Communicating Statistical Conclusions

SECTION 1

Historical Context & Motivation

Statistics has always been about drawing conclusions from data, but for centuries, people struggled with how to communicate those conclusions honestly. Early uses of data—counting crops, tracking trade, recording deaths during plagues—were straightforward tallies. The real challenge emerged when people started using data to make claims about the world. Without careful language, statistical findings could easily mislead, exaggerate, or be taken out of context.

As statistical methods matured, so did the need for precise language. Researchers learned the hard way that saying "this drug works" is very different from saying "this drug showed a statistically significant reduction in symptoms in a sample of 200 adults over six weeks." The history of statistics is filled with moments where poor communication led to bad decisions, public confusion, and even harm.

1854

John Snow's Cholera Map

John Snow used data visualization and careful reasoning to argue—not prove—that cholera spread through contaminated water, not air. His cautious language helped convince skeptics without overclaiming.

1925

Fisher Introduces Significance Testing

Ronald Fisher formalized the concept of statistical significance, giving researchers a standard way to describe how confident they could be in their results—and creating the need for qualifiers like "significant at the 0.05 level."

1954

"How to Lie with Statistics" Published

Darrell Huff's famous book exposed how misleading graphs, biased samples, and vague language could distort statistical conclusions, sparking public awareness about the importance of honest communication.

2016

ASA Statement on P-Values

The American Statistical Association issued a landmark statement warning against misinterpreting p-values, emphasizing that statistical conclusions require context, qualifiers, and transparency about limitations.

The central question this lesson addresses is: How do we translate numbers and calculations into written conclusions that are accurate, honest, and appropriately cautious? In a world overflowing with data claims—from social media posts to news headlines—this skill is more important than ever.

SECTION 2

Core Principles of Statistical Communication

Communicating statistical conclusions well requires more than just reporting numbers. You need to frame your findings within the context of how the data was collected, acknowledge what the data cannot tell you, and use language that reflects the degree of certainty your analysis actually supports. The following principles form the foundation of strong statistical communication.

Use Qualifiers

Words like "suggests," "tends to," "is associated with," and "approximately" communicate that statistical results are based on probability, not absolute proof.

State the Context

Always describe who was studied, how the data was collected, and when the study occurred. A conclusion drawn from 50 teenagers in one city may not apply to all teenagers everywhere.

Distinguish Correlation from Causation

Unless the study was a randomized controlled experiment, avoid claiming one variable causes another. Observational data can only show association.

Acknowledge Limitations

Every study has limitations—potential bias, small sample size, confounding variables. Honest communication means naming these openly.

Report Key Statistics

Include relevant numerical details such as sample size (n), measures of center, spread, and confidence levels so others can evaluate your claim.

✦ KEY TAKEAWAY

Think of a statistical conclusion like a weather forecast. A meteorologist doesn't say "it will rain tomorrow"—they say "there is a 70% chance of rain in the metro area between 2 p.m. and 8 p.m." The forecast includes the degree of certainty (70%), the scope (metro area), and the time frame (2–8 p.m.). Your statistical conclusions should work the same way—never overpromise, always give context.

SECTION 3

Anatomy of a Well-Communicated Conclusion

A well-written statistical conclusion is made up of several intentional parts. The diagram below breaks down the structure of a strong conclusion statement, showing how each component adds honesty and precision. Notice how every piece serves a purpose: there is no room for vague or sweeping claims.

The top box shows a well-structured conclusion with all four essential components labeled: context, sample details, qualifiers, and a specific finding. The bottom box illustrates a weak version that lacks every one of these elements.

As you can see, the strong conclusion is longer than the weak one—and that's the point. Precision takes more words. Every added phrase ("based on a survey," "tended to," "about 7 points") makes the statement more honest and more useful. When you communicate statistical conclusions, your goal is not to be brief; it is to be accurate and transparent.

SECTION 4

The Language Framework for Statistical Conclusions

While communicating statistical conclusions is primarily a writing skill, there is a logical structure you can follow like a formula. This section introduces a framework you can apply every time you write a conclusion from data. Think of it as a template that ensures you never accidentally overclaim or leave out important context.

The Conclusion Sentence Framework

CONCLUSION FRAMEWORK

Conclusion = Context + Qualifier + Finding + Limitation

Context = study type, sample, population; Qualifier = hedge words reflecting certainty level; Finding = specific numerical result; Limitation = known weaknesses of the study

Qualifier Language by Study Type

Match your language to the type of study that produced the data.

