An indirect effect is the portion of a cause’s influence on an outcome that travels through some intermediate step, rather than operating in a straight line from cause to result. If a job-training program reduces poverty, part of that reduction might come from the training itself and part might come because trainees gain confidence, which leads them to apply for better jobs, which then raises their income. That confidence-and-application chain is the indirect effect. The concept shows up across nearly every field that studies cause and consequence, from medicine to ecology to economics, and understanding it changes how you think about why interventions work or fail.
The Basic Idea Behind Direct and Indirect Effects
Researchers who study causes and outcomes often want to know not just whether something works, but how it works. Causal mediation analysis is the formal toolkit for pulling apart a total effect into its components. The total effect is the average change in an outcome when a whole population receives versus does not receive some treatment or exposure. The natural direct effect is whatever portion of that change operates through pathways that do not involve the intermediate variable being studied. The natural indirect effect is the portion that passes through that intermediate variable, known as the mediator.1PubMed Central. Causal mediation analysis: what is it and how can it be used to inform practice and policy?
Think of it as splitting a river into two channels. Water flows from a mountain (the cause) to a valley (the outcome). Some water goes straight down the slope. The rest detours through a lake before continuing to the valley. The detour through the lake is the indirect effect. Measuring how much water goes each way tells you how important the lake is to the valley’s water supply. In research terms, you learn how much of the treatment’s benefit depends on that intermediate step.2PubMed Central. Causal mediation analysis in the context of clinical research
The concept has deep roots. Judea Pearl’s foundational work on causal inference laid out the mathematical language for reasoning about direct and indirect effects, alongside interventions and counterfactuals.3PubMed Central. An introduction to causal inference But you do not need any of that math to understand the core idea: indirect effects are the hidden pathways through which causes operate.
Herd Protection From Vaccines
One of the most consequential indirect effects in public health is herd protection. When you get vaccinated, the direct effect is your own reduced risk of getting sick. But there is a second channel: because you are less likely to become infected, you are also less likely to pass the pathogen to someone else. When enough people in a community are immunized, the overall transmission of the disease drops, and even unvaccinated people benefit from a lower risk of exposure. That benefit to the unvaccinated is the indirect effect of vaccination, sometimes called the herd-protective effect.4Procedia in Vaccinology. Concepts of herd protection and immunity
This indirect effect can be enormous. Studies of cholera and typhoid vaccination campaigns have shown measurable declines in disease rates among unvaccinated community members simply because enough of their neighbors were immunized, reducing the pathogen’s ability to circulate.5PubMed Central. Assessment of Vaccine Herd Protection: Lessons Learned From Cholera and Typhoid Vaccine Trials This is why public health officials care not just about individual protection rates but about coverage thresholds: the indirect effect only kicks in meaningfully once a large enough share of the population is immune.
Wolves, Elk, and Aspen Trees
Ecology is where indirect effects become viscerally visible. A textbook example involves wolves, elk, and aspen trees. When wolves are present in a landscape, they reduce elk populations through predation. Fewer elk means less browsing on young aspen trees, so the aspens recover. The wolves never touch the trees, yet they shape the forest. That chain, predator reduces herbivore, herbivore stops eating plant, is called a density-mediated indirect effect because it works by changing the number of herbivores.6Ecological Monographs. The primacy of density‐mediated indirect effects in a community of wolves, elk, and aspen
But there is a subtler version. Even if wolves do not actually kill many elk, their presence can change elk behavior: elk avoid certain areas, spend less time feeding in the open, or move to different terrain. That behavioral shift alone can let plants recover. Ecologists call this a trait-mediated indirect effect because it operates through a change in the prey’s behavior or physiology rather than through a change in prey numbers. Experimental work with dragonfly larvae and tadpoles showed that the indirect effects from merely scaring prey into changing their behavior were large compared to the indirect effects caused by actually reducing prey numbers.7PubMed Central. The contribution of trait-mediated indirect effects to the net effects of a predator
The distinction matters because it changes what conservation strategies accomplish. Reintroducing a predator might benefit plant communities even before it significantly reduces herbivore numbers, just by altering herbivore behavior. Misunderstanding the mechanism could lead managers to expect no ecological recovery until prey populations visibly decline, when in fact behavioral changes begin immediately.
