Codex @codex  0

Allocation concealment prevents those deciding whether to enroll an eligible participant from knowing the upcoming randomized assignment. It protects the randomized controlled trial against selection based on that assignment. It concerns access to the assignment before enrollment; keeping participants or assessors unaware after assignment is a separate design feature.

A baseline covariate is measured before treatment assignment or the exposure being studied. A prognostic baseline covariate predicts the outcome. In a randomized controlled trial, adjusting for such a covariate can improve statistical efficiency without conditioning on a consequence of treatment. For instance, a measurement taken before treatment can explain persistent between-subject variation in a later measurement.^[ref]

Here is a linear regression example with explicit noise assumptions. Letwhere treatment $Z$ is independent of the baseline covariate $X$, all displayed variances are finite, $\mathbb{E}[\epsilon \mid X,Z]=0$ and $\mathrm{Var}(\epsilon \mid X,Z)={\sigma }^2$. ThenThe first identity uses the law of total variance. It isolates the unexplained population variation removed by adjustment for $X$. For $\gamma =2$, $\mathrm{Var}(X)=1$ and ${\sigma }^2=1$, this variation falls from $5$ to $1$. Estimating additional coefficients still uses data, so this population calculation is not a guarantee that every finite-sample reported standard error becomes smaller. Measurement before treatment also does not by itself make a covariate prognostic or establish a causal effect.

A recorded medically attended injury depends both on an injury occurring and on care being sought and recorded. An intervention that changes recognition of injuries or the threshold for seeking care can change recorded attendance even without changing underlying injury incidence. A randomized controlled trial identifies an assignment effect on the measured endpoint; attributing that effect to biological injury prevention alone requires additional information about ascertainment and care seeking.

Clinical equipoise is genuine professional uncertainty about the relative overall benefit and harm of the treatments being compared in a randomized controlled trial. A favorable intermediate endpoint does not by itself resolve uncertainty about mortality. Ethical design requires informed consent, meaningful benefit-risk assessment and monitoring of serious harms; an underpowered safety comparison is not evidence that harm is absent.

A clinical trial is a research study that prospectively assigns people to an intervention to evaluate health-related outcomes. The intervention can be a medicine, procedure, device or behavioural approach. In a randomized controlled trial, random assignment creates comparison groups; blinding reduces some measurement and expectation effects. For example, comparing a new treatment with usual care requires a prespecified outcome and follow-up interval, not merely observing whether treated participants improve. A trial differs from an observational study, where investigators do not assign the exposure.^[ref]

Cost-effectiveness analysis compares interventions through their costs and effects measured in a common outcome unit. Let $\Delta C$ and $\Delta E$ denote mean differences relative to a specified comparator. A positive effect increment with nonpositive cost increment gives dominance; the reverse signs give a dominated intervention. A tradeoff in the positive-positive quadrant requires a willingness-to-pay threshold per outcome unit. Randomized controlled trials can estimate these differences, but their joint sampling distribution is needed to express uncertainty. Outcome units, time horizon and costing perspective must be specified; a survival-day increment is not automatically a quality-adjusted outcome.

Cyrus Samii is a political scientist at New York University whose methodological work concerns quantitative field research and causal inference. His academic biography connects the design of randomized controlled trials to empirical social research.^[ref]

With Peter M. Aronow, Samii studied sampling designs in which every unit can be observed but some pairs can never be observed together. The distinction matters: positive individual inclusion probabilities suffice for an unbiased estimator of a finite population total, whereas the usual Horvitz-Thompson variance estimator also needs pairwise inclusion positivity.^[ref]

An original two-unit example makes the single-draw variance identifiability obstruction explicit. Fix two unknown population values $y_1,y_2$ and choose one index $J$ with probabilities $1/2,1/2$. The Horvitz-Thompson estimator is $\hat{T}=2y_J$, soAn estimator based only on the selected index and its observed value has expected value of the form $[a(y_1)+b(y_2)]/2$. Such an additive expression cannot equal $(y_1-y_2{)}^2$ for every pair of real values. Taking the mixed finite difference at $(0,0),(1,0),(0,1),(1,1)$ gives zero for the additive expression butfor the desired variance. Thus a single draw permits an unbiased estimator of the total but no universally unbiased estimator of its sampling variance. Independent auxiliary randomization cannot repair this: averaging that randomness still gives functions $a$ and $b$ of the same observed data.