Study Type	Appropriate Language	Language to Avoid
Observational Study	"is associated with," "tends to," "there appears to be a relationship"	"causes," "proves," "leads to"
Randomized Experiment	"the treatment resulted in," "caused a measurable change," "evidence suggests a causal effect"	"definitely causes," "proves beyond doubt"
Survey / Sample	"approximately," "in this sample," "an estimated 60% reported"	"everyone thinks," "all people do," "exactly 60%"
Census / Population Data	"the data show," "60% of residents reported," "according to the census"	"this will always be true," "proves a universal trend"

GENERALIZABILITY RULE

Generalize only to the population the sample represents

If you survey 200 students at your school, your conclusion applies to students at your school—not all students in your state, country, or the world. The sample defines the boundary of your claim.

⚠️ Causation Checkpoint

Before writing a conclusion, ask yourself: Was there random assignment to groups? If yes, you may carefully use causal language. If no—or if you are unsure—stick with association language. This single question prevents the most common error in statistical communication.

SECTION 5

Common Errors in Statistical Communication

Even experienced writers make mistakes when reporting data. Understanding the most common errors will help you recognize and avoid them. The diagram below maps out the five most frequent pitfalls, along with what they look like in practice and how to fix them.

Each card shows a common error (in the default text color) alongside its corrected version (in green). The checklist at the bottom provides a quick review tool you can use before submitting any statistical conclusion you write.

Review this checklist every time you write a statistical conclusion. Over time, these checks will become second nature. The most important habit to build is pausing before you write and asking: "Does my language match the strength of the evidence?"

SECTION 6

Worked Example: Writing a Statistical Conclusion

Let's walk through a complete example. You'll see how to take raw study information and turn it into a well-written conclusion that meets all the criteria from the framework.

📋 Scenario

A teacher wants to know whether listening to music while studying affects quiz scores. She randomly assigns 60 students in her two algebra classes: 30 students study with background music, and 30 study in silence. After one week, the music group averaged 78% on the quiz, and the silence group averaged 83%. The standard deviation for both groups was about 10 percentage points.

Building the Conclusion Step by Step

Step 1 — Identify the Study TypeStudents were randomly assigned to two groups (music vs. silence). This is a randomized experiment, which means we can consider using cautious causal language—but we must still be careful about scope.

Study type: randomized experiment

Step 2 — Record the Sample DetailsThe sample includes 60 students from two algebra classes at one school. This is a relatively small sample from a specific school, so we should not generalize to all students everywhere.

Sample: n = 60 students, two algebra classes, one school

Step 3 — State the Finding with NumbersThe silence group scored an average of 83%, while the music group scored an average of 78%. The difference is 5 percentage points. We should include this specific number rather than saying something vague like "scored higher."

Finding: 5 percentage point difference (83% vs. 78%)

Step 4 — Choose Appropriate QualifiersBecause this was a randomized experiment, we can say the music condition "resulted in" or "was associated with" lower scores. However, the difference is modest and the sample is small, so phrases like "on average" and "in this sample" are still important. We should avoid saying music "definitely hurts" performance.

Qualifiers: "on average," "in this experiment"

Step 5 — Acknowledge LimitationsThe study involved only two classes at one school, the experiment lasted just one week, the type of music was not specified, and individual study habits were not controlled. These are important limitations to state explicitly.

Limitations: small sample, one school, one week, unspecified music type

Step 6 — Write the Final ConclusionNow combine all the pieces into a cohesive paragraph.

"In a randomized experiment involving 60 algebra students at one high school, students who studied in silence scored an average of 5 percentage points higher on a quiz (83% vs. 78%) than students who studied with background music. Because students were randomly assigned to groups, the results suggest that background music may have a negative effect on quiz performance for this group. However, the small sample size, short duration, and the fact that only one school was studied limit the generalizability of these findings."

✦ WHY THIS WORKS

Notice that the final conclusion is three sentences long and includes every component from our framework: context (randomized experiment, 60 students, one school), specific findings (5 percentage points, 83% vs. 78%), cautious causal language ("suggest," "may have"), and explicit limitations. It is honest without being dismissive—the data still tells a meaningful story.

SECTION 7

Comparing Strong and Weak Conclusions

One of the best ways to improve your own writing is to compare examples side by side. The table below shows three different scenarios and contrasts a poorly written conclusion with a well-written one. Study the differences carefully—notice which specific words and phrases make each strong version better.

Notice how the strong conclusions are longer, more specific, and explicitly state what the data can and cannot show.

