A university analyzes the relationship between the number of extracurricular activities a student participates in and their GPA. When all students are analyzed together, there is a weak negative association. However, when students are separated into two groups—those with part-time jobs and those without—analysts find a moderate positive association between activities and GPA within each group. Which statement is the most likely explanation for this observation?
Answer: The overall negative association is an example of Simpson's Paradox, likely caused by the lurking variable of having a part-time job.
This scenario describes Simpson's Paradox, where a trend that appears in different groups of data reverses when the groups are combined. Here, having a job is a lurking variable. Students with jobs might have less time for both extracurriculars and studying, leading to lower GPAs and fewer activities on average than students without jobs. This creates an overall negative trend when the groups are combined, even if within each group, students who manage more activities also tend to have higher GPAs.
An education board reports that for high school students, there is a moderate, positive association between the number of advanced placement (AP) courses taken and their final high school GPA. Which of the scatterplots described below would best represent this finding?
Answer: Points are somewhat scattered in a band that goes from the bottom-left to the top-right.
A positive association means the trend goes from bottom-left to top-right. A moderate association means the points are 'somewhat scattered' but still form a discernible linear pattern. 'Tightly clustered' (Choice A) would imply a strong association. 'Widely scattered' (Choice C) would imply a weak or no association. A V-shape (Choice D) is non-linear.
A researcher studies the relationship between age and blood pressure across a wide range of adults (ages 20-80) and finds a moderate, positive, linear association. If the researcher restricts the analysis to only include adults in a narrow age range, such as 40-45 years old, how will the strength of the association likely change for this subgroup?
Answer: The association will likely become much weaker, possibly close to zero.
Restricting the range of the explanatory variable (age) tends to weaken the correlation. The original moderate correlation was visible because of the large variation in age from 20 to 80. Within a very narrow 5-year range, the corresponding variation in blood pressure will be much smaller and will likely look more like a random cloud of points. This will cause the correlation coefficient to be much closer to zero, indicating a weak or nonexistent linear association in that specific subgroup.
A regression analysis of student scores on a final exam versus the number of classes they missed produced an R2 value of 0.64. A scatterplot of the data shows a clear downward linear trend. Which statement is the best description of the association?
Answer: A moderately strong, negative, linear association.
When analyzing correlation and regression results, you need to interpret both the coefficient of determination (R2) and the direction of the relationship shown in the scatterplot.
The R2 value of 0.64 tells you that 64% of the variation in final exam scores is explained by the number of classes missed. To interpret the strength of this association, remember that R2 values around 0.25 indicate weak associations, values around 0.64 suggest moderately strong associations, and values above 0.81 indicate strong associations. The scatterplot shows a "clear downward linear trend," meaning as classes missed increases, exam scores decrease—this is a negative linear relationship.
Answer D correctly identifies this as a moderately strong, negative, linear association, combining the strength indicated by R2=0.64 with the negative direction from the scatterplot.
Answer A is wrong because R2=0.64 indicates moderate strength, not weakness. Answer B incorrectly identifies the direction as positive when the scatterplot clearly shows a downward trend. Answer C makes a fundamental error—R2 doesn't need to equal 1 for a linear association to exist. R2=1 would indicate a perfect linear relationship, but real-world data rarely achieves this.
Remember this rule: R2 tells you strength (how tightly points cluster around the line), while the scatterplot's visual trend tells you direction (positive or negative slope). You need both pieces of information to fully describe the association.
Two dietitians study the relationship between daily calorie intake and weight loss. Dietitian A studies subjects on a 1200-1500 calorie diet, while Dietitian B studies subjects on a 2000-2300 calorie diet. Both find a strong, negative, linear association (higher intake corresponds to less weight loss). The slope of the regression line for Dietitian A's group is -0.1, and for Dietitian B's group it is -0.3. Which conclusion is justified?
Answer: A steeper slope means that in Dietitian B's group, an increase in calorie intake is associated with a larger decrease in weight loss.
