Data-Based and Statistical Reasoning

Physics and Math

Kinematics and Dynamics
Work and Energy
Thermodynamics
Fluids
Electrostatics and Magnetism
Circuits
Waves and Sound
Light and Optics
Atomic and Nuclear Phenomena
Mathematics
Reasoning About the Design and Execution of Research
Data-Based and Statistical Reasoning
High-Yield Cheat Sheet

Progress

Early access Want this book as a guided course? Placement diagnostics, interactive lessons, and checkpoints, built on these books.

Original study material © Premier Exam Prep — https://premierexamprep.com/mcat/books/physics-and-math/data-based-and-statistical-reasoning · pep-wm-0bfb22d3

MCAT Books › Physics and Math › Data-Based and Statistical Reasoning

Chapter 12: Data-Based and Statistical Reasoning

Updated Mar 26, 2026

Read the entire chapter on one page Every section in order, with the sidebar tracking where you are as you scroll.

→

🎯 Diagnostic: Test Your Starting Level 24 questions (2 per section). No prior reading required, see what you already know.

Aim to answer every question before checking. Missed questions point you to the sections you need most.

1. (12.1) A histogram:

A) Shows correlation between variables
B) Lists raw values
C) Is the same as a scatter plot
D) Displays the frequency distribution of a single continuous variable (each bar represents a bin)

D. Good for seeing shape, skewness, and gaps in a dataset.

2. (12.1) A scatter plot:

A) Shows the relationship between two variables (each point is one observation)
B) Displays only a single variable
C) Cannot reveal correlation
D) Replaces histograms

A. The shape of the cloud reveals correlation (positive, negative, none) and nonlinearity.

3. (12.2) The mean of a dataset is:

A) Always the largest value
B) The sum of values divided by the number of values; sensitive to outliers
C) The middle value
D) The most frequent value

B. Report median instead when outliers or a skewed distribution are present.

4. (12.2) The median is:

A) The mean
B) The mode
C) The middle value when data are ordered; less affected by outliers than the mean
D) The largest value

C. Income is usually reported as median because of extreme high outliers.

5. (12.3) The standard deviation measures:

A) Central tendency
B) The mode
C) The mean
D) The typical spread of data around the mean; bigger SD means more variability

D. Same units as the data themselves.

6. (12.3) Variance is:

A) The square of the standard deviation (in squared units)
B) The same as the mean
C) The median
D) The range

A. Often written σ² (population) or s² (sample).

7. (12.4) The normal distribution (bell curve):

A) Is highly skewed
B) Is symmetric about the mean; 68% of data falls within 1 SD, 95% within 2 SD, 99.7% within 3 SD (empirical rule)
C) Has a U-shape
D) Has no defined shape

B. Many natural phenomena (heights, test scores) approximate a normal distribution.

8. (12.4) The central limit theorem says:

A) It is irrelevant
B) Only applies to normal data
C) The distribution of sample means approaches normal as sample size grows, regardless of the underlying population
D) Proves all data is normal

C. Why so many statistical tests rely on normal approximations for the mean.

9. (12.5) The z-score of a data point is:

A) Always 0
B) The p-value
C) The count of standard deviations from the mode
D) z = (x - μ) / σ, the number of standard deviations x lies from the mean

D. Standardizes any normal variable to a standard normal (μ = 0, σ = 1).

10. (12.5) A data point with z = 2:

A) Is 2 standard deviations above the mean (roughly the 97.5th percentile in a normal distribution)
B) Equals the mean
C) Equals twice the mean
D) Has value 2 × σ

A. About 2.5% of normal data lies above z = 2.

11. (12.6) A 95% confidence interval means:

A) 95% of individual data points lie within the interval
B) If we repeated the experiment many times, 95% of the constructed CIs would contain the true parameter
C) There is a 95% chance that the true parameter lies in any particular interval
D) The parameter is fixed at 95%

B. Common misinterpretation of CIs is answer C, which is not technically correct under the frequentist framework.

12. (12.6) A larger sample size generally:

A) Widens the confidence interval
B) Reduces precision
C) Narrows the confidence interval (standard error scales with 1/√n)
D) Has no effect

C. Quadrupling n halves the standard error.

13. (12.7) The null hypothesis H₀:

A) Proves the research hypothesis
B) Is a directional claim
C) Is the predicted effect
D) Typically says "no effect" or "no difference"; we test whether evidence is strong enough to reject it

D. You can reject or fail to reject H₀, but you never "prove" H₀.

14. (12.7) The alternative hypothesis H₁:

A) Is the claim we hope to support (there IS an effect); evidence against H₀ supports H₁
B) Is the null
C) Is always false
D) Is irrelevant

A. Can be one-tailed (directional) or two-tailed.

15. (12.8) A p-value is:

A) The probability the null is true
B) The probability of getting the observed (or more extreme) result ASSUMING H₀ is true; small p suggests data would be unlikely under H₀
C) Always 0.05
D) The same thing as the significance level

