Start With What Is Being Studied
A study description may mention people, groups, measurements, and conditions all at once. Before interpreting its results, identify what counts as one unit in the study, which variable is being used to explain or predict differences, and which outcome is recorded. If researchers deliberately assign a condition, identify that treatment too. These labels describe different roles; they are not interchangeable.
For example, a campus study might record sleep duration and reaction time for each student. The student is the unit, sleep duration may be the explanatory variable, and reaction time may be the response. If researchers instead assign students different sleep schedules and then measure reaction time, the schedule is a treatment. The names depend on what researchers did and what question they are asking—not just on the variables’ values.
Study Units and Recorded Measurements
The study unit is defined by the question and the structure of the data: what does one row of the study’s data represent? A unit may be an individual person, a household, a school, a blood sample, a pot of plants, or another entity. Do not choose a unit solely because it is the most noticeable object in the description.
A measurement is a recorded value for a unit; it is not itself the unit. A student’s reaction time of 310 milliseconds is a measurement, while the student is the unit. A student may provide several measurements, perhaps at different times, but repeated records do not automatically represent additional students. Likewise, the people who take a measurement are not necessarily the study units: researchers might record one average blood pressure for each clinic, in which case the clinics are the units represented by those averages.
A useful check is to ask, “What does one case in the analysis describe?” Then ask, “What was measured on that case?” The answers help separate the unit from the variables. A study can also collect information at more than one level, such as measurements on patients grouped within hospitals. In that situation, state clearly which level is relevant to the question being analyzed.
Worked Example: A Campus Sleep Study
Suppose researchers invite 120 students at a hypothetical university to report their average nightly sleep during the past week. Each student also completes a computer-based reaction-time task, and the researchers record the student’s average reaction time in milliseconds. Identify the units and variables if the question is whether students reporting more sleep tend to have faster reaction times.
Study units. Each student is one unit. The study has information on 120 students, so it has 120 student-level cases.
Explanatory variable. Average nightly sleep, measured in hours per night, is used to describe or predict differences in reaction time.
Response variable. Average reaction time, measured in milliseconds, is the outcome the question focuses on. Because lower reaction time means a faster response, the direction of the measurement should be described accurately.
Treatment? There is no treatment in this description: researchers record students’ existing sleep reports rather than assigning a sleep condition. Calling sleep the explanatory variable does not, by itself, make it a treatment. The labels identify roles in the question and data collection; they do not establish that sleep caused any observed reaction-time difference.
Explanatory Variables and Responses
The explanatory-response distinction is about the purpose of the investigation. The explanatory variable is the one used to account for, predict, or compare differences in the response. The response is the outcome whose variation is of interest. For a question about whether study time is related to quiz scores, study time can be explanatory and quiz score can be the response. For a question about whether quiz scores predict study time, the roles could be reversed.
These names do not necessarily describe a one-way causal relationship. In an observational study, the researcher may simply measure both variables and examine how they vary together. “Explanatory” means that a variable is being used in an explanatory role in the analysis; it does not prove that changing it would change the response. The study question and design matter when interpreting what the data can support.
A study can include several explanatory variables, several responses, or both. For instance, researchers could record sleep duration, caffeine consumption, and reaction time for each student. If the question concerns reaction time, both sleep duration and caffeine consumption might be explanatory variables, while reaction time is the response. Naming one variable as the response does not mean other recorded variables are irrelevant; it simply states the role they play in that question.
Treatments in Studies That Assign Conditions
A treatment is an assigned or applied condition, not merely a variable that happens to have categories. A treatment may be a specific dose, program, material, or level of a condition. The units receiving the conditions are the units to which the treatment is applied. When people or animals receive treatments, they are commonly called subjects; in other settings, such as an agricultural trial, the units may be pots or plots.
A treatment generally has a response measured after or during its application. For example, if researchers assign a study program and record a later quiz score, the program is the treatment and quiz score is a response. A variable’s role depends on the study description: a group label may represent an assigned treatment in one study, but a pre-existing characteristic recorded by researchers in another. Do not call every explanatory variable a treatment.
Worked Example: A Campus Sleep-Schedule Study
In a hypothetical study, researchers assign 40 student volunteers to follow one of two sleep schedules for five nights: a target of six hours in bed each night or a target of eight hours in bed each night. At the end, each student completes the same reaction-time task. Identify the study units, explanatory variable, treatments, and response.