The slope of a regression line represents the average change in the response variable (weight loss) for a one-unit change in the explanatory variable (calorie intake). A steeper slope (a value further from zero, in this case -0.3 vs -0.1) indicates a greater change. Slope does not measure the strength or linearity of an association; those are measured by the correlation coefficient r. The problem states both found a 'strong' association, so we cannot conclude one is stronger than the other based on the slope.
An analyst wants to investigate the relationship between the number of cylinders in a car's engine (a quantitative variable) and the car's country of origin (a categorical variable). They create a scatterplot with number of cylinders on the y-axis and country of origin on the x-axis. The resulting plot shows three distinct vertical lines of points. What is the most appropriate conclusion?
Answer: A scatterplot is an inappropriate graph for this data, and conclusions about association should not be drawn from it.
When analyzing relationships between variables, the type of variables involved determines which visualization methods are appropriate. This question tests your understanding of when scatterplots should and shouldn't be used.
A scatterplot requires both variables to be quantitative (numerical) because it plots points along continuous numerical axes. Here, you have number of cylinders (quantitative) and country of origin (categorical). When you force categorical data onto a numerical axis, you get the described pattern: vertical lines of points. This happens because all cars from each country cluster at discrete x-positions, with cylinders varying only vertically.
The correct answer is D because scatterplots simply aren't designed for mixed variable types. The three vertical lines don't represent any meaningful pattern about association—they're an artifact of using the wrong graph type. Drawing conclusions about relationships from this inappropriate visualization would be misleading.
Answer A is wrong because vertical lines don't indicate linear association—that would require points trending diagonally. Answer B incorrectly assumes the lack of a single line/curve means no association exists, but the issue is the graph type, not the relationship itself. Answer C misinterprets the vertical lines as indicating non-linear association, when they actually just show the categorical nature being forced onto a continuous axis.
Study tip: Always check variable types before choosing visualizations. Use scatterplots only when both variables are quantitative. For mixed types like this, consider side-by-side boxplots or comparative histograms instead.
A city finds a strong positive association between the number of firefighters responding to a fire and the amount of damage caused by the fire. A city official, reasoning that the firefighters are causing the damage, proposes sending fewer firefighters to fires to reduce damages. What is the primary flaw in this reasoning?
Answer: The observed association is due to a lurking variable; the size of the fire influences both the number of firefighters and the amount of damage.
This is a classic example of confusing correlation with causation. A third variable, the size or severity of the fire (a lurking variable), is responsible for the association. Larger fires require more firefighters to be sent and they cause more damage. The firefighters are not causing the damage; they are both a response to the fire's size.
A biologist studies the relationship between the ambient temperature of a habitat and the activity level of a certain species of lizard. Which of the following descriptions of the association is most plausible from a biological standpoint?
Answer: A non-linear (curved) association, where activity increases with temperature up to an optimal point, then decreases as it gets too hot.
Many biological processes exhibit an optimal range. Lizards are cold-blooded, so their activity increases as temperature rises from a low point. However, if the temperature becomes excessively high, they will seek shelter to avoid overheating, causing their activity to decrease. This would create a non-linear, inverted U-shaped relationship on a scatterplot.
A scatterplot shows a moderately strong, positive, linear association between the weight of a package in pounds and its shipping cost in dollars. If the package weights are converted from pounds to kilograms (1 kg ≈ 2.2 lbs) and the costs are converted from US dollars to Euros (€1 ≈ $1.10), how will the association be affected?
Answer: The strength and direction of the association will remain the same, but the slope of the regression line will change.
The correlation coefficient (r), which measures the strength and direction of a linear association, is unaffected by linear transformations (like converting units) of the variables. Therefore, the strength and direction remain the same. However, the slope of the regression line (slope=rsxsy) depends on the standard deviations (sy and sx) of the variables, which change when units are converted. Thus, the slope will change.
A scatterplot is created showing the relationship between employees' years of experience and their productivity scores. The plot reveals two distinct clusters of points. One cluster is in the bottom-left (low experience, low productivity) and the other is in the top-right (high experience, high productivity). Within each cluster, the points are randomly scattered with no discernible trend. Which is the best description of the association?