B. Common misinterpretation: a small p does NOT say the alternative is true with probability 1 - p.

16. (12.8) A p-value of 0.02 with α = 0.05 means:

A) The null is proven true
B) H₀ is definitely true
C) H₀ is rejected; the result is statistically significant at the 5% level
D) H₀ cannot be rejected

C. p < α ⇒ reject H₀. The exact value (0.02 vs. 0.049) is itself informative.

17. (12.9) A Type I error occurs when:

A) The null is correctly rejected
B) The null is not rejected when it should be
C) No decision is reached
D) H₀ is true but wrongly rejected (false positive); its probability is α

D. This is why controlling α at, say, 0.05 limits the false-positive rate.

18. (12.9) A Type II error occurs when:

A) H₀ is false but not rejected (false negative); its probability is β; statistical power = 1 - β
B) H₀ is rejected when true
C) No hypothesis is stated
D) Analysis is skipped

A. Larger samples or bigger real effects reduce β and thus boost power.

19. (12.10) Correlation between two variables:

A) Always implies causation
B) Measures the strength and direction of a linear association (r), but does not establish causation
C) Proves causation
D) Is only positive

B. Correlations can arise from reverse causation, confounders, or chance.

20. (12.10) Establishing causation usually requires:

A) Just a strong correlation
B) Chance alone
C) Controlled experiments with randomization and blinding, plus a plausible causal mechanism; observational data alone is insufficient
D) Non-controlled data

C. Hill's criteria offer additional guidelines for inferring causation from observational evidence.

21. (12.11) A t-test compares:

A) Three or more groups
B) A single group against itself
C) Non-normal data exclusively
D) Means of two groups (one-sample or two-sample) and assumes approximately normal data for small samples

D. Student's t-distribution handles small samples; as n grows it converges to the normal.

22. (12.11) ANOVA is used when:

A) Comparing means across three or more groups
B) Testing correlation
C) Computing confidence intervals
D) Calculating p-values directly without any test

A. ANOVA tells you whether group means differ, but not which pair. Post-hoc tests (Tukey HSD) are needed for pairwise comparisons.

23. (12.12) Statistical significance:

A) Is the same as practical significance
B) Tells us only that the data are unlikely under H₀ (p < α); it does NOT by itself say the effect is large or meaningful
C) Always implies a real, useful effect
D) Ensures replication

B. With large enough n, tiny and uninteresting effects can become "statistically significant."

24. (12.12) Effect size:

A) Is the p-value
B) Is the sample size
C) Quantifies the magnitude of an effect (e.g., Cohen's d, odds ratio); independent of sample size; crucial for clinical interpretation
D) Is irrelevant

C. Report effect size alongside p-values to tell the reader how much the treatment actually does.

Previous Attempts

You are reading a research passage on the MCAT. A table of patient data, a graph with error bars, and a paragraph full of p-values and confidence intervals stare back at you. If your first instinct is to panic, you are not alone. Many test-takers spend years studying amino acids and kinematics, yet arrive on test day feeling unprepared for the statistics questions that appear across every section of the exam.

Here is the good news: the MCAT does not test you like a statistics course. You will never run a t-test by hand or derive a formula for standard deviation. Instead, you need to interpret data that other people have already analyzed. Can you read a graph accurately? Do you understand what a p-value actually means? Can you tell the difference between correlation and causation? These are the skills the MCAT rewards, and they are completely learnable.

This chapter covers the full range of data and statistics content you will encounter on test day. We start with how data is presented (tables, graphs, and common visual traps), move through descriptive statistics (central tendency and variability), then tackle the normal distribution, z-scores, and confidence intervals. The second half covers the logic of hypothesis testing, p-values, error types, correlation vs. causation, and an overview of when to use common statistical tests. We finish by examining the critical distinction between statistical significance and clinical significance - a favorite MCAT topic that trips up students who have never thought carefully about what “significant” really means.

💡 Real World

Statistics is the language of evidence-based medicine. Every drug approval, every clinical guideline, and every public health recommendation is built on statistical reasoning. When a doctor says "this treatment works," what they really mean is "a well-designed study showed a statistically significant effect with a clinically meaningful effect size." By the end of this chapter, you will understand exactly what that sentence means - and you will be able to evaluate whether the evidence actually supports the claim.

In This Chapter

12.1 Data Presentation
12.2 Central Tendency
12.3 Variability
12.4 Normal Distribution
12.5 Z-Scores
12.6 Confidence Intervals
12.7 Hypothesis Testing
12.8 P-Values
12.9 Type I and Type II Errors
12.10 Correlation vs. Causation
12.11 Statistical Tests Overview
12.12 Interpreting Results
Section Test

← PreviousSection Test Next →Data Presentation

Scroll or pinch to zoom

Full chapter