Support-Aware Inequalities for Probability Calculations

Why the Support Comes First

An inequality involving a random variable describes a region of possible values. But the event occurs only if the random variable can actually take values in that region. Checking the support first can turn a complicated-looking event into a short list of outcomes, a truncated interval, or an event that is impossible or certain.

This is especially useful when an inequality has bounds outside the range of the random variable. It also prevents a frequent probability error: using a continuous interval formula when the distribution has a point mass at an endpoint. The calculations depend both on the supported values and on whether the endpoints are included.

Definition: The support of a random variable is the set of values where its probability distribution is concentrated. For a discrete random variable, list the values that have positive probability. For a continuous random variable with density $f_X$, the support is the region over which the density can be positive, including boundary points when appropriate. Probabilities are assigned only to the part of an event that intersects the distribution’s support.

For a discrete random variable, an event such as $a<X\le b$ is evaluated by adding the probabilities of supported values that satisfy both inequalities. For a continuous random variable with density, the corresponding probability is the integral over the part of $(a,b]$ that lies within the support. If the distribution is described by its cumulative distribution function $F_X(x)=P(X\le x)$, CDF differences provide a convenient calculation, but endpoint conventions still matter when point masses are present.

Reduce the Event to the Supported Region

The basic procedure is to write down the event, intersect its region with the support, and only then calculate its probability. For example, if $X$ can take values only between 0 and 4, the event $X<-2$ is empty and has probability zero. The event $X\le 7$ includes the entire support and has probability one. If the event is $X>2$, only the supported values above 2 contribute.

State the support.
List the possible discrete values or write the interval or set on which the distribution is concentrated.

Translate the inequalities into a region.
Keep strict and non-strict inequalities distinct, and split unions into separate regions when helpful.

Intersect with the support.
Discard unsupported values or portions of the region. If the intersection is empty, the probability is zero.

Choose the probability calculation.
Sum PMF probabilities, integrate the density, or use CDF values with the appropriate endpoint convention.

For a discrete random variable, values outside its support have probability zero, even if they satisfy the written inequality. For a continuous random variable, the density is integrated only over the supported portion. In both settings, the support simplifies the event; it does not change the probability assigned by the model to the remaining region.

Worked Example: Selecting Discrete Outcomes from a Union of Regions

A hypothetical claim-count variable $X$ has support $\{-1,0,2,3,6\}$, with probabilities $0.10, 0.20, 0.25, 0.15, 0.30$, respectively. Find $P((-2<X<3)\text{ or }(4<X\le 6))$.

Method. Intersect each region with the discrete support, then add the probabilities of the selected outcomes. The two regions do not overlap.

For the first region, $-2<X<3$, the supported values are $-1,0,2$. The value 3 is excluded by the strict upper inequality, and the value 6 is outside the region. Thus, $P(-2<X<3)=0.10+0.20+0.25=0.55$.

For the second region, $4<X\le 6$, the only supported value is 6. Therefore, $P(4<X\le 6)=0.30$.

The events are disjoint, so $P((-2<X<3)\text{ or }(4<X\le 6))=0.55+0.30=0.85$.

Check. The selected values are $-1,0,2,6$; the only omitted supported value is 3, whose probability is 0.15. The result is $1-0.15=0.85$, as required. The method avoids treating the inequalities as if every real number in their ranges were a possible outcome.

Truncating a Density Integral to the Support

Suppose $X$ has density $f_X$ on a support interval $[L,U]$, and an event asks for $a<X\le b$. Its probability is the integral over the intersection of the event interval with the density’s support. When the distribution is continuous, including or excluding a single endpoint does not change the probability, because a single point has probability zero. This endpoint fact must not be carried over to distributions with point masses.

Density calculation: If $X$ has density $f_X$, then $P(X\in A)=\int_A f_X(x)\,dx$ for a region $A$ within the support. If $A$ extends beyond the support, integrate only over $A$ intersected with the support. The density formula does not apply to a point mass as though that mass were an area under an ordinary density.

Worked Example: Truncating Two Intervals to a Continuous Support

A hypothetical continuous measurement $X$ has density $f_X(x)=x/8$ for $0\le x\le4$, and density zero elsewhere. Find $P((-3<X\le1)\text{ or }(2<X\le5))$.

Check the model. The density is nonnegative on $[0,4]$, and $\int_0^4(x/8)\,dx=[x^2/16]_0^4=16/16=1$.

Method. Intersect each event interval with the support $[0,4]$. The first interval becomes $[0,1]$ for integration; the second becomes $(2,4]$. Endpoint inclusion does not affect a continuous probability.

For the first part, $P(-3<X\le1)=\int_0^1(x/8)\,dx=[x^2/16]_0^1=1/16$.

For the second part, $P(2<X\le5)=\int_2^4(x/8)\,dx=[x^2/16]_2^4=(16-4)/16=12/16=3/4$.

The intervals are disjoint, so the requested probability is $1/16+3/4=1/16+12/16=13/16=0.8125$.

Check. The omitted supported region is $1<X\le2$, whose probability is $(4-1)/16=3/16$. Therefore the requested probability is $1-3/16=13/16$, matching the direct calculation. The method avoids integrating over unsupported values below 0 or above 4.

