MATH 3 • STATISTICS & PROBABILITY

Communicating Statistical Conclusions — I can communicate conclusions with appropriate uncertainty language (likely, evidence suggests) at my level.

Learn to express data-driven conclusions with the precise language of probability and uncertainty.

Historical Context & Motivation

For most of human history, people made claims about the world in absolute terms — something was either true or false, certain or impossible. It wasn't until the development of probability theory and modern statistics that thinkers began to formalize how we talk about uncertainty. The story of statistical language is really the story of how humans learned to be honest about what they know — and what they don't.

1654
Birth of Probability Theory
Blaise Pascal and Pierre de Fermat exchange letters about gambling problems, laying the mathematical groundwork for describing uncertain outcomes with numbers.
1763
Bayes' Theorem Published
Thomas Bayes' posthumous paper introduces a framework for updating beliefs as new evidence arrives, changing how scientists express degrees of confidence.
1925
Fisher Formalizes Statistical Testing
Ronald Fisher introduces p-values and significance levels, creating a standardized vocabulary — 'statistically significant' — for communicating results.
2016
ASA Statement on P-Values
The American Statistical Association warns against misusing p-values and calls for clearer, more nuanced uncertainty language in published research.

Today, the ability to communicate conclusions with appropriate uncertainty is more important than ever. Headlines routinely claim that studies "prove" things, when the data actually only suggest them. How do you write a conclusion that's honest about what the data shows without overstating or understating the evidence? That is the question this lesson addresses.

Core Principles of Uncertainty Language

When you draw a conclusion from data, you are making an inference — a statement that goes beyond the raw numbers to say something meaningful about the world. Because data always involves variability, sampling error, and limitations, your conclusion should reflect how confident you are. This is where uncertainty language comes in: a set of carefully chosen words and phrases that signal the strength of the evidence behind your claim.

1

Correlation ≠ Causation

A relationship between two variables does not mean one causes the other. Use phrases like "is associated with" or "tends to accompany" instead of "causes."
2

Sample vs. Population

Conclusions from a sample may not perfectly represent the whole population. Phrases like "the data suggest" or "based on this sample" acknowledge this gap.
3

Strength of Evidence

Strong evidence warrants phrases like "it is likely that," while weaker evidence calls for "it is possible" or "there appears to be." Match your language to the data.
4

Context Matters

A conclusion should reference the study design, sample size, and any limitations. Saying "under these conditions" or "within this group" keeps your claim precise.
KEY TAKEAWAY
Think of uncertainty language like a weather forecast. A meteorologist doesn't say "it will rain tomorrow" — they say "there is a 70% chance of rain." Similarly, a good statistical conclusion doesn't declare absolute truth. It tells the audience how strong the evidence is, so they can weigh it themselves.

The Uncertainty Language Spectrum

Uncertainty language exists on a spectrum from very cautious to very confident. The diagram below organizes common phrases along that spectrum and connects them to the type of evidence that justifies each level of confidence. Study it carefully — you'll use these phrases throughout the rest of the lesson.

This spectrum shows how language shifts from cautious to confident as the strength of the statistical evidence increases. Notice that even the strongest statistical conclusions avoid the word "proves."

The key idea in the diagram is that the phrases you choose should be proportional to the strength of the evidence. A small survey of 20 students does not justify the same language as a controlled experiment with 2,000 participants. As the sample size grows, the study design improves, and the effect size becomes larger, you can move rightward on the spectrum toward more confident language — but you should almost never reach the word "proves."

The Mathematical Framework Behind Confidence

Uncertainty language isn't just a matter of taste — it's anchored in mathematical concepts you may already know. Two of the most important are confidence intervals and p-values. Understanding what these numbers mean helps you decide which uncertainty phrase to use in your conclusion.

CONFIDENCE INTERVAL
x̄ ± z* × (s / √n)
Where is the sample mean, z* is the critical value for the chosen confidence level (e.g., 1.96 for 95%), s is the sample standard deviation, and n is the sample size. A 95% confidence interval means: if we repeated this study many times, about 95% of the intervals would contain the true population parameter.

When reporting a confidence interval, the appropriate language is: "We are 95% confident that the true mean lies between [lower bound] and [upper bound]." Notice that this does not say "there is a 95% probability" — the true mean is a fixed number, not a random variable. The confidence refers to the process, not the specific interval.

