Uniform Support and Parameters

What “Uniform” Means

A continuous uniform model is useful when a measurement can fall anywhere within a specified range and the model assigns the same likelihood to ranges of the same length. The key information is the interval’s two endpoints: they determine both the possible values and how spread out that interval is. The model is an assumption about the variable, not a claim that every real-world process is automatically uniform.

For example, a hypothetical process might produce a completion time that is modeled as equally likely across a stated interval of times. Before calculating anything, identify what is being measured, the units, and the lower and upper endpoints. A mismatch in units or a mistaken endpoint changes the distribution being described.

Definition: A continuous random variable $X$ has a uniform distribution on the interval from $a$ to $b$, written $X\sim\operatorname{Uniform}(a,b)$, when $a$ and $b$ are finite real numbers with $a<b$, and its density is constant on that interval and zero outside it: $$ f_X(x)= \begin{cases} \dfrac{1}{b-a}, & a\le x\le b,\\ 0, & \text{otherwise}. \end{cases} $$ The interval from $a$ to $b$ is the support; $a$ is the lower endpoint parameter and $b$ is the upper endpoint parameter. The model requires $a<b$. The value $b-a$ is the interval’s length.

The parameters describe the range, not a pair of likely outcomes. For instance, $X\sim\operatorname{Uniform}(4,10)$ means that the modeled values range from 4 to 10 in the stated units. It does not mean that $X$ can take only the values 4 and 10, or that those two values are more likely than the values between them.

Interpreting Equal Likelihood

In a continuous model, the density at a point is not the probability of observing that exact point. Instead, the density describes how probability is distributed over intervals. For the uniform model, the density is the same everywhere inside the support. Consequently, intervals of equal length within the support have equal probability, regardless of where they are located.

The following justification uses the constant density from the definition. For an interval $[c,d]$ contained in the support, its probability is the area under the density between $c$ and $d$. Because the density is constant, that area is a rectangle with width $d-c$ and height $1/(b-a)$.

P(c\le X\le d) =\int_c^d \frac{1}{b-a}\,dx =\frac{d-c}{b-a}.

Thus, if two intervals lie within the support and have the same length, the expression for their probabilities is the same. The total area over the support is also one:

\int_a^b \frac{1}{b-a}\,dx =\frac{b-a}{b-a} =1.

This explains why the density’s height is $1/(b-a)$: it makes the constant area across the full support equal to one. A wider support has a lower density, and a narrower support has a higher density. The density can exceed one; this is not a problem, because it is area over an interval—not height by itself—that gives probability.

“Equal likelihood across the interval” therefore has a precise meaning: equal-length subintervals inside the support have equal probabilities. It does not mean that each individual real number has a positive probability, or that intervals of different lengths are equally likely.

Reading the Model Before Calculating

A reliable first step is to write down the random variable with its units and then record the two endpoints in those same units. Check that the lower endpoint is strictly smaller than the upper endpoint. The support tells you where the model places probability; the parameters are the values that define that support.

Name the variable and its units.
State what quantity $X$ measures, such as elapsed time in seconds or a signed measurement error in millimeters.

Identify the endpoints.
Read the smallest and largest permitted values from the model description, keeping the units attached.

Write the distribution.
Use $X\sim\operatorname{Uniform}(a,b)$ with $a<b$, and state the support in context.

Interpret equal likelihood.
Compare interval lengths, not just their locations. Equal lengths within the support have equal probabilities under this model.

Worked Example: Identifying a Time Interval and Comparing Two Windows

Model and assumptions. In a hypothetical scheduling system, a task’s completion time $X$, measured in minutes, is modeled as continuous and uniform from 12 minutes to 20 minutes. Identify the parameters and support, then compare the likelihood of completion in the windows from 13 to 15 minutes and from 17 to 19 minutes.

Method. Read the endpoints in the stated units, then use the constant density to compare the probabilities of the two intervals. The model is $X\sim\operatorname{Uniform}(12,20)$, so $a=12$, $b=20$, and the support is from 12 to 20 minutes. Its length is

b-a=20-12=8\text{ minutes}.

The first window has length $15-13=2$ minutes. The second has length $19-17=2$ minutes. Both windows are inside the support. Each therefore has probability equal to its length divided by the support length:

P(13\le X\le15)=\frac{2}{8}=0.25,\qquad P(17\le X\le19)=\frac{2}{8}=0.25.

The two windows are equally likely under the stated model. Check: each probability is one quarter, and the windows have equal lengths. The common mistake avoided here is treating a window nearer the upper endpoint as more likely merely because it is later in time.

Worked Example: A Signed Measurement Error

Model and assumptions. Suppose, hypothetically, that a device’s signed calibration error $E$, measured in millimeters, follows a continuous uniform distribution from $-0.6$ mm to $0.6$ mm. Identify its support and parameters, and determine whether the intervals from $-0.5$ to $-0.2$ mm and from $0.1$ to $0.4$ mm have equal probabilities.