Answer: A strong positive association, which is primarily an artifact of having two distinct groups (e.g., junior and senior employees).
The overall trend from the lower-left cluster to the upper-right cluster is positive and appears strong because the groups are well-separated. However, this overall association is created by the difference between the two groups, not by a continuous trend across all employees. Within the junior group and within the senior group, there is no relationship. This is a case where an aggregate trend is misleading if the subgroup structure is not considered.
A scatterplot of two variables shows that for low values of the explanatory variable, the response variable tends to be high. As the explanatory variable increases, the response variable tends to decrease. The points on the plot are clustered in a broad, cloud-like pattern around a discernible downward trend. Which of the following is the most appropriate description?
Answer: A weak, negative, linear association.
The description that the response variable decreases as the explanatory variable increases indicates a negative direction. The fact that the points are in a 'broad, cloud-like pattern' suggests that the relationship is not precise, meaning it is weak. A 'discernible downward trend' confirms that there is an association, and in the absence of information about curvature, a linear form is the default description for a general trend.
Four scatterplots are shown. Which plot displays a relationship where the Pearson correlation coefficient r would be closest to 0.8?
Answer: Plot I
A correlation coefficient of r=0.8 indicates a strong, positive, linear association. Plot I shows points tightly clustered in a band sloping upward from left to right, which corresponds to a strong positive correlation. Plot II shows a moderate negative correlation (r approx -0.5). Plot III shows a random cloud of points, indicating no linear correlation (r approx 0). Plot IV shows a perfect negative correlation (r=−1.0).
A scatterplot shows three data points: A(2, 2), B(4, 5), and C(6, 3). Which of the following fourth points, when added to the plot, would most substantially weaken any existing linear association?
Answer: (5, 9)
The original three points form a weak, slightly positive association. To weaken an association, one should add a point that deviates significantly from the established trend. Point (8, 4) is roughly consistent with the trend. Point (4, 3.5) is in the center of the data and would have little effect. Point (1, 1) is also reasonably consistent with the trend. Point (5, 9) is a significant outlier in the y-direction, lying far above the trend of the other points. Adding it would pull the regression line strongly upwards and increase the scatter, thus substantially weakening the linear association and decreasing the correlation coefficient.
The scatterplot shows data for two different groups of athletes. Which statement provides the best description of the association shown?
Answer: There is a moderate negative association within each group, but the overall association appears strongly positive.
This is a visual example of Simpson's Paradox. If you examine the points for Group A (circles) alone, they show a downward (negative) trend. Similarly, the points for Group B (squares) also show a negative trend. However, because Group B's points are generally in the upper-right portion of the plot compared to Group A's, the overall trend when all points are considered together is clearly positive. Therefore, the trend within the subgroups is opposite to the overall aggregate trend.
Consider the scatterplot showing the relationship between variables X and Y. Point P is a data point with a high value of X compared to the other points. Which of the following is the best description of Point P and the likely effect of its removal?
Answer: P is a high-leverage point that conforms to the trend; its removal would decrease the strength of the positive association.
Point P has an extreme X-value, making it a high-leverage point. It appears to follow the general positive linear trend of the other points. Such points, when they conform to the pattern, tend to 'anchor' the regression line and increase the correlation coefficient (r). Therefore, removing this point would likely cause the correlation to decrease, meaning the association would become weaker.
An analyst examines the scatterplot showing the relationship between hours of sleep the night before a test and the number of problems missed on the test. Which of the following best describes the association?
Answer: There is a strong, non-linear association.
The points form a clear, predictable pattern (a U-shaped curve), which indicates a strong association. However, the pattern is not a straight line, so it is non-linear. A common mistake is to confuse a correlation coefficient (r) near 0 with a lack of association. The Pearson correlation coefficient r measures the strength of linear relationships. For this dataset, r would be close to 0, but there is clearly a strong relationship between the variables.
