---
title: Copas-Jackson Bound in Meta-Analysis
url: https://www.emergentmind.com/topics/copas-jackson-bound-c-j-bound
type: topic
---

# Copas-Jackson Bound in Meta-Analysis

Searching arXiv for papers on the Copas-Jackson bound and closely related selection-model work in meta-analysis.
arXiv search query: "Copas-Jackson bound publication bias meta-analysis"
The **Copas-Jackson bound (C-J bound)** is a worst-case bound for publication-bias-induced bias in random-effects meta-analysis. In the formulation revisited in recent work, the meta-analytic outcome satisfies \(y \sim N(\mu,\sigma^2)\) with \(\sigma^2=s^2+\tau^2\), and publication occurs according to a selection probability \(p(y,\sigma)\) [2508.17716]. The original C-J construction provides an analytic envelope for the bias of the naive observed-data estimator over a nonparametric class of selection models defined by monotonicity of the marginal publication probability in study standard error. More recent work extends this construction to a broader class of selection mechanisms, especially those driven by \(t\)-statistics or statistical significance, while separate work on Copas’ parametric selection method emphasizes that publication-bias adjustment can be highly sensitive to the assumed selection mechanism [2508.17716]; [2007.15955].

## 1. Statistical formulation

In the C-J framework, the random-effects meta-analysis model is
\[
y \sim N(\mu,\sigma^2), \qquad \sigma^2=s^2+\tau^2,
\]
where \(s\) is the within-study standard error and \(\tau^2\) is the between-study heterogeneity [2508.17716]. Publication bias is represented through a selection probability \(p(y,\sigma)\), and the marginal publication probability given \(\sigma\) is defined by
\[
p(\sigma)=\Pr(\text{selected}\mid \sigma)=\int \Pr(\text{selected}\mid y,\sigma)f(y\mid \sigma)\,dy.
\]

The key structural restriction in the original C-J bound is **Assumption \(A_0\)**: \(p(\sigma)\) is a **non-increasing** function of \(\sigma\) [2508.17716]. Its interpretation is that studies with smaller standard errors, typically larger studies, are more likely to be published.

This formulation is explicitly nonparametric at the level of the publication mechanism. Rather than specifying a fully parametric selection model, it bounds the bias induced by publication selection over a class of admissible selection functions satisfying the monotonicity condition. That feature distinguishes the C-J bound from parametric Copas-style sensitivity analyses, which posit a latent selection equation with fixed sensitivity parameters and then vary those parameters over a grid [2007.15955].

## 2. Original analytic bound

Under Assumption \(A_0\), Copas and Jackson derived an analytic worst-case bound for the publication-bias-induced bias \(b\) of the naive observed-data estimator [2508.17716]. In the notation reproduced in the later paper, the bound is
\[
|b| \leq \frac{E_O\left(\sigma^{-1}\right)}{E_O\left(\sigma^{-2}\right)} \frac{\phi\left\{\Phi^{-1}(p)\right\}}{p},
\]
where \(p=\Pr(\text{selected})\) is the overall marginal publication probability, \(f(\sigma)\) is the population density of study standard errors, and \(E_O\) denotes expectation among published studies.

With observed standard errors \(\sigma_i=(s_i^2+\tau^2)^{1/2}\), the empirical C-J bounds are given by
\[
L_{CJ}=-\frac{\sum_{i=1}^N \sigma_i^{-1}}{\sum_{i=1}^N \sigma_i^{-2}} \frac{\phi\left\{\Phi^{-1}(p)\right\}}{p},
\qquad
U_{CJ}=\frac{\sum_{i=1}^N \sigma_i^{-1}}{\sum_{i=1}^N \sigma_i^{-2}} \frac{\phi\left\{\Phi^{-1}(p)\right\}}{p}.
\]

The practical significance of the original C-J result is that it yields a closed-form worst-case expression. This allows publication-bias sensitivity analysis without fully specifying the selection process. The bound is symmetric around zero in the reported formulation, and its width depends on the published-study distribution of \(\sigma\) and on the assumed overall publication probability \(p\) [2508.17716].

## 3. Scope of the original model class

The main limitation of the original C-J bound is the restrictiveness of Assumption \(A_0\). In the later generalization, this restriction is described as allowing only selection mechanisms whose **marginal publication probability is monotone in standard error** [2508.17716]. That accommodates some intuitive size-based publication preferences, but it does not cover all practically relevant publication processes.