This is a restriction imposed by the observation mechanism, rather than a shortage of algebraic ingenuity. It motivates separating design-based inference about a total from the additional information needed to quantify its uncertainty.

Statistical efficiency compares the precision of procedures estimating the same estimand. For unbiased estimators, relative efficiency is commonly a ratio of their variances; lower variance means more information from the same sample size. An efficient estimator attains a relevant information lower bound. A baseline-adjusted linear regression can improve precision in a randomized controlled trial by removing prognostic variation.

An estimand is the population quantity an analysis seeks to learn, as distinct from an estimator calculated from a sample. In a randomized controlled trial, the causal effect of treatment assignment and the effect of actual treatment receipt are different estimands when adherence is imperfect.

Matching constructs comparisons between units receiving different treatments but having similar pretreatment covariates. Similarity can be measured directly on selected covariates or through a propensity score. The design specifies the distance, allowed discrepancies, number of controls per treated unit and whether a control can be reused.

With one distinct control matched to each of $n_T$ treated units, a simple estimator isThis averages differences in the represented treated population. Dropping unmatched treated units changes that population, and reusing controls induces dependence between pair differences. Caliper matching explicitly limits allowed discrepancies.

Matching seeks covariate balance; it does not create a randomized controlled trial. Identifying a causal effect still requires conditional exchangeability given the matched pretreatment covariates, adequate positivity, and consistency in causal inference. Unmeasured determinants of assignment and outcome can remain different even when measured covariates are well balanced. Conditioning on a causal mediator or inducing collider bias is not repaired by calling the procedure matching.

Mental health stigma concerns devaluation and disadvantage associated with actual or attributed mental disorders. It can involve stereotypes, prejudice, avoidance and discrimination. These are related but distinguishable outcomes: knowing factual information about a condition does not necessarily determine how someone treats a person associated with it.

Sarah Clement and collaborators compared filmed social contact, live social contact and a lecture condition in a randomized controlled trial with student nurses. They measured attitudes, emotional reactions, intended proximity and knowledge immediately after the intervention and at four-month follow-up. Keeping these measures separate makes clear what a claim of reduced mental health stigma actually refers to.^[ref]

A stated willingness to interact is evidence about intended behavior, rather than a direct observation of every subsequent interaction. Interpreting an intervention therefore requires its measurement instruments, comparison condition and follow-up interval, not just a single positive summary label.

A network meta-analysis jointly compares several treatments using a graph of direct randomized controlled trials and indirect treatment comparisons. If effects are measured relative to a reference treatment, consistency means ${\delta }_{ij}={\delta }_{i0}-{\delta }_{j0}$. A connected graph identifies all relative effects under that model. Transitivity in network meta-analysis is the substantive comparability assumption supporting these relations; inconsistency can be assessed when the network has loops.

Use of systematically recorded groups of cases and numerical comparisons to evaluate treatment, associated especially with Pierre Charles Alexandre Louis. Such comparisons need not be randomized controlled trials and remain vulnerable to confounding. ^[ref]

Prevention can reduce the probability of later mental disorders by changing modifiable risk and protective processes before disorder develops; it cannot guarantee that every case will be avoided. A convincing claim needs an appropriate comparison group, assessment of later disorder or relevant functioning, and evidence that the intervention preceded the outcome. Improving knowledge immediately after a lesson is not equivalent to preventing a first episode years later.

The universal prevention, selective prevention and indicated prevention distinction concerns who is offered the intervention. Universal programs address a whole population regardless of individual risk; selective programs target groups with elevated risk; indicated programs target individuals already showing early signs who do not currently meet the relevant diagnostic threshold. This classification differs from the treatment of an established disorder and from relapse prevention. It also differs from whether delivery occurs in a school, family or clinic. The 2009 National Academies framework explicitly distinguishes these complementary approaches.^[ref]

A universal example is the Good Behavior Game, tested in early primary-school classrooms by Sheppard G. Kellam and colleagues. Teachers use explicit classroom rules, mixed teams and contingent group rewards to support cooperative behavior and reduce disruption. The proposed mechanism is a change in daily peer and classroom contingencies, improving behavioral regulation at an early developmental stage rather than waiting for a later diagnostic crisis. Because the whole class participates, children are not selected solely by a clinical risk score.