An analyst creates a scatterplot with a person's height as the explanatory variable (X-axis) and their foot length as the response variable (Y-axis). The plot shows a strong, positive, linear association. If the analyst swaps the axes, what will be the effect on the Pearson correlation coefficient r and the slope of the least-squares regression line?
Answer: The correlation coefficient r will remain the same, but the slope will change.
When you encounter questions about swapping variables in correlation and regression analysis, focus on the fundamental mathematical properties of these statistical measures.
The Pearson correlation coefficient r measures the strength and direction of the linear relationship between two variables. Crucially, correlation is symmetric — it doesn't matter which variable you call X and which you call Y. The formula r=∑(xi−xˉ)2∑(yi−yˉ)2∑(xi−xˉ)(yi−yˉ) produces the same result regardless of variable order. So swapping height and foot length won't change r.
The slope of the regression line, however, is not symmetric. When height predicts foot length, the slope tells you how much foot length increases per unit increase in height. When foot length predicts height, you get a completely different slope representing how much height increases per unit increase in foot length. Mathematically, if the original slope is b, the new slope becomes br2 (not simply b1).
Looking at the options: B is wrong because while r stays the same, the slope definitely changes. C is wrong because r is symmetric and doesn't change when axes swap. D is wrong because it incorrectly assumes both measures change.
Study tip: Remember "correlation is symmetric, regression is not." The correlation coefficient treats both variables equally, but regression slopes depend entirely on which variable is predicting which. This distinction appears frequently on statistics exams.
A researcher creates a scatterplot of student study hours per week versus their grade point average (GPA). The plot shows a moderate, positive, linear association. A new data point is added for a student whose data was mistyped: 25 study hours and a GPA of 0.5. This point is a significant outlier compared to the general trend of the original data. How will this new point most likely affect the description of the association?
Answer: The strength of the positive association will decrease, and the form may appear less linear.
The new point (high study hours, very low GPA) is an influential outlier that contradicts the existing positive trend. It will 'pull' the regression line towards it, decreasing the slope and thus weakening the positive correlation (decreasing the value of r). The presence of this single point far from the general pattern could also make the overall form appear less linear.
Which of the following statements most accurately describes the distinction between the form and the strength of an association on a scatterplot?
Answer: Form refers to the general pattern of the points (e.g., linear, curved), while strength refers to how closely the points follow that pattern.
When analyzing scatterplots, you need to distinguish between two key characteristics of associations: form and strength. These concepts help you describe the relationship between variables systematically.
Form describes the overall shape or pattern that the data points create on the scatterplot. This could be linear (points roughly follow a straight line), curved (quadratic, exponential, etc.), or have no discernible pattern. Strength measures how tightly the points cluster around whatever pattern exists - whether the points closely follow the form or are widely scattered around it.
Answer A correctly captures this distinction: form identifies the general pattern (linear, curved, etc.), while strength indicates how closely points adhere to that pattern. A strong linear relationship shows points tightly clustered around a straight line, while a weak linear relationship has the same straight-line pattern but with points more scattered.
Answer B confuses form with direction. Whether an association is positive or negative describes the direction, not the form. Answer C incorrectly links strength to slope and form to y-intercept - these are components of the regression equation, not fundamental characteristics of scatterplot associations. Answer D misunderstands both concepts entirely by bringing in causation versus correlation, which relates to interpretation of relationships, not their visual characteristics on scatterplots.
Study tip: When examining any scatterplot, always ask yourself two questions in order: "What pattern do I see?" (form), then "How closely do the points follow that pattern?" (strength). This systematic approach will help you accurately describe any association.
The scatterplot shows the relationship between a country's per capita income and the life expectancy of its citizens. Which statement best describes the features of the data presented in the plot?
Answer: A positive, non-linear (curved) association with more variation in life expectancy at lower income levels.
The plot shows that as per capita income increases, life expectancy also tends to increase, so the association is positive. The relationship is not linear; it shows a steep increase at lower incomes which then levels off, indicating a curved (non-linear) form. Furthermore, the vertical spread (variation) of the points is much larger for countries with low per capita income than for countries with high per capita income. This feature is called heteroscedasticity.