Study units. Each of the 40 students is a study unit and a subject. There are 40 students, not 80 units, even though there are two schedule conditions.
Explanatory variable and treatments. The assigned sleep schedule is the explanatory variable. Its two treatment conditions are the six-hour target and the eight-hour target. The assigned target describes the treatment; it is not the same thing as the actual sleep duration a student achieves.
Response variable. The recorded reaction time, in milliseconds, is the response. Researchers could compare the reaction-time outcomes between the two assigned schedule conditions.
The description identifies the treatment because researchers deliberately assign a schedule. It does not specify how the assignment is carried out, so no claim about a particular assignment procedure follows from this example. Nor does the treatment label alone settle what conclusions a study can support.
When the Measurement Is Not the Treatment Unit
A study may apply a treatment to one unit and take measurements on several smaller objects associated with that unit. Be careful not to mistake the number of measurements for the number of units receiving the treatment. The treatment-application unit is the entity that receives a condition; the measured object or value may be at a different level.
Worked Example: A Fertilizer Trial
Imagine a hypothetical trial with 24 pots, each containing four seedlings. Researchers assign one of three fertilizer conditions to each pot and later measure the height increase of every seedling. The study question is whether the fertilizer condition is related to plant growth. Identify the units, treatment, and response, and explain how many pots receive a condition.
Study and treatment units. Each pot is a unit because the fertilizer is assigned and applied to the pot. There are 24 pots, and each pot receives one fertilizer condition.
Treatment. The fertilizer condition is the treatment. The three fertilizer conditions are the three versions applied to the pots; the particular product or concentration would need to be specified to describe them fully.
Measurements and response. Seedling height increase, measured in centimeters over the stated growing period, is the response. The researchers record four seedling measurements within each pot. If the question is framed at the pot level, they could summarize those four measurements—for example, by calculating a mean height increase for each pot. There would then be 24 pot-level summaries.
The arithmetic makes the levels clear: \(24 \times 4 = 96\) seedlings are measured, but the treatment is applied to 24 pots. The count of 96 measurements does not mean that 96 pots received a fertilizer condition. The treatment comparison concerns conditions applied to pots, so report the number of treated units as 24, not 96.
A Reliable Way to Parse a Study Description
Use the following sequence before interpreting a study. It keeps the unit, variable roles, and treatment distinct and makes it easier to notice when the description does not provide enough detail.
Identify the outcome or relationship the study is meant to investigate. This helps determine which variable is the response.
Complete the sentence, “One case in this study is one ___.” Specify whether that is a student, pot, school, or another entity.
Name each variable and include its meaning and units of measurement where relevant. Separate a measured value from the unit it describes.
Identify the response for the stated question, then name the variable or variables used to explain or predict differences in it.
If researchers assign or apply a condition, name its treatment levels and say which units receive them. If conditions are only observed, do not invent a treatment.
Why These Distinctions Matter
These labels prevent errors in both description and interpretation. If a report says “96 plants were studied,” but fertilizer was applied to 24 pots, a reader needs to know which count refers to measurements and which refers to units receiving treatment. Similarly, if sleep duration is recorded rather than assigned, describing it as a treatment changes the meaning of the study.
A concise description is often enough to clarify a study: “The units are 120 students; sleep duration in hours is the explanatory variable; reaction time in milliseconds is the response; no treatment was assigned.” Or: “The units are 24 pots; fertilizer condition is the treatment; seedling height increase is the response.” Naming the context and measurement units makes these statements more informative than simply writing “independent variable” and “dependent variable.”
Check Your Understanding
For each description, identify the study units, explanatory variable, response, and treatment if one is assigned.
- A hypothetical survey records each student’s weekly exercise hours and resting pulse rate. What are the units and the explanatory and response variables if the question is whether exercise hours are associated with pulse rate? Is exercise a treatment here?
- Researchers assign 30 pots to one of two watering schedules and measure the average plant height in each pot. What is the treatment, and what is one study unit?
- A researcher records one average wait time for each of 18 clinics. What is the unit represented by one recorded average: a patient or a clinic?
- In a study, researchers measure caffeine consumption and quiz score for each participant. Explain why calling caffeine consumption an explanatory variable does not by itself make it a treatment.
- A trial applies one condition to each of 12 trays, with five seedlings in every tray. How many units receive a condition, and how many seedling measurements are collected if every seedling is measured once?