P-VALUE INTERPRETATION
p-value = P(observed result or more extreme | H₀ is true)
The p-value is the probability of obtaining your observed data (or something more extreme) if the null hypothesis were true. A small p-value (typically < 0.05) suggests the observed result is unlikely under the null hypothesis. This does NOT mean the probability the null is true equals the p-value.
⚠️ Common Misconception
A p-value of 0.03 does not mean there is a 3% chance the null hypothesis is true. It means: assuming the null hypothesis is true, there is a 3% chance of seeing data this extreme. When writing your conclusion, say: "The results are statistically significant at the 0.05 level, providing evidence against the null hypothesis."

Matching Language to Evidence Strength

Now that you understand the math behind confidence, you need a practical guide for choosing the right words. The table below maps different statistical scenarios to the phrases that best communicate your findings. Use it as a reference whenever you write a statistical conclusion.

Reference table for matching statistical evidence to appropriate conclusion language.
ScenarioAppropriate LanguageExample
Small sample, observational study"It is possible…" / "The data hint…""Based on a survey of 15 students, it is possible that screen time is associated with lower sleep quality."
Moderate sample, correlation found"Evidence suggests…" / "It is likely…""Evidence from 200 respondents suggests that students who exercise regularly tend to report higher focus in class."
Large sample, controlled experiment, p < 0.05"The data strongly support…" / "There is convincing evidence…""A randomized experiment with 1,500 patients provides convincing evidence that the new drug reduces symptoms compared to a placebo."
No significant result found, p > 0.05"There is insufficient evidence to conclude…" / "The data do not support…""Based on these results, there is insufficient evidence to conclude that the tutoring method improves test scores."
Confidence interval reported"We are __% confident that…""We are 95% confident that the true average commute time for city residents is between 22 and 28 minutes."
Follow this flowchart when constructing a statistical conclusion. Start with your study type, consider sample size and significance, then choose the matching language tier. Always include the contextual elements shown at the bottom.

Worked Example: Writing a Statistical Conclusion

Let's walk through a realistic scenario step by step. Suppose a school administrator wants to know whether a new breakfast program improves students' math test scores. She conducts a study with 120 students randomly assigned to two groups: 60 who eat the school breakfast and 60 who don't. After 8 weeks, the breakfast group's average score is 78 and the non-breakfast group's average is 73. The p-value from a two-sample t-test is 0.02, and the 95% confidence interval for the difference in means is (1.8, 8.2).

Writing a Conclusion for the Breakfast Study
1
Step 1 — Identify the Study TypeStudents were randomly assigned to two groups. This is a randomized experiment, which means we can potentially make a causal claim — not just an association.
Study type: Randomized experiment → causal language is possible
2
Step 2 — Assess the SignificanceThe p-value is 0.02, which is less than the standard significance level of 0.05. This means the observed difference is statistically significant — it is unlikely to have occurred by random chance alone if the breakfast program had no effect.
p = 0.02 < 0.05 → statistically significant
3
Step 3 — Consider Sample Size and ContextThe total sample size is 120, which is moderate for a school-level study. The study lasted 8 weeks, which is a reasonable duration. However, results may not generalize to all schools or demographics. We should acknowledge this limitation.
n = 120, moderate sample, limited generalizability
4
Step 4 — Report the Confidence IntervalThe 95% confidence interval for the difference in means is (1.8, 8.2). Since this interval does not contain zero, it is consistent with the significant p-value. We are 95% confident the true difference in average scores lies between 1.8 and 8.2 points.
95% CI: (1.8, 8.2) — does not include 0
5
Step 5 — Write the Final ConclusionCombining all of the above, here is a well-crafted conclusion:
"Based on a randomized experiment with 120 students, there is convincing evidence that the school breakfast program leads to higher math test scores. The breakfast group scored an average of 5 points higher (95% CI: 1.8 to 8.2), and this difference was statistically significant (p = 0.02). While these results are promising, they may not generalize to schools with different student populations."
💡 Notice the Key Elements
The conclusion above includes: (1) the study type ("randomized experiment"), (2) the sample size ("120 students"), (3) the direction and size of the effect ("5 points higher"), (4) the confidence interval, (5) the p-value, and (6) a limitation. This is the template you should follow for every statistical conclusion.

Common Mistakes vs. Strong Conclusions

Even after learning the right phrases, students often fall into predictable traps. The table below compares weak or incorrect conclusion language with improved versions. Study each pair to develop a feel for what good statistical writing looks like.