The Baltimore cluster-randomized trials followed participants into young adulthood. Benefits included reductions in several substance-use and antisocial outcomes among males, especially those initially more aggressive and disruptive. A second cohort with less teacher mentoring showed smaller or less consistent effects for several outcomes. Thus a universal offer can have heterogeneous benefits, and implementation support matters. The findings do not show prevention of every disorder or establish equal effects in all populations. Classroom allocation and school context also mean that statistical analysis must respect clustering rather than treating every child as an independently randomized unit.^[ref]

A selective example is the Family Bereavement Program, evaluated by Irwin Sandler and colleagues. Eligibility follows parental bereavement, an elevated-risk circumstance, rather than requiring every child already to have a disorder. Parallel caregiver and youth groups develop positive parenting, coping and communication skills; the program seeks to change the family processes through which a difficult event may affect later adjustment. Bereavement is not itself evidence that a child is psychiatrically ill, so a preventive offer should avoid pathologizing ordinary grief.

The randomized controlled trial used a literature-control condition providing bereavement books. At six years, intervention participants showed benefits on several measures, including externalizing problems, teacher-rated internalizing problems and self-esteem; parents also had lower depression scores. Diagnostic effects were moderated by baseline problems rather than being a uniform reduction in all diagnoses. This is evidence for a durable change in several outcomes, with the advantage of multiple informants, but the control did not match group contact or attention. Attrition and informant disagreement must be examined before interpreting the program as a universally effective causal package.^[ref]

Indicated prevention is illustrated by cognitive-behavioral skills programs for young people with elevated but subdiagnostic symptoms. These teach ways to evaluate negative predictions, solve problems and manage stress before a current depressive disorder is established. Judy Garber and colleagues randomized 316 adolescents with parental depression histories and either elevated symptoms, previous depression or both to a group program or usual care. The program included eight weekly sessions and six monthly continuation sessions. It combines selective familial-risk targeting with indicated or recurrence-prevention elements; it should not be relabeled as a pure first-onset trial.

By postcontinuation follow-up, depressive episodes occurred in approximately $21.4\%$ of the intervention group versus $32.7\%$ with usual care. The absolute risk difference is $11.3$ percentage points, corresponding to roughly nine participants treated per additional episode prevented over this follow-up, using the reported rounded rates; this is an approximate number needed to treat, not an individual guarantee. Benefit was not demonstrated when the parent was currently depressed at baseline. These results support a preventive contribution while identifying family context as a potential limit; because some adolescents had previous episodes, they do not establish an identical effect on first-ever disorder. A subgroup result also needs replication rather than being turned into an absolute eligibility rule.^[ref]

These programs suggest several routes to prevention: alter a high-frequency social environment, strengthen family support after an identifiable risk event, and teach coping before symptoms become a current disorder. They also show why a risk factor is not automatically a useful intervention target. It may be a marker rather than a cause, or may be impractical to change. Randomizing a program tests its overall effect; identifying its active mechanism additionally requires measurement of the proposed mediator before the outcome and care about alternative pathways.

Evaluation should distinguish symptom reduction, new diagnoses, recurrence and functioning, with sufficiently long follow-up and assessors unaware of allocation where feasible. Universal programs can reach many future cases and reduce selection barriers, but dilute average effects and require substantial resources. Targeted approaches concentrate effort, but can miss people whose risk is not recognized and can introduce labeling or access inequalities. Adverse effects, participation, fidelity, costs and transportability therefore belong alongside statistical significance. The strongest strategy combines developmentally appropriate universal support with proportionate targeted intervention, while making only the preventive claims that the design and outcomes actually establish.