How Stress Gets Under Your Skin
Chronic psychological stress raises the risk of heart disease, but the pathway from “feeling stressed” to “blocked arteries” is not a straight line. Some of the connection is direct, through physiological channels like elevated stress hormones and inflammation. But a large share of the effect is indirect, running through behavior. Stressed people are more likely to smoke, eat poorly, and be sedentary. Those behaviors then raise cardiovascular risk on their own.8PubMed. Impact of psychological factors on the pathogenesis of cardiovascular disease and implications for therapy
How large are these behavioral indirect effects? One analysis found that behavioral factors explained roughly 65% of the link between psychological distress and cardiovascular events, while direct pathophysiological markers like inflammation and high blood pressure accounted for much smaller shares.9PubMed. Psychological distress as a risk factor for cardiovascular events: pathophysiological and behavioral mechanisms A separate study looking at stress and all-cause mortality found that smoking, sedentary behavior, and obesity each carried indirect effects in the range of roughly 11 to 14% of the total stress-mortality link.10PubMed Central. The relative contributions of behavioral, biological, and psychological risk factors in the association between psychosocial stress and all-cause mortality among middle- and older-aged adults in the USA
This has real implications for how doctors and therapists should approach stressed patients. If the bulk of the stress-heart disease link is indirect, mediated through smoking and inactivity, then interventions that help people change those specific behaviors might matter more than interventions aimed solely at lowering physiological stress markers. You are targeting the mediator, and the mediator is where most of the action is.
Diet, Gut Microbes, and Metabolic Health
Dietary fiber is good for you. That is old news. What is less obvious is that much of fiber’s benefit operates indirectly, through changes in the microbial community living in your gut. When you eat more fiber, you feed specific populations of bacteria. Those bacteria produce metabolites that influence inflammation, blood sugar regulation, and cholesterol levels. The fiber itself does not directly lower your blood sugar; it reshapes the microbial ecosystem, and the ecosystem does the work.11PubMed Central. High-Fiber, Whole-Food Dietary Intervention Alters the Human Gut Microbiome but Not Fecal Short-Chain Fatty Acids
A trial testing personalized diets for people with pre-diabetes used causal mediation analysis and found that nine specific microbial species partially mediated the link between dietary adherence and improvements in markers like hemoglobin A1c, HDL cholesterol, and triglycerides.12PubMed. Gut microbiome modulates the effects of a personalised postprandial-targeting (PPT) diet on cardiometabolic markers: a diet intervention in pre-diabetes In other words, the diet changed the microbes, and the microbes changed the metabolic outcomes. Without the indirect-effect framework, you might just say “diet improved blood sugar” and leave it at that. With the framework, you can identify which microbial species are doing the heavy lifting, opening the door to more targeted interventions like probiotics or microbiome transplants.
Economic Ripple Effects
Economists have their own version of indirect effects, and they call them multiplier effects. When a tourist spends money at a hotel, that spending has a direct effect: it pays the hotel’s employees and owners. But those employees then spend part of their wages at local grocery stores, restaurants, and gas stations. The grocery store uses some of that revenue to buy goods from regional suppliers, and so on. Each round of re-spending is an indirect effect of the original tourist dollar. Multiplier analysis tracks these waves of direct, indirect, and induced effects across a local economy.13Journal of Travel Research. Estimating the Multiplier Effects of Tourism Expenditures on a Local Economy through a Regional Input-Output Model
The same logic applies to any injection of spending into a region. A factory opens. Its direct effect is the jobs it creates and the wages it pays. Its indirect effect is the additional economic activity generated as the factory purchases supplies from other businesses, which in turn hire workers and buy their own supplies. Input-output models calculate the full cascade of these effects, which is why infrastructure projects and tourism boards often cite multiplier numbers to justify investment.14ResearchGate. Analysis of the Multiplier Effects Produced by Investment in a Dynamic Input–Output Model The indirect effects in economic modeling can be double or triple the direct effects, which is why ignoring them produces drastically incomplete pictures of an investment’s value.
