moment correlation, Fisher z confidence intervals, bivariate correlation testing, correlation matrices, scatterplot interpretation and outlier diagnostics.

Related Statistical Guides