Exploring Relationships with Bivariate Analysis in SPSS

When researchers want to understand whether two variables are connected – say, whether education level relates to income, or whether gender influences political participation – they need a method that goes beyond describing individual variables in isolation. That is exactly what bivariate analysis does. It examines the relationship between two variables at a time, testing whether a pattern exists, how strong it is, and in which direction it runs. SPSS makes this process structured and accessible, offering three core tools for the job: cross-tabulation, correlation coefficients, and linear regression. Knowing when and how to use each one is a foundational skill for any researcher working with quantitative data.

What is bivariate analysis?
Cross-tabulation: examining categorical variables
The chi-square test of independence
Correlation coefficients: measuring relationships between continuous variables
Running and reading Pearson’s r in SPSS
Before you run: check with a scatterplot
Linear regression: predicting outcomes from variables
Running linear regression in SPSS
Reading the SPSS regression output
Choosing the right method
Common pitfalls to avoid
Applications in social research

What is bivariate analysis?

Bivariate analysis is the statistical examination of the relationship between exactly two variables. Unlike univariate analysis, which describes a single variable (its mean, frequency, or spread), bivariate analysis asks a different question: do these two variables move together in some meaningful way? It is the starting point for hypothesis testing in most social research. The researcher identifies one variable as the independent variable (IV) – the presumed cause or predictor – and the other as the dependent variable (DV) – the presumed outcome or response.

The method you use in SPSS depends on what kind of variables you have. As a general rule: when both variables are categorical (nominal or ordinal), cross-tabulation with a chi-square test is appropriate; when both are continuous (interval or ratio), correlation or regression analysis applies; and when one is categorical and the other continuous, techniques like independent samples t-tests or ANOVA come into play, though these sit slightly outside the core bivariate toolkit covered here.

Cross-tabulation: examining categorical variables

Cross-tabulation – often called a crosstab or contingency table – is the go-to method when both your variables are categorical. It organises data into a grid where rows represent the categories of one variable and columns represent the categories of another. Each cell in the grid shows how many cases fall into that particular combination of categories, which makes it easy to spot patterns at a glance.

In SPSS, you run a crosstab by going to Analyze → Descriptive Statistics → Crosstabs. Place the independent variable in the Columns box and the dependent variable in the Rows box. Once the table is generated, reading it correctly matters. Column percentages are typically the most useful: they show how the dependent variable is distributed within each category of the independent variable, making comparisons across groups straightforward.

The chi-square test of independence

Crosstabs tell you what the data looks like, but the chi-square test of independence tells you whether that pattern is statistically meaningful. This test evaluates whether the observed distribution in the table differs from what you would expect if the two variables were completely unrelated. The logic is straightforward: if gender and political party preference are truly independent, you would expect roughly the same party distribution across men and women. The chi-square statistic measures how far the observed counts deviate from those expected counts.

In SPSS, you request the chi-square by clicking the Statistics button in the Crosstabs dialog and selecting Chi-square. The key number to look at in the output is the p-value (labelled “Asymptotic Significance” in SPSS). If the p-value is less than 0.05, you reject the null hypothesis and conclude that a statistically significant association exists between the two variables. A p-value above 0.05 means you cannot rule out that the pattern occurred by chance alone.

Beyond statistical significance, you will also want to assess the strength of the association. Measures like Cramér’s V (suitable for any table size) or Phi (for 2×2 tables) provide this information. These are chi-square-based measures of association that help quantify how strongly the variables are related, independent of sample size effects.

Correlation coefficients: measuring relationships between continuous variables

When both your variables are measured on a continuous scale – income in dollars, age in years, test scores out of 100 – correlation analysis is the appropriate bivariate technique. A correlation coefficient captures two things simultaneously: the direction of the relationship (positive or negative) and its strength, expressed as a single number ranging from -1 to +1.

A positive value means that as one variable increases, the other tends to increase as well; a negative value means that as one increases, the other tends to decrease; and a coefficient of zero indicates no linear relationship between the variables. The closer the coefficient is to +1 or -1, the stronger the relationship.

Running and reading Pearson’s r in SPSS

The most widely used correlation coefficient for continuous data is Pearson’s r. In SPSS, you access it through Analyze → Correlate → Bivariate. Move the variables of interest into the Variables box, ensure “Flag significant correlations” is checked, and click OK. SPSS produces a correlation matrix showing the coefficient, its p-value (labelled “Sig. 2-tailed”), and the sample size (N) for each pair of variables.

For ordinal data, Spearman’s rho is more appropriate. You can also request Spearman correlation through the same Bivariate dialog in SPSS by selecting the Spearman option instead of Pearson. For nominal data with a 2×2 structure, Phi is suitable; for larger nominal tables, Cramér’s V applies – both can be requested through the Crosstabs statistics menu.

One critical point in interpreting correlation results: correlation does not imply causation. A strong Pearson’s r between two variables tells you they move together – it says nothing about which one causes the other, or whether a third variable is driving both.