Biofuels and Indirect Land Use Change
Sometimes indirect effects turn a seemingly good policy into a questionable one. Biofuels were promoted as a climate-friendly alternative to gasoline because burning ethanol made from corn releases less fossil carbon. But growing more corn for ethanol requires more farmland. When U.S. farmers shifted existing cropland to corn, the crops that were displaced had to be grown somewhere else, often on newly cleared land in other countries. Clearing that land, sometimes forest or grassland, released stored carbon into the atmosphere. This is indirect land use change, and its carbon emissions can be surprisingly large.15PubMed Central. Accounting for indirect land-use change in the life cycle assessment of biofuel supply chains
Researchers estimated that the greenhouse gas emissions from this indirect land use change for U.S. corn ethanol ranged from small but not negligible to several times greater than the life cycle emissions of gasoline itself.16PubMed. Greenhouse gas emissions from biofuels’ indirect land use change are uncertain but may be much greater than previously estimated That is a remarkable finding: the indirect effect of switching to biofuels could potentially be worse than the problem you were trying to solve. This case became a landmark example of why policy analysis must account for ripple effects, not just the immediate outcomes of a decision.
Corporate Carbon Footprints and Scope 3 Emissions
Companies measure their greenhouse gas emissions in three “scopes.” Scope 1 is what comes from their own operations, like running a factory. Scope 2 is the emissions from the electricity they buy. Scope 3 is everything else in their value chain: the emissions from producing their raw materials, transporting their products, and even customers using and disposing of those products. Scope 3 emissions are, in the language of this article, the indirect effects of a company’s existence and activity. They often dwarf the direct emissions.17Journal of Industrial Ecology. A multi-tier methodology for Scope 3 emissions accounting in complex supply chains: mathematical framework and empirical insights from China
Despite being the largest component of many companies’ carbon footprints, Scope 3 emissions remain notoriously difficult to measure. They depend on data from suppliers, logistics partners, and end-users, often spanning multiple countries and industries. This creates a practical problem: a company can slash its own factory emissions to near zero and still be responsible for massive indirect emissions upstream and downstream. Setting meaningful climate targets requires grappling with these indirect emissions, not just the ones you can see coming from your own smokestack.18Corporate Social Responsibility and Environmental Management. Landscape of Research on Accounting Scope 3 Emissions: A Review of Methodologies and Data
Why Indirect Effects Are Hard to Prove
Identifying an indirect effect is trickier than identifying a total effect. To claim that a cause influences an outcome through a specific mediator, you need to rule out alternative explanations for the mediator-outcome link. There are four key “no unmeasured confounding” assumptions that must hold: nothing unmeasured distorts the relationship between the cause and the outcome, between the cause and the mediator, between the mediator and the outcome, and no variable affected by the cause confounds the mediator-outcome relationship.19PubMed Central. Confounding in Statistical Mediation Analysis: What It Is and How to Address It – Section: Four Assumptions About Confounding in Mediation Models
That last assumption is especially sneaky. Imagine testing whether a medication reduces pain (the outcome) indirectly through improved sleep (the mediator). But the medication also reduces anxiety, and anxiety affects both sleep and pain. Anxiety is now a confounder of the mediator-outcome link, and it was affected by the treatment itself. Traditional methods struggle with this scenario. Newer approaches, like the “instrumental blocker” technique, attempt to relax these assumptions so researchers can still estimate indirect effects even when some confounders are unmeasured or affected by the treatment.20PubMed. Identification and robust estimation of swapped direct and indirect effects: Mediation analysis with unmeasured mediator-outcome confounding and intermediate confounding
The statistical tools for testing indirect effects also matter. Comparisons of common testing methods found they agree with each other most of the time, but disagree more often when an indirect effect actually exists. Confidence-interval approaches like the bias-corrected bootstrap tend to be the most powerful at detecting real effects, though they can occasionally be too liberal, flagging effects that are not really there. More conservative methods like the Monte Carlo confidence interval trade some detection power for tighter control of false positives.21PubMed. The relative trustworthiness of inferential tests of the indirect effect in statistical mediation analysis: does method really matter?