Two model families are particularly informative in this regard. The later paper examines the **Copas-Heckman selection model** and several **\(t\)-statistics type selection models**, including 1-logit, mlogit, exp, 2-probit, and 2-logit [2508.17716]. It shows that the Copas-Heckman model satisfies \(A_0\) because its marginal selection probability is monotone in \(s\) when \(\gamma_1\ge 0\). By contrast, at least some \(t\)-statistics type models do not satisfy \(A_0\). For the 2-probit model, the derived marginal representation can have a turning point, so it is not globally non-increasing in \(\sigma\).

A concise comparison is useful.

| Feature | Original C-J bound | Generalized C-J-type bound |
|---|---|---|
| Structural assumption | \(A_0\): \(p(\sigma)\) non-increasing in \(\sigma\) | \(A_0 \cup A_1\) |
| Model classes covered | Copas-Heckman-type mechanisms | Copas-Heckman-type and \(t\)-statistics type mechanisms |
| Form of bound | Closed-form analytic bound | Monte Carlo + nonlinear programming |

This restriction matters because many publication processes are driven more directly by statistical significance or \(t\)-statistics than by standard error alone. The later paper therefore treats the original C-J class as too narrow for a broad robustness analysis [2508.17716].

## 4. Generalization via the \(\mu\)-representation

To weaken the monotonicity restriction, the generalized framework rewrites the random-effects model as
\[
y \sim N(\tilde{\mu}, s^2), \qquad \tilde{\mu}\sim N(\mu,\tau^2),
\]
and introduces the **\(\mu\)-representation** of the selection model [2508.17716]. For a selection function \(p_0(y,s)\), it defines
\[
p_1\left(\tilde{\mu}, s\right)=\int p_0\left(y, s\right) \frac{1}{s} \phi\left(\frac{y-\tilde{\mu}}{s}\right) d y,
\]
followed by
\[
p_2\left(s\right)=\int p_1\left(\tilde{\mu}, s\right) \frac{1}{\tau} \phi\left(\frac{\tilde{\mu}-\mu}{\tau}\right) d \tilde{\mu}.
\]

The relaxed structural condition is **Assumption \(A_1\)**: there exists \(\mu_*\) such that either \(p_1(\tilde{\mu}, s)\) is non-increasing in \(s\) for \(\tilde{\mu}\le \mu_*\) and non-decreasing for \(\tilde{\mu}>\mu_*\), or the reverse monotonicity pattern holds [2508.17716]. The paper characterizes this as a **single turning point** structure.

The associated model class is
\[
\boldsymbol{\mathscr A_1}=\{p(y,s): \text{its }\mu\text{-representation }p(\tilde{\mu},s)\text{ satisfies Assumption }A_1\}.
\]
The generalized theory then considers the class
\[
\mathscr A_0 \cup \mathscr A_1.
\]

The importance of this extension is that it captures the behavior of many \(t\)-statistics type selection rules. The paper proves that all \(t\)-statistics type selection models in its list satisfy \(A_1\), stated as
\[
\mathscr B_1 \bigcup \mathscr B_2 \subset \mathscr A_1.
\]
This broadens the admissible selection mechanisms from monotone standard-error selection to a family that can encode threshold-like or turning-point behavior around a latent mean parameter [2508.17716].

## 5. Bias representation and numerical construction

Under the generalized model class, the overall marginal selection probability becomes
\[
p=\int p_2\left(s\right) f\left(s\right) d s,
\]
and the bias is represented by an integral expression involving \(p_0\), \(f(s)\), and the latent-normal variables \(z\) and \(w\) [2508.17716]. The published-study density of \(s\) is
\[
f_o(s)=\frac{\Pr(\text { select }\mid s) f(s)}{\Pr(\text { select })}=\frac{p_2(s) f(s)}{p}.
\]

Unlike the original C-J bound, this generalized bias expression does **not** simplify to a closed analytic bound [2508.17716]. The paper therefore develops a computational procedure based on Monte Carlo approximation and constrained optimization. The inner integrals are approximated with draws \(w_{k_1}\sim N(0,1)\) and \(z_{k_2}\sim N(0,1)\), and the extremal bias is characterized through an optimization problem \(OPT1\) with constraints:

- **(C1)** \(0 \le p_{i,k_1,k_2}\le 1\)
- **(C2)** averaging identities from the definitions of \(p_1\) and \(p_2\)
- **(C3)** monotonicity structure implied by Assumption \(A_1\)
- **(C4)** overall selection probability

Because unpublished studies are unavailable, the paper replaces (C4) with
\[
p \le 1 / N \sum_{i=1}^N p_i,
\]
obtaining a relaxed problem \(OPT2\), justified by a Cauchy–Schwarz argument [2508.17716]. The optimization is solved using **nonlinear programming with linear constraints**, implemented with software such as the **OPTMODEL procedure in SAS**, with \(K_1=K_2=1000\) in the reported examples.