Before you run: check with a scatterplot

Pearson’s r only detects linear relationships. If your variables have a curved or non-linear relationship, the coefficient may be close to zero even when a real relationship exists. Producing a scatterplot first – through Graphs → Legacy Dialogs → Scatter/Dot – helps you visually confirm whether the relationship between your variables appears linear before committing to Pearson’s correlation. This step is considered good practice, not optional.

Linear regression: predicting outcomes from variables

Correlation tells you whether a relationship exists and how strong it is. Linear regression goes further: it quantifies exactly how much change in one variable is associated with a unit change in the other, and it lets you build a predictive equation. In simple bivariate linear regression, one continuous independent variable is used to predict one continuous dependent variable.

The regression equation takes the form Ŷ = a + bX, where Ŷ is the predicted value of the dependent variable, X is the independent variable, a is the constant (the y-intercept, or the predicted value of Y when X equals zero), and b is the unstandardised regression coefficient showing how much Y changes for each one-unit increase in X.

Running linear regression in SPSS

In SPSS, go to Analyze → Regression → Linear. Move your dependent variable into the Dependent box and your independent variable into the Independent(s) box, then click OK. The output window produces four tables: Variables Entered/Removed, Model Summary, ANOVA, and Coefficients. Each plays a distinct role in interpreting your results.

Reading the SPSS regression output

Start with the Model Summary table. The key figure here is R² (R-Square), also known as the coefficient of determination. R² tells you what proportion of the total variation in the dependent variable is explained by the independent variable. An R² of 0.45, for instance, means that 45% of the variance in the outcome is accounted for by the predictor – a moderately strong result in social science research.

Next, check the ANOVA table. The Sig. value in this table tells you whether the overall regression model is statistically significant. A p-value below 0.05 confirms that the model predicts the dependent variable significantly better than chance.

Finally, the Coefficients table gives you the numbers needed to write the regression equation. The value of the constant (a) and the unstandardised coefficient B for the independent variable (b) appear in the “B” column under Unstandardized Coefficients. These figures let you construct the equation Ŷ = a + bX and use it to make predictions. The Sig. column in this same table shows whether each coefficient is statistically significant – a p-value below 0.05 means the independent variable is a meaningful predictor of the outcome.

Choosing the right method

Selecting between cross-tabulation, correlation, and regression is not arbitrary – it depends on your variable types and your research question. Use cross-tabulation with chi-square when both variables are categorical. Use Pearson’s r or Spearman’s rho when both variables are continuous and you want to assess the strength and direction of their relationship. Use linear regression when you want to quantify how much one continuous variable predicts another and need a working equation for that prediction. Regression also provides standardised beta coefficients, which allow you to compare the relative influence of different predictors when you extend to multiple variables.

Common pitfalls to avoid

Even well-designed analyses can produce misleading results if certain issues go unchecked. Three mistakes come up repeatedly in bivariate analysis. First, assuming causation from correlation: a strong coefficient between two variables does not mean one causes the other. Always consider whether a third, unmeasured variable might be driving both. Second, ignoring outliers: a single extreme value can dramatically inflate or deflate a correlation coefficient or regression slope, making a weak relationship appear strong or masking a genuine one. Always inspect your data visually before running formal tests. Third, over-relying on statistical significance with large samples: in very large datasets, even trivial relationships may reach statistical significance. Always consider both the statistical significance and the practical, substantive meaning of your results in the context of your research question.

Bivariate analysis in SPSS is not just a classroom exercise – it is a standard tool in professional and academic research across disciplines. Sociologists use cross-tabulation to examine whether survey responses differ by gender, ethnicity, or social class. Policy researchers use Pearson’s r to assess whether socioeconomic indicators correlate with health outcomes. Criminologists use linear regression to explore how education levels predict crime rates. In media and representation research, crosstabs have been used to examine whether certain studios show significantly different rates of female directors compared to others – a direct application of cross-tabulation to a social equity question. These analyses form the empirical backbone of evidence-based arguments in policy, advocacy, and academic publication.

Bivariate analysis is also typically the second step in a larger analytical process. Researchers usually begin with univariate descriptive statistics to understand their variables individually, then move to bivariate analysis to test initial hypotheses, and finally proceed to multivariate models when they want to account for multiple factors simultaneously. Getting the bivariate stage right – choosing the correct test, running it properly in SPSS, and interpreting the output accurately – builds the foundation for all of that subsequent work.

What do you think? When you encounter a headline claiming that two things are “strongly correlated,” what questions would you now ask about the nature of those variables and the method used to measure that relationship? And how might the choice between correlation and regression change the kinds of conclusions a researcher is able to draw from the same dataset?

How useful was this post?

Click on a star to rate it!

Average rating 0 / 5. Vote count: 0

No votes so far! Be the first to rate this post.

References

PDF 📄

Table of Contents

What is bivariate analysis? 🔗

Cross-tabulation: examining categorical variables 🔗

The chi-square test of independence 🔗

Correlation coefficients: measuring relationships between continuous variables 🔗

Running and reading Pearson’s r in SPSS 🔗

Before you run: check with a scatterplot 🔗

Linear regression: predicting outcomes from variables 🔗

Running linear regression in SPSS 🔗

Reading the SPSS regression output 🔗

Choosing the right method 🔗

Common pitfalls to avoid 🔗

Applications in social research 🔗