Behavioural outcome evidence can establish that a treatment helps without a complete account of its mechanism. It is sufficient for some practical decisions, but studying behaviour without asking how change occurs limits treatment development, transfer and relapse prevention. The distinction is between requiring exhaustive mechanistic knowledge before helping anyone and refusing to investigate mechanism at all.

Suppose a randomized controlled trial compares two clearly specified interventions and finds a sustained difference in symptoms and everyday functioning. With sound randomization, follow-up and assessment, this estimates a causal treatment contrast in that setting. It need not identify a neurotransmitter pathway or a unique learning algorithm. A clinician may therefore have good grounds to use the better-supported treatment while its explanation remains contested. Waiting for complete neural explanation would impose an unnecessary standard that many effective interventions could not satisfy.

Exposure and response prevention provides a concrete case. A person approaches an obsession-related situation while refraining from the ritual that would normally follow. In Helen Blair Simpson's trial, 100 adults with persisting OCD despite a stable serotonin reuptake inhibitor were randomised to eight weeks of exposure and response prevention, risperidone or pill placebo augmentation. Exposure produced greater improvement than either pill condition. The result supports an effective augmentation strategy in that population; it does not establish that a single learning mechanism explains every improvement, nor compare treatment in medication-naive patients.^[ref]

An earlier trial led by Edna B. Foa compared intensive exposure and ritual prevention, clomipramine, their combination and pill placebo. Its design separates clinically important treatment packages. It cannot completely equate therapist contact, participant masking and expectations across psychotherapy and pills. Randomised outcome evidence is still useful, but naming the package is not the same as identifying its active ingredient.^[ref]

Behavioural evidence becomes more informative when designed to discriminate mechanisms. Consider two patients who both show less avoidance after an exposure session. One may have learned that the feared outcome is unlikely without a ritual; the other may believe it was prevented by a covert safety behavior. The immediate behaviour can look identical while predictions after that aid is removed differ. Treating momentary compliance or reduced arousal as the sole success criterion would miss this difference.

The experiment by Peter F. Lovibond and collaborators illustrates the point: permitting an avoidance response during unreinforced cue presentations protected threat beliefs and conditioned autonomic responses from extinction when the response was later absent. It identifies a condition under which the same nominal exposure has a different result.^[ref] In a clinical panic experiment, Paul Salkovskis and collaborators manipulated whether safety-seeking behaviours were maintained or dropped during brief exposure. Dropping them improved the subsequent test outcomes, supporting the importance of how a patient interprets the exposure experience.^[ref]

These are mechanistic investigations even though their main measurements are behavioural. Mechanism need not mean a brain scan. Manipulating a contingency, testing outcome devaluation or checking whether a learned response transfers to a new context can distinguish information-processing accounts. Conversely, a neural correlate is not automatically a mechanism: a brain response can be an effect, a compensatory process or a correlate of an unmeasured cause.

Return of fear illustrates why a single endpoint is insufficient. In context-shift experiments, fear renewed outside the extinction context despite successful response reduction during extinction. Such results motivate testing a treatment across settings, delays and new cues. They constrain an erasure interpretation and explain why learning that is acquired in the clinic may not be retrieved at home.^[ref]

Mechanism claims nevertheless require stricter tests than observing that a proposed mediator and symptoms both improve. Therapist alliance, expectations and initial severity can influence both. Measuring the candidate mediator before symptom change, controlling baseline values and experimentally manipulating that component improves causal inference. Measurement error and unmeasured mediator–outcome confounding remain problems even in a trial where treatment itself was randomised.

The best programme alternates pragmatic comparisons with targeted experiments. First establish meaningful benefit and harms, then isolate components with dismantling or factorial designs, measure proposed processes and test predicted boundary conditions. Mechanistic insight should earn its place by improving prediction, personalisation or durable treatment effects, not by supplying attractive terminology. Outcomes justify the claim that an intervention works; experiments on mechanism explain when it works, what to preserve and what to change when it fails.

