Real-world examples of determining strength of correlation

Everyday examples of determining strength of correlation

Most people first meet correlation in a classroom, but the best examples come from everyday life. Any time you say “those two things seem related,” you’re informally talking about correlation.

A classic example of a strong positive correlation is height and weight in adults. Taller people tend to weigh more. If you collected data on 1,000 adults and ran a correlation analysis, you’d likely see a Pearson correlation coefficient (r) around 0.6–0.8 in many samples. That range is often interpreted as a moderate to strong positive correlation: as height increases, weight usually increases too, though with plenty of individual variation.

On the other hand, consider shoe size and reading ability in children. At first glance, you might see a strong correlation: older kids have bigger feet and can read better. But the strength of correlation here is driven by age, a third variable. This is a perfect example of determining strength of correlation while also asking whether the relationship is meaningful or just a side effect of something else.

These are simple examples, but they illustrate the main point: correlation strength is not just a number. It’s a story about how two variables move together, how tightly they move, and whether that movement makes sense in the real world.

Health and medicine: real examples of determining strength of correlation

Health research is packed with examples of determining strength of correlation, because medical decisions often start with patterns in data.

Example: Smoking and lung cancer

Decades of epidemiological studies show a very strong positive correlation between cigarette smoking and lung cancer incidence. In many cohort studies, the correlation between pack-years of smoking and lung cancer risk is high, even after adjusting for other variables. Organizations like the National Cancer Institute summarize this body of evidence.

If you plotted cigarettes per day against lung cancer cases per 100,000 people, the points would cluster along an upward trend. Correlation coefficients in such datasets can easily exceed 0.8 at the group level. This is one of the best examples where a strong correlation supported a causal conclusion (after additional research), changing public policy and clinical practice worldwide.

Example: Physical activity and resting heart rate

Another example of a negative correlation in health: minutes of moderate-to-vigorous exercise per week and resting heart rate. People who exercise more tend to have lower resting heart rates. In observational data, you might see r values around -0.4 to -0.6, which many analysts treat as a moderate negative correlation.

If you were a health researcher in 2024 using wearable data from smartwatches, you’d likely compute this correlation across thousands of users. Determining the strength of correlation here helps answer questions like: Is the relationship strong enough to use resting heart rate as a rough indicator of someone’s activity level? The answer is often “yes, but with caveats,” which is exactly why understanding correlation strength matters.

Example: BMI and type 2 diabetes risk

Body mass index (BMI) and type 2 diabetes risk provide another widely studied relationship. Large datasets, including those summarized by the Centers for Disease Control and Prevention (CDC), show that higher BMI is associated with higher diabetes prevalence.

In population-level data, the correlation between BMI and diabetes status (often coded 0/1) is usually moderate: strong enough to be statistically significant and clinically important, but not so strong that BMI perfectly predicts who will develop diabetes. This is a real example where a moderate correlation still has major public health implications.

Education and learning: examples of examples of determining strength of correlation

Education research gives some of the clearest examples of examples of determining strength of correlation, because outcomes like test scores and graduation rates are easy to measure.

Example: Study hours and exam scores

Imagine tracking 300 college students in an introductory statistics course. You record how many hours they study for the final exam and their exam scores. When you calculate Pearson’s r, you might get something like 0.5.

A correlation of 0.5 is often interpreted as a moderate positive correlation. It tells you that, on average, more study hours are linked to higher scores, but the scatterplot will still show plenty of students who studied a lot and did poorly, or studied little and did surprisingly well.

This is one of the best examples to teach that:

• A moderate correlation can still be practically useful.
• Correlation does not guarantee an outcome for any single individual.

Example: Attendance and course grades

Now take course attendance percentage and final course grade. Many instructors who have analyzed their own class data find correlations around 0.6 or higher, especially in discussion-heavy courses. Education researchers, including many at universities like Harvard and others, routinely examine these patterns when studying student engagement.

Here, the strength of correlation supports decisions such as:

• Encouraging attendance policies.
• Flagging students with low attendance as at-risk.

It’s a real example of determining strength of correlation leading directly to an intervention: if the correlation were weak (say 0.1), attendance would be a poor predictor of performance and a weaker target for policy.

Example: SAT/ACT scores and first-year GPA

Standardized test scores (SAT/ACT) and first-year college GPA are another widely studied pair. Meta-analyses often find correlations in the 0.3–0.4 range. That’s a moderate relationship: test scores tell you something about likely performance, but not everything.

This is a powerful example of determining strength of correlation in a high-stakes context. Colleges that rely heavily on test scores are implicitly treating a moderate correlation as if it were strong. As test-optional policies expand in 2024–2025, many institutions are revisiting how much weight to give a predictor with only moderate correlation to outcomes.

Economics, income, and markets: real examples of correlation strength

In economics and finance, correlation is a workhorse. Analysts constantly ask how tightly variables move together over time.

Example: Education level and income

Across the United States, individuals with more years of education tend to earn higher incomes. Data from the U.S. Bureau of Labor Statistics consistently show higher median weekly earnings for people with bachelor’s degrees compared with high school diplomas.

When you compute correlation between years of education and income in a large survey, you often see a moderate positive correlation, sometimes around 0.4–0.5. This is a real example of determining strength of correlation where:

• The relationship is clearly positive.
• The correlation is strong enough to matter for policy.
• The scatterplot still shows plenty of outliers (high-earning tradespeople with less formal education, for example).