Scenario	Weak Conclusion	Strong Conclusion
Survey: 150 students at a school asked about screen time and sleep	"Screen time causes sleep problems in teenagers."	"In a survey of 150 students at Jefferson High, those who reported more than 3 hours of daily screen time tended to report fewer hours of sleep. However, because this was observational data based on self-reports, a causal relationship cannot be established."
Experiment: 40 plants given fertilizer A vs. fertilizer B over 4 weeks	"Fertilizer A is better than Fertilizer B."	"In a controlled experiment with 40 tomato plants over 4 weeks, plants given Fertilizer A grew an average of 2.3 cm taller than those given Fertilizer B. While the random assignment supports a causal claim, the results may not apply to other plant species or growing conditions."
Census data: City records on bike lanes and traffic accidents over 5 years	"Bike lanes reduce accidents everywhere."	"City traffic data from 2018–2023 show that intersections with dedicated bike lanes had 28% fewer cyclist injuries than those without. These findings describe this particular city and time period, and other factors such as traffic volume may also play a role."

✦ KEY TAKEAWAY

Think of a statistical conclusion like testimony in a courtroom. A good witness says exactly what they saw, when they saw it, and acknowledges what they couldn't see from where they stood. A bad witness exaggerates and claims to know things they don't. Your data is the witness—your job is to let it testify honestly and within its limits.

SECTION 8

Connection to Advanced Statistical Reasoning

The communication skills you are learning now form the foundation for more advanced statistical reasoning. As you move into AP Statistics, college courses, or data science, the stakes get higher and the language gets more precise—but the core principles stay the same. The table below shows how the concepts in this lesson connect to ideas you'll encounter later.

Your current skills are the building blocks for the more formal statistical reasoning used in college and professional research.

This Lesson	Advanced Version
Using qualifiers like "suggests" or "tends to"	Reporting confidence intervals (e.g., "We are 95% confident the true mean falls between 74 and 82")
Stating the sample size and who was studied	Calculating margin of error and discussing statistical power
Distinguishing observational studies from experiments	Understanding confounding variables, lurking variables, and Simpson's Paradox
Saying "the data suggest" instead of "the data prove"	Interpreting p-values and understanding Type I and Type II errors
Noting limitations of a single study	Evaluating meta-analyses that combine results from many studies

The good news is that if you master the habits in this lesson—using qualifiers, providing context, respecting the scope of your data, and naming limitations—you will already be ahead of many college students when you encounter formal inference. The language of statistics becomes more technical, but the underlying principle never changes: say what the data shows, say how confident you are, and say what the data cannot tell you.

SECTION 9

Practice Problems

PROBLEM 1 — CONCEPTUAL

A classmate writes: "Our survey proved that students who play sports get better grades." Identify at least two problems with this conclusion and explain why each is an issue.

PROBLEM 2 — BASIC CALCULATION

A researcher surveys 200 adults in a city and finds that 120 of them (60%) support a new park. Write a one-sentence conclusion that includes the sample size, the result, and an appropriate qualifier.

PROBLEM 3 — INTERMEDIATE

A school nurse conducts an observational study and finds that students who eat breakfast score an average of 6 points higher on morning tests (n = 80). Write a complete conclusion paragraph that includes context, the finding, a qualifier, and at least one limitation.

PROBLEM 4 — APPLIED

A local news headline reads: "New Study: Drinking Coffee Makes You Live Longer!" The study was an observational study that followed 5,000 adults over 10 years and found that coffee drinkers had a 12% lower mortality rate. Rewrite the headline and write a two-sentence summary that accurately reflects the study's design and findings.

PROBLEM 5 — CRITICAL THINKING

Two students analyze the same data set about study hours and test scores. Student A writes: "Students who study more get higher scores." Student B writes: "Among the 45 juniors surveyed at our school, there was a moderate positive correlation (r ≈ 0.58) between self-reported weekly study hours and semester exam scores. This suggests that more study time is associated with higher scores in this group, though other factors—such as prior knowledge, tutoring, and course difficulty—were not measured." Compare the two conclusions. Which is stronger, and why? What, if anything, would you add to Student B's conclusion?

SUMMARY

Lesson Summary

Communicating statistical conclusions requires more than reporting numbers—it demands honest, precise language that reflects the actual strength of the evidence. Every strong conclusion includes four essential components: context (describing the study type and sample), qualifiers (using words like "suggests," "tends to," and "approximately" to match the certainty level), specific findings (including actual numbers such as means, percentages, or correlation values), and explicit limitations (acknowledging small sample sizes, potential confounders, or lack of random assignment).

The most critical rule to remember is the causation checkpoint: only use causal language when the study involved random assignment to treatment groups. For observational studies and surveys, you must use association language instead. By applying the Conclusion Framework—Context + Qualifier + Finding + Limitation—you will consistently produce conclusions that are accurate, transparent, and credible. These skills are not just for math class; they are essential for interpreting data in the news, in science, and in everyday decision-making.