Give Readers the Data Behind the Model
A regression equation is easier to understand when readers know what its observations represent. Before interpreting a slope or describing a prediction, a report should tell readers who or what was observed, how the data were collected, and what the variables measure. A concise method and data summary paragraph supplies that foundation.
In “Communicating a Prediction and Its Accuracy,” you practiced reporting a fitted prediction with its input, residual scatter, and observed range. The paragraph in this tutorial comes earlier in a written report: it introduces the data to which the model applies. It does not interpret the regression results or claim that one variable causes another.
The individuals are the people or objects described by the data. Usually, each row represents one individual, though a row could also represent a defined unit such as a household, classroom, or route. Be precise: if one row summarizes a whole route, the individuals are the routes, not the passengers counted along them.
The variables are characteristics recorded for each individual. For a simple regression model, identify the explanatory variable \(x\), used to predict, explain, or account for variation in another variable, and the response variable \(y\), whose values the model predicts. A list of two variables by itself does not tell you which is explanatory and which is response. Their roles depend on the question the analysis is trying to answer.
Build the Paragraph from Specific Details
A useful summary answers a small set of questions in a sensible order. It names the cases, tells how they were selected or observed if that information is available, describes how the measurements were recorded, and makes the model’s prediction direction clear. Include the number of individuals and the time or setting when known. Use units for both variables, not just for the response.
Say what one observation or row represents, and give the number of individuals if known. Avoid a vague description such as “the data are about deliveries.”
State the known sampling, assignment, or observation method and the setting or time period. If the information is not provided, do not guess or imply random selection.
State what \(x\) and \(y\) measure and the units in which each was recorded. Make clear whether a measurement is, for example, a distance, a duration, a percentage, or an average.
Explain which variable is used to predict the other. Use the question being investigated to assign explanatory and response roles; do not infer a direction just from the order of a variable list.
These details help readers understand what the data can support. If the cases were volunteers, for instance, say so rather than calling them a random sample. If the data were recorded without assigning treatments, describe them as observed data. Be equally careful not to say that a variable was measured if the description only says it was reported by participants.
A good paragraph is accurate rather than elaborate. Include important collection details that are known, but do not invent a sampling frame, a measurement tool, or a reason for missing observations. If the method is unknown, describe the data that are given and be transparent about what cannot be specified.
Worked Example: Describe a Commute-Time Data Set
Worked Example: Name the Individuals and Prediction Direction
Situation. An invented school transportation project records data for 36 students who commute by bicycle. Students volunteered for the project and reported the one-way distance from home to school in kilometers and their usual one-way commute time in minutes. The analysis question is whether distance can be used to predict commute time.
Plan. Identify the individuals as the students, not the bicycles or trips. State the volunteer method and the source of the measurements accurately. The question gives the direction: distance is explanatory, and commute time is the response. Include the units for both.
Do. A concise method and data summary paragraph is: “The data describe 36 students who volunteered for an invented school transportation project. Each student reported the one-way distance from home to school, in kilometers, and their usual one-way bicycle commute time, in minutes. We use distance as the explanatory variable to predict commute time.”
Conclude. The paragraph makes clear what each row represents, how students entered the data set, what was recorded, and which variable the model predicts. It does not imply that the students were randomly selected or that a longer distance necessarily causes a particular student’s commute to take longer.
Variable Roles Come from the Question
The same two recorded variables can be used in different regression questions. For example, a data set might contain average passengers per bus route and average delay per route. If the question asks whether passenger count can predict delay, passenger count is explanatory and delay is the response. If the question instead asks whether delay can predict passenger count, the roles switch. The data listing does not choose the direction.
This distinction matters even when one direction seems more natural. A paragraph should report the direction actually used in the analysis, not quietly select one for the reader. If no prediction question or model direction is supplied, name the two variables and say that their explanatory and response roles cannot yet be determined.
Worked Example: Avoid Assuming the Regression Direction
Worked Example: Bus Routes and Average Delay
Situation. An invented data set has one row for each of 32 city bus routes. For each route, it records the average number of passengers during one weekday, in passengers per route, and the average delay, in minutes per route. First, consider the description without a stated prediction question. Then consider the question: “Can average passengers per route be used to predict average delay?”
Plan. The individuals are the 32 bus routes because each row summarizes one route. The two variables and units are specified, but their explanatory and response roles are not settled until the prediction question is stated.
Do. Without a prediction question, report: “The data set contains 32 city bus routes, with each route described by its average weekday passenger count, in passengers per route, and its average delay, in minutes per route. The description alone does not specify which variable is used to predict the other.” For the stated question, passenger count is explanatory and delay is the response.