Example: Stock market indices

Consider daily returns of two major stock indices, such as the S&P 500 and the NASDAQ Composite. Historically, their daily returns show a strong positive correlation, often above 0.8. For an investor, this is one of the best examples of a strong correlation: the indices tend to move in the same direction on most days.

Meanwhile, if you compare an S&P 500 index fund with long-term U.S. Treasury bonds, the correlation in daily or monthly returns is often weak or even negative over some periods. Portfolio managers constantly assess these correlations to decide how to diversify risk.

Here, determining the strength of correlation is not academic; it’s directly tied to how people manage retirement funds and institutional portfolios.

Tech, data science, and 2024–2025 trends: examples include algorithmic decisions

Modern data science provides fresh examples of determining strength of correlation, especially as organizations rely on algorithms to make predictions.

Example: Click-through rate and time-on-site

In 2024, digital marketing teams routinely analyze how click-through rate (CTR) on ads relates to time spent on a website. In many cases, there’s a weak to moderate positive correlation: users who click are somewhat more likely to stay longer, but not always.

A data scientist might find r ≈ 0.3 between CTR and average session duration across campaigns. This is a real example where a weak-to-moderate correlation still influences strategy: campaigns with higher CTR are promising, but you can’t assume they automatically drive deep engagement.

Example: App notifications and daily active users

Mobile app teams often track the number of push notifications sent and daily active users (DAU). If the correlation is very weak or near zero, it suggests that blasting more notifications does not reliably increase engagement and might even annoy users.

If, however, a product analyst finds a moderate positive correlation (say 0.4) for targeted notifications, that’s an example of determining strength of correlation guiding product decisions: invest in smarter, personalized notifications rather than raw volume.

Example: Machine learning features and target variables

In machine learning, correlation analysis is an early step in feature engineering. For instance, a health-tech startup building a model to predict risk of hospitalization might examine correlations between variables like age, number of chronic conditions, prior ER visits, and future admissions.

Features with stronger correlations to the target variable (such as prior ER visits) are often prioritized, though they’re never used alone. This is one of the best examples of determining strength of correlation as a practical tool: it helps teams decide which features are worth modeling more deeply, and where to invest data collection efforts.

How analysts interpret “strong” vs “weak” correlation in practice

Different fields use slightly different rules of thumb, but many statisticians informally interpret the absolute value of Pearson’s r like this:

• Around 0.1: very weak
• Around 0.3: weak to moderate
• Around 0.5: moderate
• Around 0.7 and above: strong

These are not hard rules. In some areas of psychology or education, a correlation of 0.3 can be considered impressive because human behavior is noisy and influenced by many factors. In tightly controlled physics experiments, a correlation of 0.3 would look flimsy.

So when you see examples of determining strength of correlation in the literature, pay attention to:

• Context: What’s normal in this field?
• Sample size: In huge datasets, even tiny correlations can be statistically significant but practically unimportant.
• Shape of the relationship: Pearson’s r captures linear correlation. A curved relationship can have r near zero even if the variables are strongly related in a non-linear way.
• Outliers: A few extreme points can inflate or deflate correlation.

Real analysts always pair the statistic with a scatterplot and domain knowledge. The number is part of the story, not the whole story.

FAQ: examples of correlation strength in everyday questions

What is an example of a strong correlation in real life?

A widely cited example of a strong correlation is the relationship between the number of cigarettes smoked and the risk of lung cancer at the population level. Groups with higher smoking rates have much higher lung cancer rates, and the correlation between exposure and outcome is very high. This relationship helped drive major public health campaigns and policy changes.

What are examples of weak correlations that still matter?

In social science and education, many examples include correlations around 0.2–0.3 that still inform policy. For instance, a weak-to-moderate correlation between class size and student achievement can still influence how districts allocate resources, even if the effect on any individual student is modest.

Can you give examples of negative correlations?

Yes. Common examples of negative correlations include:

• Exercise time and resting heart rate (more exercise, lower resting heart rate).
• Price of a product and quantity demanded (higher price, lower demand), at least within a normal range.
• Hours of TV watched per day and hours available for homework.

These are all real examples of determining strength of correlation where the variables move in opposite directions.

Is a correlation of 0.3 considered strong or weak?

A correlation of 0.3 is usually described as weak to moderate. It indicates a real relationship, but with plenty of scatter in the data. Whether it’s “strong enough” depends on context. In medicine or public health, even a 0.3 correlation can be important if it affects millions of people, as seen in many risk factor studies summarized by agencies like the National Institutes of Health (NIH).

Why do we use examples of examples of determining strength of correlation when teaching statistics?

Because correlation is abstract until you see it in action. Using examples of examples of determining strength of correlation from health, education, economics, and tech helps students and practitioners connect the number (r) to real decisions: Should we change a policy, adjust a marketing campaign, or redesign an app feature? Concrete cases make it easier to remember that correlation is about patterns in real lives, not just symbols on a whiteboard.

In short, the best way to understand correlation is to keep collecting and comparing examples. When you see how analysts in different fields use the same basic idea—variables moving together—to answer very different questions, the concept stops feeling theoretical and starts feeling like a practical tool you can use on your own data.