Causal Inference glossary
49 causal inference terms, each defined in plain language with what the number hides once real ad spend runs through it. Definitions are written for ecommerce operators, not statisticians.
Causal Inference terms
- Association vs. CausationAssociation indicates a relationship between two variables. Causation means a change in one variable directly produces a change in another.
- Attrition BiasAttrition Bias occurs when participants who leave a study or marketing funnel differ systematically from those who remain. This skews results and leads to inaccurate conclusions.
- Average Treatment Effect (ATE)Average Treatment Effect (ATE) is the average causal effect of a treatment on an outcome in a population. It is the difference between average outcomes with and without treatment.
- Average Treatment Effect on the Treated (ATT)Average Treatment Effect on the Treated (ATT) is the average causal effect of a treatment on those who received it. It measures the impact of a program on its participants.
- Backdoor CriterionBackdoor Criterion identifies a sufficient set of variables to control for confounding between a treatment and an outcome. It blocks all 'backdoor paths' on a directed acyclic graph (DAG).
- BlindingBlinding is a procedure where one or more parties in an experiment do not know which treatment subjects received. It prevents bias in study results.
- Causal ChainA Causal Chain is a sequence of events where each event causes the next, leading from an initial cause to a final effect.
- Causal DiscoveryCausal Discovery infers causal relationships from data, using statistical methods and machine learning to uncover a system's causal structure.
- Causal ForestsCausal Forests are a machine learning method that estimates heterogeneous treatment effects, extending random forests for causal inference.
- Causal ModelA Causal Model is a mathematical representation describing the causal relationships between variables, used to reason about and estimate intervention effects.
- Causal PathwayA Causal Pathway is the sequence of events through which a cause produces an effect.
- CausalityCausality is the relationship where one event directly causes another, essential for identifying specific actions that drive desired outcomes in marketing.
- ColliderA Collider is a variable in a directed acyclic graph (DAG) that two or more other variables cause. Conditioning on a collider opens a non-causal path between its causes, leading to spurious association.
- ConfoundingConfounding is a distortion of the estimated treatment effect when a third variable, a confounder, associates with both the treatment and the outcome. Causal inference methods control for confounding to isolate the true treatment effect.
- CounterfactualCounterfactual is a hypothetical outcome that would have occurred if a subject had received a different treatment.
- d-separationd-separation is a graphical criterion that determines conditional independence between sets of variables in a Directed Acyclic Graph (DAG). It identifies confounding and informs which variables to control for in causal analysis.
- Directed Acyclic Graph (DAG)Directed Acyclic Graph (DAG) is a graphical representation of causal relationships between variables. Nodes represent variables, and directed edges represent causal relationships without feedback loops.
- Do-CalculusDo-Calculus is a set of rules for manipulating probability distributions to estimate the causal effect of an intervention from observational data.
- Double Machine LearningDouble Machine Learning is a statistical method for estimating causal parameters when high-dimensional confounding exists.
- EconometricsEconometrics applies statistical methods to economic data to provide empirical content to economic relationships. It provides tools for causal inference, such as regression analysis, instrumental variables, and difference-in-differences.
- ExogeneityExogeneity is a property where a variable in a statistical model does not correlate with the error term. This allows for a causal interpretation of the variable's effect.
- ExperimentsExperiments are scientific procedures that test hypotheses or demonstrate facts. In marketing, experiments like A/B tests determine the causal effect of campaign changes, enabling data-driven decisions.
- Frontdoor CriterionFrontdoor Criterion identifies a mediating variable to estimate the causal effect of a treatment on an outcome when unmeasured confounding exists.
- g-computationg-computation is a method for estimating the causal effect of a time-varying treatment from longitudinal data. It uses a parametric model to estimate outcome distributions under different treatment strategies.
- Hawthorne EffectHawthorne Effect is a type of reactivity where individuals modify their behavior because they know they are being observed.
- Heterogeneous Treatment EffectsHeterogeneous treatment effects are variations in a treatment's causal impact across different population subgroups. Understanding these effects is crucial for personalizing marketing and maximizing ROI.