Endpoint Conventions for CDF Calculations

The CDF is defined by $F_X(x)=P(X\le x)$. Consequently, $F_X(b)-F_X(a)$ includes values strictly greater than $a$ and up to and including $b$. To handle other endpoint combinations, use the left limit $F_X(a-)=\lim_{x\uparrow a}F_X(x)=P(X<a)$. The left limit matters at a point where the CDF jumps, which is exactly where the distribution has positive probability at that point.

Endpoint formulas: For $a<b$, $$ \begin{aligned} P(a<X\le b)&=F_X(b)-F_X(a),\\ P(a\le X\le b)&=F_X(b)-F_X(a-),\\ P(a<X<b)&=F_X(b-)-F_X(a),\\ P(a\le X<b)&=F_X(b-)-F_X(a-). \end{aligned} $$ The point probability at $a$ is $P(X=a)=F_X(a)-F_X(a-)$.

These formulas follow by partitioning the CDF events. The event $\{X\le b\}$ consists of the disjoint events $\{X\le a\}$ and $\{a<X\le b\}$. Thus $F_X(b)=F_X(a)+P(a<X\le b)$, which gives the first formula. Replacing $\{X\le a\}$ by $\{X<a\}$ gives the second. Applying the same reasoning with $b$ replaced by values approaching $b$ from below gives the formulas with $F_X(b-)=P(X<b)$. In a continuous distribution, $F_X(c)=F_X(c-)$ at every point $c$, so the endpoint distinctions disappear; in a distribution with point masses, they may change the answer.

Worked Example: A Mixed Distribution with a Point Mass at the Endpoint

A hypothetical duration $X$ has probability 0.20 at 0. The remaining probability 0.80 is spread uniformly over the interval $(1,4)$. Thus its CDF is $F_X(x)=0$ for $x<0$, $F_X(x)=0.20$ for $0\le x\le1$, $F_X(x)=0.20+0.80(x-1)/3$ for $1<x<4$, and $F_X(x)=1$ for $x\ge4$. Find $P(0<X\le2)$, $P(0\le X\le2)$, and $P(X>4)$.

Method. Use CDF differences for the intervals, distinguishing $F_X(0)$ from $F_X(0-)$. The only point mass is at 0; the uniform component has no point masses.

At 2, $F_X(2)=0.20+0.80(2-1)/3=0.20+0.80/3=1/5+4/15=7/15$. At 0, $F_X(0)=1/5$ and $F_X(0-)=0$.

For the open lower endpoint, $P(0<X\le2)=F_X(2)-F_X(0)=7/15-1/5=7/15-3/15=4/15$.

For the closed lower endpoint, $P(0\le X\le2)=F_X(2)-F_X(0-)=7/15-0=7/15$. The difference is $7/15-4/15=3/15=1/5=0.20$, exactly the point mass at 0.

Finally, $F_X(4)=1$, so $P(X>4)=1-F_X(4)=0$. The distribution has no probability beyond its upper support endpoint.

Check. The event $0\le X\le2$ includes the 0.20 atom and the 4/15 probability from the continuous portion: $1/5+4/15=3/15+4/15=7/15$. This agrees with the CDF calculation. The method avoids losing the atom by using $F_X(0)$ where $F_X(0-)$ is required for a closed lower endpoint.

Common Mistakes and Exam Traps

Counting unsupported values. An inequality may describe many real numbers, but only supported outcomes contribute. For a discrete variable, filter the support list before summing.
Forgetting that support may truncate an interval. If $X$ is supported on $[0,4]$, an integral from $-3$ to 5 must use only the portion from 0 to 4.
Using the wrong CDF difference at an atom. The calculation $F_X(b)-F_X(a)$ is for $a<X\le b$, not $a\le X\le b$. For the closed lower endpoint use $F_X(a-)$.
Assuming every endpoint has probability zero. This is true for a continuous distribution with a density, but not for a discrete or mixed distribution. Check for jumps in the CDF.
Adding probabilities of overlapping regions without accounting for overlap. Confirm that the event pieces are disjoint before adding their probabilities; otherwise subtract the probability of their intersection.

Key takeaway: First intersect the event with the support. Then sum, integrate, or use a CDF according to the distribution model. When using a CDF, match each endpoint to the event’s strict or non-strict inequality.

Check Your Understanding

For each problem, reduce the event to the supported region before calculating its probability.

A discrete random variable has support $\{0,1,4,7\}$ with probabilities $0.10,0.25,0.40,0.25$, respectively. Find $P(-2<X\le4)$. (Answer: $0.75$)
A continuous random variable has density $f_X(x)=1/5$ for $2\le x\le7$, and zero elsewhere. Find $P(X<5\text{ or }X>6)$. (Answer: $0.80$)
A CDF satisfies $F_X(3)=0.62$ and $F_X(3-)=0.47$. Find $P(X=3)$ and $P(X\le3)$. (Answer: $P(X=3)=0.15$; $P(X\le3)=0.62$)
A random variable has support $\{-2,0,1,5\}$, with probabilities $0.15,0.20,0.35,0.30$. Find $P(X<-2\text{ or }1<X\le5)$. (Answer: $0.30$)
A mixed distribution has $P(X=1)=0.12$, and its CDF values are $F_X(4)=0.70$ and $F_X(1)=0.30$. Find $P(1<X\le4)$ and $P(1\le X\le4)$. (Answer: $0.40$ and $0.52$, respectively)

What You'll Learn