Google

Correlation is a statistical association observed between two variables. When high values of one variable tend to occur with high values of another, the variables are positively correlated. Conversely, when high values of one variable are associated with low values of another, there is a negative correlation. It is crucial to note that correlation does not imply causation; a statistical link between two variables does not necessarily mean that one causes the other.

Correlation

It is important to understand the correlation does not mean causation.

- Correlations may be direct. This is where the two variables directly influence one another. The influence may run in one or both directions. 

- Correlations may be indirect. There is where a third, unknown factor affects the two variables being measured

Direction of Correlations

A city analyst observes that neighborhoods with a higher number of public libraries also report, on average, a higher median household income. Based solely on this observed statistical association, which of the following is the most accurate conclusion?

An elementary school study finds a strong positive statistical association between students' shoe sizes and their scores on a reading comprehension test. Based on this finding, a researcher concludes that having larger feet improves a child's ability to read. Analyze the researcher's conclusion. What is the primary logical error in this reasoning?

Analyzing a Faulty Conclusion

Read the following scenario and evaluate the proposed policy. In your evaluation, explain whether the reasoning is sound and provide a more likely explanation for the observed data.

Evaluating a Public Health Proposal

An economist publishes a study showing a strong positive statistical association between the number of coffee shops per capita in a city and the city's overall economic productivity. A local newspaper reports on this study with the headline: 'Want a Stronger Economy? Build More Coffee Shops!' 

Analyze the newspaper's conclusion. In your response, first describe the relationship that the economist actually identified. Then, explain why the newspaper's causal claim might be misleading. Finally, propose at least one alternative explanation for the observed association.

Interpreting Economic Data

A study reveals a strong positive statistical association between the monthly sales of ice cream and the number of drowning incidents at a beach resort. This finding proves that eating ice cream before swimming increases the risk of drowning.

Match each statistical concept with the real-world scenario that best illustrates it.

An economist observes that in years when the average price of coffee beans increases, the average quantity of tea sold in cafes also increases. What type of statistical relationship does this observation describe between the price of coffee and the quantity of tea sold?

An analyst finds a strong positive statistical association between the number of firefighters sent to a fire and the total dollar amount of damage caused by that fire. The analyst concludes that sending more firefighters to a fire causes more damage. Briefly explain why this conclusion is likely flawed and propose an alternative explanation for the observed association.

Evaluating a Causal Claim

An automotive industry analyst observes that as the age of a specific model of car increases, its average resale value consistently decreases. This statistical relationship, where an increase in one variable is associated with a decrease in another, is described as a ______ correlation.

A researcher studying elementary school children finds a strong positive statistical association between their shoe size and their scores on a standardized reading test. As shoe size increases, reading scores tend to be higher. Which of the following statements provides the most logical analysis of this finding?

A numerical index, commonly represented as Pearson's $$r$$, that measures the direction and strength of a linear relationship between two quantitative variables. The coefficient is bounded between $$-1.00$$ and $$+1.00$$, where values further from zero indicate a tighter, more predictable association, and values near zero indicate a negligible relationship. In research, this coefficient frequently serves as an effect size measure to evaluate how strongly variables are connected in the population.

Correlation Coefficient

Correlations have predictive value. Once a statistical relationship between two variables is observed and its strength is determined, this information can be used to predict one variable based on the other. For instance, observing strong correlations in existing data helps in predicting relative success or forecasting future outcomes in similar scenarios.

Predictive Value of Correlation

This dataset is presented as a scatterplot illustrating the relationship between the US inflation rate and the electoral margin for the incumbent party in presidential elections. A line of best fit is drawn through the data points, sloping downward to indicate that the ruling party typically performs better when inflation is low. The chart omits two years of deflation; including this data would make the observed negative correlation between inflation and electoral success even stronger.

Figure 5.1: Inflation and US Presidential Election Outcomes (1912–2020)

Empirical data from 2009-2019, visualized in Figure 7.16, shows a strong positive correlation between the average rate of a currency's depreciation and its average CPI inflation rate. While this correlation suggests that fixing the exchange rate could be a solution to high inflation, it is important to note that this statistical relationship does not in itself establish a causal link between the two phenomena.

Strong Correlation Between Exchange Rate Depreciation and Inflation (2009-2019)

The scatter plot in Figure E5.1 illustrates a clear negative correlation between the degree of central bank independence in the mid-1980s and the average inflation rate from 1962 to 1990 for a sample of OECD countries. The plot shows that nations with more independent central banks, such as Germany and Switzerland, tended to have lower inflation, while countries with less independent central banks, like Portugal and Greece, experienced higher inflation.

Negative Correlation Between Central Bank Independence and Inflation (OECD, 1962-1990) [Figure E5.1]

The data from Figure 7.16 provides evidence that countries maintaining their nominal exchange rates within a narrow range against the U.S. dollar also tended to experience comparatively low inflation rates. This observation extends the findings beyond strictly fixed exchange rate regimes, suggesting that even managed stability can be associated with price stability.

Empirical Link Between Relatively Stable Exchange Rates and Low Inflation

A nonlinear relationship is a statistical association between quantitative variables where the data points on a scatterplot are better represented by a curved line rather than a single straight line. Unlike linear relationships, nonlinear correlations exhibit a changing rate or direction across the range of the variables. Because standard correlation metrics like Pearson's $$r$$ only measure linear trends, they can misleadingly indicate a weak relationship even when a strong nonlinear pattern is present.

Nonlinear Relationship

When analyzing the relationship between variables like television viewing and exam scores, it is crucial to distinguish between correlation and causation. The observation that "students who watch more television perform more poorly on their exams" describes a negative correlation. However, statements such as "watching too much television leads to poor exam performance" or "viewing television interferes with a student’s ability to prepare" are causal claims. Correlational data alone does not support these causal conclusions.

Distinguishing Correlation from Causation: Television and Exams

Scatterplots are a graphical method used to display correlations and relationships between quantitative variables. They are particularly useful when the variable plotted on the $$x$$-axis—typically the independent variable—has a large number of distinct levels. In a scatterplot, scores are plotted at the intersection of the values on the $$x$$-axis and $$y$$-axis, with each individual data point representing a single participant's scores across the two variables. Unlike line graphs, the points in a scatterplot stand alone and are not connected by lines.

Scatterplots

A linear relationship is a statistical association between quantitative variables in which the data points on a scatterplot are reasonably well fit by a single straight line. Both positive and negative correlations are examples of linear relationships, provided their points form a straight linear trend rather than a curve.

Learn Before

Related