- Immortal Time BiasImmortal Time Bias occurs in observational studies when the follow-up period includes time where the outcome cannot happen. This inflates the perceived treatment effect.
- Interaction EffectAn Interaction Effect occurs when one variable's effect on an outcome depends on another variable's level.
- Interrupted Time Series (ITS)Interrupted Time Series (ITS) is a quasi-experimental design that evaluates an intervention's effect by comparing outcome trends before and after its implementation.
- InterventionAn Intervention is an action taken to produce a change in an outcome.
- Inverse Probability Weighting (IPW)Inverse Probability Weighting (IPW) is a statistical method that estimates causal effects from observational data by weighting subjects based on their treatment probability.
- Local Average Treatment Effect (LATE)Local Average Treatment Effect (LATE) is the average causal effect of a treatment on individuals whose treatment status changes due to an instrumental variable. It measures the treatment effect for a specific subpopulation.
- Marginal Structural Model (MSM)A Marginal Structural Model (MSM) is a statistical model that estimates the causal effect of a time-varying treatment from longitudinal data. It uses inverse probability weighting to account for confounding variables.
- MatchingMatching is a statistical technique that reduces bias in observational studies. It pairs treated subjects with similar control subjects.
- Mediation AnalysisMediation analysis is a statistical method that explains how a treatment affects an outcome. It separates direct effects from indirect effects through a mediator variable.
- Natural ExperimentNatural Experiment is an empirical study where experimental and control conditions are determined by nature or external factors. This estimates causal effects when randomization is not feasible.
- Observational StudyObservational Study observes the effects of a treatment or intervention without controlling exposure. It does not use randomization, making it susceptible to confounding and selection bias.
- Placebo EffectThe Placebo Effect is a beneficial outcome from a treatment attributed to a patient's belief, not the treatment itself. In marketing, customer expectations influence product perception.
- Potential Outcomes FrameworkPotential Outcomes Framework defines the causal effect of a treatment as the difference between potential outcomes under treatment and control. This framework reasons about causality and designs randomized experiments and observational studies.
- Quasi-ExperimentA quasi-experiment estimates the causal impact of an intervention without random assignment. It applies when random assignment is not feasible or ethical.
- RandomizationRandomization: The process of assigning subjects to treatment and control groups by chance. This minimizes confounding and selection bias, allowing for unbiased estimation of treatment effects.
- Randomized Controlled Trial (RCT)Randomized Controlled Trial (RCT): A study design where subjects are randomly assigned to a treatment or control group. RCTs are the standard for causal inference, minimizing bias and directly measuring treatment effects.
- Regression to the MeanRegression to the Mean describes the phenomenon where an extreme variable measurement tends to be closer to the average on subsequent measurements. This can bias before-and-after studies, falsely attributing change to an intervention.
- Simpson's ParadoxSimpson's Paradox shows a trend in data that reverses when groups combine. It proves association does not equal causation.
- Spurious CorrelationSpurious Correlation is a statistical relationship between variables that are not causally linked. It occurs due to coincidence or an unobserved third factor.
- Structural Causal Model (SCM)Structural Causal Model (SCM) is a mathematical framework representing causal relationships between variables. It uses equations and directed acyclic graphs to describe how variables influence each other.
- Survivorship BiasSurvivorship bias is the logical error of focusing on successful outcomes while ignoring failures. This leads to false conclusions by overlooking unseen data.
- Synthetic Control MethodThe Synthetic Control Method estimates the causal effect of an intervention in a single case study. It constructs a 'synthetic' control unit from a weighted average of control units to isolate the intervention's impact.
- Treatment EffectTreatment Effect is the causal impact of an intervention on an outcome. In marketing, this means the change in a metric like conversion rate directly caused by a campaign or pricing adjustment.
Other glossary sections
Last-click guesses.We run the math.
Causal attribution for ecommerce brands. Watch the model work on a sample store first, then upload your GA4 export and see which channels really drove revenue in 5–10 minutes. €99, pay-per-use. Pro at €299/mo when you want it continuous.
No signup for the demo. Book a 20-min call or compare plans.