Conclude. With the model direction supplied, a suitable summary is: “The data describe 32 city bus routes, with one observation per route. For each route, the data record the average number of passengers during one weekday, in passengers per route, and the average delay, in minutes per route. The analysis uses average passenger count to predict average delay.” The direction is justified by the question, not guessed from the variable names.
Use a Method Description That Matches the Data
The word method can refer to how cases were selected, how measurements were obtained, or whether a treatment was assigned. State only what is known. A sample of volunteers is not automatically a random sample. A record of two variables does not establish that one was controlled or manipulated. These differences affect how readers should understand the analysis, so choose verbs carefully.
For example, “researchers recorded” describes observation; “participants reported” describes self-report; and “students were randomly assigned to” describes an assignment process only if random assignment actually occurred. Do not upgrade a vague source statement into a more specific method. If a prompt says only that a data set contains 32 routes, it supports saying that there are 32 routes in the data—not claiming how those routes were chosen.
The units should also match what was measured. “Average delay, in minutes per route” is more informative than “delay,” and “average passenger count, in passengers per route” distinguishes a route-level average from the count of individual passengers. If a variable is a percentage, identify what the percentage is a percentage of when that information is available. Clear units prevent readers from confusing individual measurements with group summaries.
Worked Example: Summarize a Plant-Growth Project
Worked Example: Distinguish Cases, Measurements, and Units
Situation. In an invented greenhouse project, a class records data for 24 tomato seedlings grown in separate pots. For each seedling, students record the amount of water applied each day, in milliliters per day, and the plant’s height after four weeks, in centimeters. The question is whether daily water amount can be used to predict four-week height. The description does not say how seedlings were selected or whether water amounts were assigned by a controlled plan.
Plan. The individuals are the 24 seedlings, with one row per seedling. Daily water amount is explanatory because the question uses it to predict height; four-week height is the response. State the known measurements and units, and do not invent a random-sampling or experimental-assignment method.
Do. A suitable paragraph is: “The data describe 24 tomato seedlings grown in separate pots in an invented greenhouse project, with one observation per seedling. Students recorded the amount of water applied each day, in milliliters per day, and each plant’s height after four weeks, in centimeters. The analysis uses daily water amount as the explanatory variable to predict four-week plant height; the available description does not specify how seedlings were selected or how water amounts were determined.”
Conclude. This summary names the cases, setting, recorded variables, measurement timing, units, and model direction. It also makes the missing method information explicit instead of implying that water was randomly assigned or that the seedlings represent a broader population.
Common Mistakes and AP Exam Tips
- Calling a variable an individual. In a data set with one row per route, the routes are the individuals; passenger count and delay are variables measured for those routes. Tell the reader what a row represents.
- Assigning roles from a variable list. “Passengers and delay were recorded” does not specify which predicts which. Full-credit wording names the analysis direction, such as “passenger count is used to predict delay,” or explicitly notes that the roles cannot be determined without the question.
- Leaving out units. “Distance and time were recorded” is incomplete if the units are available. Name both measurements with units, such as kilometers and minutes.
- Inventing a collection method. Do not call cases randomly sampled, volunteers representative, or treatments randomly assigned unless the information supports those statements. A careful report states what is known and identifies important unknown details.
- Mixing up the observational unit and the people counted. If a data set contains one summary per route, the route is the individual even though its passenger count describes many riders. Match the individual to the unit represented by each row.
- Making a causal claim in the method paragraph. Saying that \(x\) is used to predict \(y\) does not establish that changing \(x\) causes a change in \(y\). Describe the model direction without overstating what the data-collection method supports.
For a strong response, make the row’s subject explicit, accurately describe how observations were obtained when that information is supplied, define each variable with its units, and state the prediction direction. If the direction is missing, do not guess. This brief paragraph gives readers the context needed to understand the regression analysis that follows.
Check Your Understanding
For each question, be precise about the individuals, variables, units, and any information that is not specified.
- A data set has one row per apartment building and records average monthly electricity use in kilowatt-hours and number of occupied units. What are the individuals, and what details would you still need before assigning explanatory and response roles?
- A project records distance in kilometers and travel time in minutes for 40 volunteer walkers. If the question is to predict travel time from distance, identify the explanatory and response variables and their units.
- A report says that 18 classrooms were included but does not explain how they were chosen. What can you say about the individuals, and what should you avoid claiming about selection?
- Why is “the study recorded water amount and plant height” insufficient to tell readers which variable the regression uses to predict the other?
- Write one sentence explaining why an observed association between two variables, by itself, does not show that the explanatory variable causes the response to change.