The resulting bounds are denoted \(L_{\mathscr A_1}\) and \(U_{\mathscr A_1}\). The paper’s practical recommendation is the **extended bound**
\[
L_{\mathrm{extended}} = \min (L_{\mathscr A_1}, L_{CJ}), \qquad
U_{\mathrm{extended}} = \max (U_{\mathscr A_1}, U_{CJ}),
\]
which combines the original analytic C-J bound with the broader numerical bound over \(\mathscr A_1\) [2508.17716].

## 6. Empirical performance and interpretive consequences

The simulation evidence in the generalized-bound paper is organized around two data-generating families: **Copas-Heckman** and **2-probit** selection models [2508.17716]. For Copas-Heckman scenarios, the original C-J bound performed well and covered the true bias in **most cases**, especially when \(p\ge 0.3\). For 2-probit selection, the original C-J bound often failed badly: in one setting, the bias exceeded the C-J bound in **over 90%** of simulated datasets, and in another, in **over 60%**. The proposed extended bound had much better coverage; failures were mostly limited to **extremely low \(p\)** values, and the bound length was typically about **1 to 3 times** the C-J bound length.

The same paper reports two real-data applications. In the **prophylactic corticosteroids meta-analysis**, with 14 RCTs and essentially no between-study heterogeneity (\(\hat\tau=0\)), several standard \(t\)-statistics type sensitivity analyses produced adjusted estimates not covered by the C-J bound, whereas the **extended bound almost always covered** the adjusted estimates across \(p=0.9\) down to \(0.1\) [2508.17716]. In the **Clopidogrel meta-analysis**, with 12 studies and small heterogeneity (\(\hat\tau=0.241\)), the C-J bound again failed to cover several sensitivity-analysis estimates, while the extended bound covered almost all adjusted estimates for realistic \(p\).

A related, but methodologically distinct, line of evidence comes from work on **Copas’ method**, which is a **selection-model approach** for correcting publication bias in an aggregated-data meta-analysis [2007.15955]. That paper does **not** derive or analyze a Copas-Jackson Bound explicitly; instead, it studies the broader Copas selection-model framework. Its simulations show that Copas’ method performs best when the data are generated under Copas selection, becomes less effective when selection is based on significance, and performs poorly when selection is based on standardized effect size \(D_i/S_i\), especially with heterogeneity. The paper’s stated practical message is that Copas’ method is **not robust** to publication-bias mechanisms other than the one it assumes and recommends improving it “to make it more robust against different forms of publication bias” [2007.15955].

Taken together, these results support a precise interpretation. The original C-J bound is most credible when the publication process is plausibly monotone in standard error, while the generalized C-J-type bound is intended for a broader class that includes \(t\)-statistics-driven selection [2508.17716]. A plausible implication is that publication-bias sensitivity analysis should be read as mechanism-conditional unless the admissible selection class is itself sufficiently broad.

## 7. Terminological ambiguity and unrelated uses of “Copas-Jackson”

The expression **“Copas-Jackson bound”** refers, in the meta-analysis literature, to the publication-bias bound associated with Copas and Jackson and its later generalizations [2508.17716]. It should not be confused with unrelated uses of the surname **Jackson** in other fields.

A prominent example is the comment on Jackson and Sonnenschein’s linked-decisions mechanism, which corrects a bound in the definition of **approximate truthfulness** [2205.03352]. In that mechanism-design setting, the original bound-based notion is
\[
\#\text{lies} = K\, d(\marg u_i, P_i^K),
\]
whereas the corrected notion is
\[
\#\text{lies} \le (\#U_i-1)\,K\, d(\marg u_i, P_i^K).
\]
The paper shows that **permutation-truthful** strategies imply the loosened inequality, not the original exact equality, and that this weaker bound is still sufficient because the **fraction of lies vanishes** as \(K\to\infty\) [2205.03352].

That correction is conceptually separate from publication-bias sensitivity analysis. It concerns incentive compatibility, permutation-truthfulness, and asymptotic efficiency in linked decisions, not worst-case bounds for publication selection in meta-analysis. The overlap in surnames can therefore generate a nomenclatural confusion that is substantive only at the level of terminology, not theory [2205.03352].

Source: https://www.emergentmind.com/topics/copas-jackson-bound-c-j-bound