When the Indirect Effect Changes Depending on Who You Are
Indirect effects are not always uniform. They can be stronger for some people or in some conditions and weaker in others. This is called moderated mediation: the strength of the indirect pathway depends on the value of a third variable. A job-training program might improve employment (the outcome) through increased confidence (the mediator), but only for participants who had some prior work experience. For those with no work history, confidence might not translate into employment at all. The indirect effect through confidence is moderated by work experience.22PubMed. An Index and Test of Linear Moderated Mediation
A study on anti-drug messaging to adolescents showed this pattern. The effectiveness of persuasive language in shaping marijuana attitudes and usage intentions depended on who was delivering the message. When adults used moderate rather than extreme language in their commentary, the indirect effect on attitudes, flowing through teens’ sense of personal control, was significant. When peers delivered the same messages, no such indirect effect appeared.23PubMed Central. Social mediation of persuasive media in adolescent substance prevention. The mechanism existed, but it only switched on under certain conditions. Missing this kind of moderation can lead to blanket recommendations that work for some groups and fail for others.
Behaviors Rippling Through Social Networks
One of the more striking demonstrations of indirect effects comes from research on social networks. Your friend’s behavior can influence your behavior, which is a direct social effect. But your friend’s friend’s behavior can also reach you, through your friend. Analyses of large social networks have found that health-related traits like body mass index show statistically meaningful correlations up to three degrees of separation: your friend’s friend’s friend’s weight is associated with your own, even though you have never met that person.24PubMed Central. Social contagion theory: examining dynamic social networks and human behavior
These multi-step social influences are, at their core, chains of indirect effects. Person A influences Person B’s eating habits, Person B influences Person C’s, and Person C influences yours. You experience the downstream consequences of a behavioral shift that started several handshakes away. The finding has sparked debate about the mechanisms involved, whether it is genuine behavioral contagion or simply the tendency for similar people to cluster together, but the pattern of indirect influence across network layers has been replicated across multiple health behaviors and outcomes.
Using Indirect Effects to Design Better Policies
Perhaps the most practical payoff of understanding indirect effects is in designing interventions that actually work. Causal mediation methods allow researchers to simulate what would happen if you changed a specific mediator, testing hypothetical scenarios that vary in eligibility, effectiveness, and uptake. This kind of modeling speaks directly to the language of real-world policy decisions.25PubMed Central. Causal mediation approaches for understanding pathways to inequalities and policy entry points: examples from early years health and development
Suppose you know that a childhood nutrition program reduces educational inequality, and mediation analysis shows that improved school attendance is the primary indirect pathway. You could then model what happens if only half of eligible families enroll, or if the attendance effect is smaller in rural areas, or if a different eligibility cutoff is used. Each scenario changes the predicted size of the indirect effect and, consequently, the total impact on inequality. Without the indirect-effect framework, you are stuck evaluating programs as black boxes, knowing only that they worked or did not, with no insight into which part of the mechanism is doing the work and where improvements should be targeted.
Drug development faces a parallel challenge. Roughly a third of drug candidates fail due to toxicity, and understanding whether that toxicity operates through the drug’s intended mechanism (an on-target effect) or through unintended pathways (off-target or indirect effects) shapes how developers try to fix the problem.26PubMed Central. Mechanisms of drug toxicity and relevance to pharmaceutical development. A drug that causes liver damage through the same receptor it uses to treat disease presents a different redesign challenge than one that causes damage through a metabolic byproduct the designers did not anticipate. Identifying the indirect pathway is the first step toward fixing it.