Side-by-side comparisons of weak and strong statistical conclusions.
Weak / IncorrectWhat's WrongImproved Version
"This proves that coffee improves grades."Statistics never "prove" anything; also claims causation without specifying study type."Evidence suggests that coffee consumption is associated with higher grades in this sample."
"The data failed to show a relationship."Implies data has a duty to "succeed"; absence of evidence ≠ evidence of absence."There is insufficient evidence to conclude that a relationship exists between the two variables."
"The p-value is 0.03, so there is a 3% chance the null is true."Misinterprets the p-value — it measures the probability of the data, not the hypothesis."The result was statistically significant (p = 0.03), meaning data this extreme would be unlikely if the null hypothesis were true."
"Everyone should switch to this diet because our study showed weight loss."Overgeneralizes from one sample; makes a recommendation beyond the scope of the data."Among the 80 participants in this study, those on the new diet lost an average of 4 lbs more, suggesting the diet may be effective for similar populations."
KEY TAKEAWAY
Think of a statistical conclusion like testimony in a courtroom. A good witness reports what they observed and how confident they are, but doesn't pretend to be all-knowing. They say "I believe," "it appeared," or "in my view" — not "I am 100% certain." Your data is the witness. Your job is to let it testify accurately, neither exaggerating nor hiding what it shows.

Connecting to Advanced Statistical Communication

The uncertainty language skills you are building now form the foundation for more advanced statistical communication you'll encounter in AP Statistics, college research courses, and professional settings. The table below shows how the concepts from this lesson connect to their more advanced counterparts.

How today's uncertainty language connects to advanced statistical reasoning.
What You Learn NowWhere It Leads
Choosing phrases like "evidence suggests" vs. "proves"Writing formal research conclusions with effect sizes, confidence intervals, and Bayesian credible intervals
Distinguishing correlation from causationIdentifying confounding variables, designing controlled experiments, and evaluating causal inference methods
Interpreting p-values in plain languageDiscussing Type I/II errors, statistical power, and the replication crisis in science
Noting limitations and sample scopeWriting full "limitations" sections in research papers and assessing external validity

In college and professional settings, poor communication of statistical results can have real consequences. Medical researchers who overstate findings may lead to unsafe treatments. Journalists who confuse correlation with causation can mislead millions. By learning to communicate with appropriate uncertainty now, you're building a skill that matters far beyond your math class.

Practice Problems

PROBLEM 1CONCEPTUAL
Explain why statisticians avoid using the word "proves" when writing conclusions, even when the evidence is very strong. What phrase would be more appropriate?
PROBLEM 2BASIC CALCULATION
A researcher surveys 50 students and finds a correlation of r = 0.35 between hours of sleep and GPA. The p-value is 0.01. Which of the following is the best conclusion? (a) "Sleep causes higher GPAs." (b) "Evidence suggests that sleep and GPA are positively associated in this sample of students." (c) "There is insufficient evidence to conclude a relationship."
PROBLEM 3INTERMEDIATE
A randomized experiment tests whether a new fertilizer increases tomato yield. The experimental group (n = 40) averaged 12.3 lbs per plant, while the control group (n = 40) averaged 10.8 lbs per plant. The 95% confidence interval for the difference in means is (0.5, 2.5), and the p-value is 0.004. Write a complete statistical conclusion using appropriate uncertainty language.
PROBLEM 4APPLIED
A news headline reads: "Study Proves Smartphones Make Teens Anxious." The study was observational, surveying 300 teenagers about smartphone use and self-reported anxiety levels, with a correlation of r = 0.28 and p = 0.03. Rewrite the headline and a one-paragraph summary that accurately reflects the study's findings using appropriate uncertainty language.
PROBLEM 5CRITICAL THINKING
Two studies investigate whether music lessons improve spatial reasoning. Study A is a randomized experiment with 200 participants and finds a significant effect (p = 0.04, 95% CI for difference: 0.2 to 4.8 points). Study B is an observational study with 1,000 participants and finds a significant association (p = 0.001, 95% CI for difference: 1.5 to 3.5 points). Compare the two studies and write a conclusion that synthesizes both, explaining which aspects of each study support stronger or weaker claims.

Lesson Summary

Communicating statistical conclusions is about more than just reporting numbers — it requires choosing uncertainty language that honestly reflects the strength of your evidence. The words you choose exist on a spectrum: cautious phrases like "it is possible" suit small samples and observational data, moderate phrases like "evidence suggests" fit significant correlations and larger studies, and confident phrases like "convincing evidence" are reserved for well-designed experiments with significant results. Critically, you should never use the word "proves" in a statistical conclusion.

Every strong conclusion references the study type (observational vs. experimental), the sample size, and any limitations of the data. When a confidence interval or p-value is available, include it — but interpret it correctly. Remember: only randomized experiments support causal claims, and the distinction between correlation and causation is one of the most important ideas in all of statistics.

Varsity Tutors • Math 3 • Communicating Statistical Conclusions