The variable is the signed error, so negative values indicate errors in one direction and positive values indicate errors in the other. The model is $E\sim\operatorname{Uniform}(-0.6,0.6)$. Thus $a=-0.6$ mm, $b=0.6$ mm, and the support runs from $-0.6$ mm to $0.6$ mm. Its length is

b-a=0.6-(-0.6)=1.2\text{ mm}.

The first interval’s length is $-0.2-(-0.5)=0.3$ mm. The second interval’s length is $0.4-0.1=0.3$ mm. Both are within the support and have the same length, so

P(-0.5\le E\le-0.2)=\frac{0.3}{1.2}=0.25,\qquad P(0.1\le E\le0.4)=\frac{0.3}{1.2}=0.25.

The probabilities match even though the intervals lie on opposite sides of zero. Check: both intervals use the same measurement unit and span 0.3 mm. The common mistake avoided is comparing the signed locations rather than the intervals’ lengths.

Worked Example: Converting the Support to Different Units

Model and assumptions. A hypothetical processing time $T$ is modeled as continuous and uniform from 6 seconds to 18 seconds. Express the same model in minutes, identify the parameters and support in each unit, and compare the intervals from 6 to 9 seconds and from 12 to 15 seconds.

In seconds, the model is $T\sim\operatorname{Uniform}(6,18)$, with support from 6 to 18 seconds. The support length is $18-6=12$ seconds. To express the endpoints in minutes, divide each by 60:

6\text{ seconds}=\frac{6}{60}=0.1\text{ minutes},\qquad 18\text{ seconds}=\frac{18}{60}=0.3\text{ minutes}.

Therefore the same physical-time model, expressed in minutes, is uniform from 0.1 to 0.3 minutes, with support length $0.3-0.1=0.2$ minutes. The first stated window has length $9-6=3$ seconds; the second has length $15-12=3$ seconds. Each is one quarter of the support length in seconds:

\frac{3}{12}=0.25.

Equivalently, each window is $3/60=0.05$ minutes long, and $0.05/0.2=0.25$. Thus the two windows have equal probability under the model. Check: converting the whole support and both windows to minutes leaves these ratios unchanged. The common mistake avoided is writing the numerical endpoints 6 and 18 while labeling them as minutes.

Common Modeling Traps

Confusing the endpoints with the only possible values. The parameters $a$ and $b$ bound a continuum of possible values; the model is not a two-outcome distribution.
Assuming all intervals are equally likely. The equal-likelihood property applies to intervals of equal length within the support. A longer interval has a larger probability under the uniform model.
Assigning probability to a point from the density. The density $1/(b-a)$ is not the probability of any particular exact value. Probability is assigned to intervals, whose lengths matter.
Ignoring units or the model assumption. Endpoints must use the same units as the variable. In addition, a variable is not uniform merely because its values are bounded; the constant-density assumption must be appropriate to the situation.

Key takeaway: For $X\sim\operatorname{Uniform}(a,b)$, $a<b$ gives the lower and upper endpoint parameters and the support. The density is constant across that support, so equal-length intervals contained in it have equal probabilities.

Check Your Understanding

For each problem, identify the model features or use the equal-length principle as appropriate.

A hypothetical response time $X$ is modeled uniformly from 3 to 11 seconds. State its parameters and support, then compare the intervals from 4 to 6 seconds and from 8 to 10 seconds. (Answer: $a=3$, $b=11$; support 3 to 11 seconds; both intervals have length 2 seconds and probability $2/8=0.25$.)
A signed delay $D$, in minutes, follows a continuous uniform model from $-4$ to $2$. What is its support length, and what is the probability of each of the intervals from $-3.5$ to $-3$ and from 1 to 1.5? (Answer: support length 6 minutes; each interval has length 0.5 minute and probability $0.5/6=1/12\approx0.0833$.)
A hypothetical measurement $M$ is uniform from 40 to 100 centimeters. Express the support and endpoint parameters in meters. (Answer: $a=0.4$ m, $b=1.0$ m; support 0.4 to 1.0 meters.)
A proposed continuous uniform model has support from 0 to 30 units. An analyst says that each integer value is equally likely and therefore has probability $1/31$. Explain the error and state what the model does imply about equal-length intervals. (Answer: the density is not a point probability; each exact value has probability zero, while equal-length intervals within 0 to 30 have equal probabilities.)
A hypothetical repair time $R$ is uniform from 2.5 to 7.5 hours. Compare the probability of a repair time from 3.5 to 4.5 hours with the probability of a repair time from 6 to 7 hours. (Answer: support length is 5 hours; each interval has length 1 hour, so each probability is $1/5=0.20$.)

What You'll Learn