{
  "id": "multiple-imputation-by-chained-equations",
  "title": "Multiple Imputation by Chained Equations",
  "lang": "en",
  "source_lang": "en",
  "summary": "Multiple imputation by chained equations handles missing data by iteratively fitting variable-specific conditional models, generating several completed datasets, and combining their statistical analyses.",
  "category": "mathematics",
  "subcategory": "probability-statistics",
  "type": "concept",
  "aliases": [
    "MICE",
    "Chained equations imputation",
    "Fully conditional specification",
    "FCS",
    "Sequential regression multiple imputation"
  ],
  "generated_at": "2026-11-06",
  "model": "pro/gpt-6.1-sol",
  "url": "https://aiwiki.page/en/multiple-imputation-by-chained-equations/",
  "translations": {
    "zh-Hans": {
      "url": "https://aiwiki.page/zh/multiple-imputation-by-chained-equations/",
      "json": "https://aiwiki.page/zh/multiple-imputation-by-chained-equations.json"
    }
  },
  "sections": [
    {
      "heading": "",
      "text": "Multiple imputation by chained equations (MICE) is a method of missing-data imputation that represents uncertainty about missing observations by generating several plausible completed datasets. It specifies a separate conditional model for each variable with missing values and repeatedly updates the missing entries using the current values of the other variables. Each completed dataset is analyzed separately, and the resulting estimates and uncertainty measures are combined. MICE is particularly useful for datasets containing a mixture of continuous, binary, categorical, and other variable types."
    },
    {
      "heading": "Statistical basis",
      "text": "MICE implements multiple imputation, a framework in which missing values are treated as uncertain rather than replaced by a single definitive guess. Differences between completed datasets reflect uncertainty about the unobserved values; uncertainty remaining within each completed dataset is represented by the statistical analysis performed on that dataset.\n\nFor example, a dataset might contain income, age, employment status, and educational attainment, with missing entries in several columns. A chained-equations procedure can model income using the other variables, then model employment status using income and the remaining variables, and continue through the incomplete columns. Later updates use values imputed during earlier updates.\n\nUnlike methods that begin with one explicit joint probability distribution for all variables, MICE specifies their conditional distributions individually. This approach is called fully conditional specification. It provides modeling flexibility, but separately chosen conditional models do not necessarily correspond to a coherent joint distribution."
    },
    {
      "heading": "Missingness assumptions",
      "text": "The validity of an imputation analysis depends on both its models and the process that generated the missing values. Three standard missing-data mechanisms distinguish the main cases:\n\n- Missing completely at random (MCAR): missingness does not depend on either observed or unobserved data values.\n- Missing at random (MAR): conditional on the observed data, missingness does not additionally depend on the unobserved values.\n- Missing not at random (MNAR): missingness still depends on unobserved values after conditioning on the observed data.\n\nStandard MICE analyses commonly rely on MAR, together with suitable model specification and conditions that permit the missingness process to be ignored for the intended inference. MAR does not mean that missingness is arbitrary. For example, income may be more often missing among younger respondents without violating MAR if age is observed and adequately represented in the imputation models.\n\nMAR generally cannot be distinguished from MNAR using the observed data alone. Standard chained equations do not automatically address MNAR. Such analyses require additional assumptions about missingness or the unobserved values, often examined through sensitivity analysis."
    },
    {
      "heading": "Algorithm",
      "text": "Let the dataset contain variables \\(X_1,\\ldots,X_p\\), and let \\(X_{-j}\\) denote all variables except \\(X_j\\). For each incomplete variable, a conditional model is specified:\n\n\\[p_j(X_j\\mid X_{-j},\\theta_j),\\]\n\nwhere \\(\\theta_j\\) denotes that model’s parameters.\n\nA typical implementation proceeds as follows:\n\n- Initialize missing entries. Supply provisional values, for example by sampling from observed entries.\n- Select an incomplete variable. Temporarily regard its originally missing entries as unknown.\n- Fit its conditional model. Use records in which that variable was observed, with the current observed or imputed values of its predictors.\n- Draw model parameters. In a Bayesian implementation, draw \\(\\theta_j\\) from its posterior distribution. Other implementations use suitable resampling or approximate procedures to represent parameter uncertainty.\n- Draw replacement values. Generate values for the missing entries from the conditional predictive distribution.\n- Cycle through the remaining incomplete variables.\n- Repeat the cycles. After an initial stabilization period, retain a completed dataset.\n\nThe process is repeated to obtain \\(m\\) completed datasets. Implementations may use separately initialized chains or retain appropriately spaced states from a longer run. Observed entries remain unchanged in ordinary imputation.\n\nAt cycle \\(t\\), a sequential update has the schematic form\n\n\\[X_j^{\\mathrm{mis},(t)} \\sim p_j\\!\\left( X_j^{\\mathrm{mis}} \\mid X_1^{(t)},\\ldots,X_{j-1}^{(t)}, X_{j+1}^{(t-1)},\\ldots,X_p^{(t-1)}, \\theta_j^{(t)} \\right).\\]\n\nHere each \\(X_k^{(t)}\\) combines its original observed values with its current imputations. Variables visited earlier in a cycle contribute their newly updated values; variables visited later contribute values from the preceding cycle.\n\nThe number of cycles and the number of completed datasets serve different purposes. Additional cycles may improve stabilization within a chain, whereas additional imputations reduce simulation error in pooled results. Increasing either number does not correct a misspecified model."
