Results Interpretation and Figure Analysis

Results versus Discussion

The Results section reports what was found; the Discussion section interprets what it means. Mixing them is the single most common error in reading scientific papers -- and the MCAT exploits this confusion systematically. The Results section should present data objectively: descriptive statistics, inferential test results, effect sizes, and figures/tables. It states 'the treatment group had significantly lower blood pressure (p = 0.03, d = 0.42)' -- this is a finding. The Discussion section says 'this suggests that the drug is an effective antihypertensive' -- this is an interpretation that goes beyond the data. MCAT questions often ask: 'Which of the following conclusions is supported by the data in Figure 1?' The correct answer stays within what the data actually show; wrong answers over-interpret by adding causal claims, generalizing beyond the sample, or drawing conclusions that require additional evidence. A passage may present a statistically significant but very small effect (d = 0.10) and then claim clinical importance in the Discussion. The MCAT expects you to recognize that statistical significance does not equal practical significance. Similarly, a non-significant result does not prove the null hypothesis -- it means the study failed to detect an effect, which could be due to low power rather than a true absence. The core skill: when reading a Results section, ask only 'what did they find?' When reading a Discussion, ask 'does the data actually support this interpretation or are they overreaching?' If an answer choice adds words like 'proves,' 'demonstrates,' or 'establishes causation,' it is likely an over-interpretation unless the study design actually warrants that language.

Figure Analysis

Figure analysis is the systematic extraction of information from data visualizations. Every figure has three questions to answer: (1) What is on each axis? (2) What comparison is being shown (between groups? across time? correlation between variables?)? (3) What do the error bars, p-values, or significance markers tell you about whether the effect is real? Common figure types on the MCAT include bar charts (comparing group means, often with error bars representing SD or SEM), line graphs (tracking change over time or dose-response relationships), scatterplots (showing correlations or individual data points), box plots (displaying median, quartiles, and outliers), and scientific diagrams (gel images, microscope images, anatomical schematics). Error bars are critical: if they represent standard deviation (SD), they show the spread of the data; if they represent standard error of the mean (SEM), they show precision of the mean estimate and are always smaller than SD. Non-overlapping SEM error bars strongly suggest a significant difference, but overlapping error bars do not rule out significance -- this is a common MCAT trap. Significance markers (, , *) directly indicate p-values (typically , p < 0.05; , p < 0.01; *, p < 0.001). When multiple comparisons are shown (e.g., a figure with several pairwise bars), check whether the asterisks correspond to specific comparisons or are globally applied -- unmarked bars may or may not be significant. For images (gels, blots, micrographs), look for band intensity, presence/absence of bands, and labeling of lanes/conditions. The MCAT often asks you to identify the control lane in a gel or determine which condition produced a particular result.

Graph Interpretation

Graph interpretation builds on figure analysis with a focus on extracting quantitative trends and relationships. The first step is always axis inspection: what variable is on the x-axis (independent) and y-axis (dependent)? What are the units? Is the scale linear or logarithmic? A log scale compresses large ranges and can make exponential relationships look linear -- if you miss this, you will misinterpret the trend. Next, identify the shape of the relationship: linear (constant rate of change), exponential (increasing rate of change, characteristic of population growth and radioactive decay), logarithmic (decelerating increase, characteristic of dose-response curves and enzyme saturation), sigmoidal (S-shaped, characteristic of cooperative binding and titration curves), or threshold/exponential (little effect until a critical concentration, then rapid change). Pay attention to intercepts, inflection points, and asymptotes -- these often carry biological meaning (Km in enzyme kinetics, carrying capacity in population biology, maximum response in pharmacology). When multiple lines or curves appear, check the legend: does each line represent a different condition, treatment group, or time point? Dose-response curves show effect magnitude against drug concentration; look for EC50 (concentration producing 50% of maximal effect) and Emax (maximum effect). Survival curves (Kaplan-Meier) plot the proportion of subjects surviving over time; look for separation between curves and median survival times. Standard curves are used to determine unknown concentrations from known standards -- interpolation from the linear portion is valid; extrapolation beyond the standard range is not. The MCAT frequently asks you to interpolate a value from a graph, identify which condition produced the strongest effect, or determine whether a relationship is positive, negative, or absent.