    },
    {
      "heading": "Conditional models",
      "text": "Different variables can use different model families:\n\nVariable type | Common conditional model\nApproximately continuous | [[linear-regression\nBinary | [[logistic-regression\nUnordered categorical | Multinomial logistic regression\nOrdered categorical | Ordinal regression\nCount | Poisson or negative binomial regression\nContinuous with a suitable observed donor pool | Predictive mean matching\n\nChoosing a model involves more than matching the broad variable type. Important features include skewness, bounds, excess zeros, nonlinear relationships, and interactions. For example, ordinary Gaussian regression can generate impossible negative values for a strictly positive variable.\n\nPredictive mean matching\n\nPredictive mean matching uses a fitted prediction model to identify observed records whose predicted values are close to those of a record with a missing value. One of these donors is selected, and its observed value supplies the imputation.\n\nBecause it uses observed donor values, the method can preserve the observed support and reduce sensitivity to some distributional misspecifications. It does not eliminate modeling assumptions. Sparse donor pools, poor predictors, or inadequate representation of a subgroup can produce weak imputations, and the method cannot naturally extrapolate beyond the available donor values.\n\nParameter and residual uncertainty\n\nFor proper multiple imputation, imputations generally need to represent both uncertainty in fitted model parameters and unexplained variation conditional on those parameters. Replacing missing entries only with fitted conditional means usually understates uncertainty and compresses variability.\n\nDrawing regression coefficients alone is also insufficient for an ordinary continuous-response model: predictive draws ordinarily include residual variation. The precise uncertainty-generating procedure depends on the imputation method."
    },
    {
      "heading": "Combining completed-data analyses",
      "text": "Suppose a scalar quantity \\(Q\\), such as a regression coefficient, is estimated in each of \\(m\\) completed datasets. Let \\(\\hat Q_k\\) be its estimate and \\(U_k\\) its estimated complete-data variance in dataset \\(k\\).\n\nThe pooled estimate is\n\n\\[\\bar Q=\\frac{1}{m}\\sum_{k=1}^{m}\\hat Q_k.\\]\n\nThe average within-imputation variance is\n\n\\[\\bar U=\\frac{1}{m}\\sum_{k=1}^{m}U_k,\\]\n\nand the between-imputation variance is\n\n\\[B=\\frac{1}{m-1}\\sum_{k=1}^{m}(\\hat Q_k-\\bar Q)^2.\\]\n\nRubin’s rules give the total variance estimate\n\n\\[T=\\bar U+\\left(1+\\frac{1}{m}\\right)B.\\]\n\nThe pooled standard error is \\(\\sqrt{T}\\). The between-imputation component accounts for uncertainty associated with the missing data, while the factor \\(1+1/m\\) includes a finite-imputation adjustment.\n\nConfidence intervals and significance tests commonly use a Student’s t-distribution with an appropriate degrees-of-freedom approximation. Small-sample adjustments may be important. Parameters with strongly asymmetric distributions may require transformation before pooling; for example, odds ratios are commonly pooled on the logarithmic scale.\n\nPooling is performed on analysis results, not by averaging the completed datasets into a single dataset. Likewise, simply averaging separate \\(p\\)-values is not a general multiple-imputation testing procedure. Multivariate tests and some performance measures require specialized pooling methods."