Table Analysis

Tables present numerical results in structured rows and columns. The first task is reading the title and column headers to understand the organization: typically, each row is an observation or condition, and each column is a variable or measurement. Key numbers to scan for: sample size (n) -- small n reduces statistical power and generalizability; means and standard deviations -- the mean tells you the central tendency, the SD tells you spread; p-values -- the probability that an effect at least as large as observed would occur if the null hypothesis were true; confidence intervals -- the range within which the true population parameter likely falls (usually 95% CI: 'we are 95% confident the true mean lies in this range'); and effect sizes -- measures of practical significance like Cohen's d (small = 0.2, medium = 0.5, large = 0.8). When comparing numbers across rows, check whether differences are meaningful: an effect can be statistically significant (p < 0.05) but too small to matter (d = 0.05). Tables often use superscript letters or asterisks for significance -- look for a footnote explaining the notation. A table that reports only means without any measure of variability (SD, SEM, CI) is incomplete -- you cannot assess whether differences reflect real effects or sampling noise. The MCAT tests this by presenting a table and asking: 'Based on Table 1, which conclusion is supported?' The correct answer will be a statement directly verifiable from the numbers, not an interpretation that requires assumption of causation or generalization. Also watch for tables that report group sizes but not individual-level data -- summary statistics can hide outlier-driven effects and non-normal distributions. If a table reports medians instead of means, the data is likely skewed -- this is an important clue about the distribution.

Statistical Reporting

Statistical reporting is the language through which results communicate their reliability. Complete reporting includes: the test statistic (t, F, chi-square, r), degrees of freedom, p-value (exact value preferred over '< 0.05'), effect size (Cohen's d, eta-squared, r-squared), and confidence intervals. The MCAT tests your ability to read this reporting and draw correct conclusions. A p-value of 0.04 and a p-value of 0.001 are both 'significant' at alpha = 0.05, but they convey very different strength of evidence. The p-value is NOT the probability that the null hypothesis is true -- it is the probability of observing data at least as extreme as the data obtained, assuming the null hypothesis is true. This distinction is one of the most tested concepts on the MCAT. When multiple comparisons are performed without correction (Bonferroni, Tukey), the familywise error rate inflates -- if you run 20 independent tests at alpha = 0.05, you expect one false positive by chance alone. The MCAT expects you to flag this. Confidence intervals provide more information than p-values alone: a 95% CI that does not cross zero (for a difference) or one (for a ratio) corresponds to p < 0.05, but the CI also tells you the plausible range of the effect magnitude. A wide CI indicates low precision (small sample); a narrow CI indicates precise estimation. Statistical power is the probability of detecting a true effect -- it depends on sample size, effect size, and alpha level. Underpowered studies are vulnerable to Type II errors (false negatives). When a study reports 'no significant difference,' ask: was the study adequately powered to detect a meaningful difference? If n = 10 and the effect was small, non-significance is unsurprising even if a real effect exists. The MCAT may also test recognition of inappropriate statistical choices: using a t-test for ordinal data, applying parametric tests to highly skewed data without transformation, or treating repeated measures as independent observations.

How it works

Approach every MCAT figure and table with a three-pass method. Pass 1: Orientation -- read the title, axis labels, legend, and column headers. What variables are being displayed? What comparison is the figure/table designed to make? Pass 2: Extraction -- identify the key numbers: group means, p-values, error bar types, sample sizes. Note the direction and magnitude of any differences. Is the difference visually obvious or barely there? Pass 3: Match to claim -- compare what the data actually show to what the passage text or answer choice claims. Does the claim stay within the data or does it over-interpret? Common over-interpretations flagged by the MCAT: claiming causation from a correlation, generalizing from a small/unrepresentative sample, interpreting non-significance as proof of no effect, and treating statistical significance as automatically meaningful (ignoring effect size). This three-pass method works for any data display the MCAT presents.

How it works

Approach every MCAT figure and table with a three-pass method. Pass 1: Orientation -- read the title, axis labels, legend, and column headers. What variables are being displayed? What comparison is the figure/table designed to make? Pass 2: Extraction -- identify the key numbers: group means, p-values, error bar types, sample sizes. Note the direction and magnitude of any differences. Is the difference visually obvious or barely there? Pass 3: Match to claim -- compare what the data actually show to what the passage text or answer choice claims. Does the claim stay within the data or does it over-interpret? Common over-interpretations flagged by the MCAT: claiming causation from a correlation, generalizing from a small/unrepresentative sample, interpreting non-significance as proof of no effect, and treating statistical significance as automatically meaningful (ignoring effect size). This three-pass method works for any data display the MCAT presents.