    },
    {
      "heading": "Compatibility and convergence",
      "text": "Compatibility of conditional models\n\nA collection of conditional models is compatible if there exists a joint distribution having those models as its full conditional distributions. Arbitrarily chosen regression equations need not satisfy this requirement.\n\nWhen the updates are genuine full conditional draws from a coherent joint model, the procedure is closely related to Gibbs sampling, a form of Markov chain Monte Carlo. General MICE procedures do not automatically have this interpretation.\n\nUnder incompatible specifications, theoretical guarantees can be weaker, and limiting behavior may depend on the update order. Nevertheless, incompatibility alone does not establish that a particular inferential result is invalid: performance also depends on the estimand, the missingness pattern, and the relevant model properties.\n\nCompatibility with the analysis\n\nA second issue is the relationship between the imputation models and the eventual analysis model. An imputation model that omits an important interaction or nonlinear term can weaken that relationship in the completed data.\n\nFor example, imputing a covariate using only additive linear relationships may distort an analysis that depends on its squared value or its interaction with another variable. Substantive-model-compatible approaches construct imputations to respect specified features of the analysis model. More generally, congeniality concerns whether the imputation and analysis procedures support mutually consistent inferential assumptions.\n\nDiagnostics\n\nCommon diagnostics include:\n\n- Trace plots of imputed means, standard deviations, or selected model parameters across cycles.\n- Comparisons between independently initialized chains.\n- Comparisons of observed and imputed distributions, including comparisons within relevant subgroups.\n- Checks for impossible combinations, boundary violations, and implausible relationships.\n- Assessment of the stability of pooled estimates as the number of imputations increases.\n\nObserved and imputed distributions need not match: under MAR, the records with missing values may systematically differ from fully observed records. Distributional differences therefore require interpretation rather than automatic rejection.\n\nStable chains do not establish that MAR is correct or that the imputation models are adequate. Posterior predictive checking and related model checks can expose some shortcomings but cannot verify assumptions about unobserved values."
    },
    {
      "heading": "Model construction and data structure",
      "text": "Imputation models often include variables used in the final analysis, predictors of incomplete variables, and observed predictors of missingness. Auxiliary variables can make MAR more plausible and improve precision, although excessive or poorly supported predictors can cause instability.\n\nIn explanatory analyses, an observed outcome is often included when imputing missing predictors so that their associations are preserved. In predictive machine learning, however, evaluation requires a separate consideration: preprocessing and imputation must not use information from held-out observations that would be unavailable at prediction time. Imputation belongs within the training portion of each relevant resampling split.\n\nDerived variables require consistent handling. A total, ratio, or squared term may be recalculated from its component variables after updates, a procedure often called passive imputation. Maintaining an arithmetic identity in this way does not by itself guarantee compatibility with the analysis model.\n\nClustered, longitudinal, and survey data introduce additional structure. Ordinary single-level MICE can misrepresent within-group dependence or uncertainty if it ignores clustering. Multilevel conditional models, appropriately specified cluster effects, or survey-aware procedures may be needed. Structural missingness—such as a response that is not applicable—must also be distinguished from an unknown but meaningful value."
    },
    {
      "heading": "Applications, limitations, and software",
      "text": "MICE is used in survey research, observational studies, economics, and other settings with incomplete multivariable data. Its main practical advantage is the ability to combine variable-specific models without requiring a simple joint distribution for the entire dataset.\n\nImportant limitations include sparse categories, separation in logistic models, strongly correlated predictors, weak information about heavily incomplete variables, and unsupported extrapolation. Conditional models may need regularization or other stabilization. No imputation procedure can reliably recover information absent from both the observed data and the assumptions supplied by the model.\n\nThere is no universally sufficient number of imputations. The required number depends on the amount of missing information and the desired precision of estimates, standard errors, intervals, and tests. A number adequate for a stable point estimate may be inadequate for stable uncertainty estimates.\n\nWidely used implementations include the R package mice and Stata’s mi impute chained. Software differs in its conditional models, predictor specifications, handling of parameter uncertainty, diagnostics, and support for complex data structures. The method name alone therefore does not fully specify an imputation analysis."
    }
  ],
  "links": [
    "missing-data-imputation",
    "multiple-imputation",
    "joint-probability-distribution",
    "sensitivity-analysis",
    "posterior-distribution",
    "linear-regression",
    "logistic-regression",
    "statistical-interaction",
    "variance",
    "standard-error",
    "confidence-interval",
    "students-t-distribution",
    "gibbs-sampling",
    "markov-chain-monte-carlo",
    "posterior-predictive-checking",
    "machine-learning",
    "regularization"
  ],
  "sources": [],
  "license": "https://creativecommons.org/licenses/by/4.0/"
}