Comparisons

• B/B (Experimental passages): Biology passages routinely present bar charts of enzyme activity, line graphs of reaction rates, and gel images. Apply the three-pass method: identify the comparison (treated vs. control, different concentrations), extract the direction and significance, and check whether the passage's conclusion overreaches the data.
• C/P (Chemical and Physical Sciences): Chemistry and physics passages present calibration curves, absorbance spectra, and concentration-time graphs. Axis units matter critically here -- a log scale on a concentration axis changes the interpretation entirely. Interpolation from standard curves is a common task.
• P/S (Psychology and Sociology): P/S passages use tables of survey results, bar charts of group differences, and scatterplots of correlations. Pay special attention to whether reported group differences are statistically significant (check for p-values or CI overlap) and whether operational definitions match what the graph displays.
• RM-008 (Descriptive Statistics): Every figure and table is a descriptive display -- box plots show median/IQR, bar charts show means/SD, scatterplots show correlation. RM-012 applies these display types to passage-based interpretation.
• RM-009 (Inferential Statistics): The p-values, error bars, and significance markers in figures and tables represent inferential statistics. RM-012 teaches you to read these in context; RM-009 teaches the underlying statistical logic.
• RM-010 (Correlation and Regression): Scatterplots and regression lines are common MCAT figures. RM-012 covers how to read them; RM-010 covers what the r and R-squared values mean.
• RM-013 (Identifying Conclusions): RM-012 trains you to extract what the data actually show; RM-013 builds on this to identify when conclusions are justified versus when they overreach.

Common confusions

• Confusing SD and SEM error bars: SEM is always smaller than SD (SEM = SD / sqrt(n)). Non-overlapping SEM bars suggest significance at p < 0.05, but overlapping SEM bars do NOT rule out significance. Non-overlapping SD bars suggest the groups are very different, but statistical significance still requires a formal test. The MCAT may show you error bars without specifying which type they are -- check the figure caption.
• Missing the log scale: A relationship that looks linear on a log scale is actually exponential. The MCAT often hides this in the axis label ('log[substrate]') or tick marks (1, 10, 100, 1000). If you miss the scale, you will misinterpret the relationship entirely.
• Over-interpreting p > 0.05 as 'no effect': A non-significant result means the study failed to detect an effect, not that the effect is zero. Check the sample size -- if n is small, the study may simply be underpowered. The MCAT expects you to distinguish 'no evidence of an effect' from 'evidence of no effect.'
• Causation from correlation in scatterplots: A strong correlation (r = 0.8) in a scatterplot does not establish that the x-axis variable causes the y-axis variable. Reverse causation or a third-variable confound are always alternatives. The MCAT will ask: 'Which of the following conclusions is supported by Figure 2?' and the correct answer will describe an association, not a causal claim.
• Extrapolating beyond the data range: Standard curves are valid for interpolation (estimating values within the measured range) but not extrapolation (estimating values beyond the highest or lowest standard). A question asking you to estimate a concentration of 500 ng/mL from a standard curve that only goes to 200 ng/mL is testing this trap.
• Ignoring asterisk footnotes: Tables often use asterisks or superscript letters to denote significance, but the notation may compare each group to a specific reference, not to every other group. An asterisk on group B may mean 'significantly different from control,' not 'significantly different from group C.' Always read the footnote.
• Confidence intervals that include the null value: A 95% CI for a difference that includes zero (e.g., -0.3 to 1.7) corresponds to p > 0.05. A CI for a ratio that includes 1.0 corresponds to non-significance. The MCAT expects you to recognize this equivalence.

Quick review

• Results section: What was found (objective, data-focused). Discussion section: What it means (interpretation, speculation).
• Three-pass figure method: (1) Orientation (axes, legend, headers), (2) Extraction (key numbers, direction, magnitude), (3) Match to claim (does the data support the conclusion?).
• SD error bars: Show spread of the data. SEM error bars: Show precision of the mean; always smaller (SEM = SD/sqrt(n)).
• Non-overlapping SEM bars strongly suggest p < 0.05. Overlapping error bars do NOT rule out significance.
• Log scale: Check tick marks. 1, 10, 100, 1000 = log scale. Linear-looking trend on log scale = exponential relationship.
• Standard curves: Interpolate within the range (valid). Extrapolate beyond the range (invalid).
• p-value: Probability of data this extreme IF null hypothesis is true. NOT the probability that the null is true.
• 95% CI that includes zero (difference) or 1.0 (ratio) = non-significant result (p > 0.05).
• Statistical significance (p < 0.05) does not equal practical significance (check effect size).
• Non-significant result (p > 0.05) does not prove the null -- the study may be underpowered (small n, small effect).
• Multiple comparisons without correction inflate Type I error rate. Bonferroni: divide alpha by number of tests.
• Correlation does not imply causation: scatterplots show association, not causal direction.
• Tables must report variability (SD, SEM, or CI) -- means alone cannot distinguish real effects from noise.
• Check asterisk footnotes in tables: significance markers may compare to a specific reference, not every other group.
• Box plots: Show median (line), IQR (box), range (whiskers), and outliers (individual points).