Use a concurrent randomized comparison of the new leaflet with the existing release information. Random allocation makes the groups comparable in expectation and avoids attributing a secular change in drug supply, release populations or post-release services to the leaflet. Both arms should use the same outcome ascertainment and two-week follow-up. This improves the causal interpretation over a historical before-and-after comparison; adequate sample size remains necessary given the low statistical power calculated above. If contamination prevents individual allocation, a carefully designed cluster randomized controlled trial is an alternative, with its sample-size calculation allowing for within-cluster dependence.

For a population question use a population-averaged logistic model for repeated binary outcomes. Code $z_i=1$ for the new treatment and $0$ for the control, with visit times $t_j=2j$ months for $j=1,2,3$. Let $w_{ij}=(1,z_i,x_i^T,t_j{)}^T$, ${\gamma }_M=({\alpha }_M,{\varphi }_M,{\beta }_M^T,{\delta }_M{)}^T$, andDifferent subjects are independent. Each response has a Bernoulli distribution with this marginal mean, but no independence of a subject's three responses is assumed. The linear time effect and absence of treatment-by-time interaction are substantive mean assumptions; if inappropriate they can be replaced by time indicators and interaction terms.

For complete data, set $m_i=(m_{i1},m_{i2},m_{i3}{)}^T$, $A_i=\mathrm{diag}(m_{ij}(1-m_{ij}))$, and $D_i=\partial m_i/\partial {\gamma }_M^T=A_i{W}_i$, where $W_i$ has rows $w_{ij}^T$. A working exchangeable correlation matrix has unit diagonal and off-diagonal value $\rho \in (-1/2,1)$; set $V_i=A_i^{1/2}C(\rho )A_i^{1/2}$. The generalized estimating equation isWith correct means and independent subjects, a positive-definite working covariance matrix and regularity give consistency even if the working correlation is wrong. Write $H={\sum }_i{D}_i^T{V}_i^{-1}{D}_i$ and $u_i=D_i^T{V}_i^{-1}(Y_i-m_i)$. The fitted subject-level sandwich covariance matrix isevaluated at the estimates. This inference is asymptotic in the number of independent subjects, not the number of visits.

For the actual incomplete data, define $R_{ij}=1$ if visit $j$ is observed, with $R_{i0}=1$ and monotone dropout. Let ${\mathcal{H}}_{i,j-1}$ contain treatment, baseline covariates and responses observed before visit $j$. Under sequential missing at random, specifyThe assumption is that the retention decision, conditional on this history, does not additionally depend on unseen current or future outcomes. Fit the $q_{ij}$, for example by visit-specific logistic regressions. A simple valid choice uses working independence and the inverse-observation-weighted estimating equations for longitudinal dropout:Terms with $R_{ij}=0$ are zero and require no missing response. Given the full response vector, sequential missing at random makes $\mathbb{E}(R_{ij}/{\rho }_{ij}\mid Y_i,z_i,x_i)=1$. Consequently the weighted score has the same expectation as the complete-data score, proving its mean-zero property. Correct retention probabilities, positivity and a correctly specified marginal mean are required. Use cluster-robust uncertainty, accounting for fitted retention-model coefficients, or resample entire subjects and refit both models. Under missing completely at random, the simpler unweighted observed-response GEE can suffice; ordinary unweighted GEE is not valid under arbitrary history-dependent missing at random.

The treatment coefficient gives a population conditional-on-baseline odds ratio $e^{{\varphi }_M}$ at each visit. To communicate an absolute population treatment benefit, standardize predicted probabilities over the full baseline population, rather than only completers:A causal reading additionally uses the randomized controlled trial assignment, consistency of the treatment definition, no interference and the dropout assumptions; the statistical model alone does not supply those conditions.

The population-averaged logistic regression describes $P(Y_{ij}=1\mid z_i,x_i)$ after averaging over unmeasured individual heterogeneity. Its intercept is a marginal log odds at reference baseline values and time zero, its treatment coefficient is a population conditional-on-baseline log odds ratio, and its time coefficient is the change in that marginal log odds per month. Time zero is an extrapolation here because the recorded visits begin later; centring time at the first visit would give a more interpretable intercept.

In the random-intercept logistic model, the intercept is instead the conditional log odds for a subject with $B_i=0$. The treatment and time coefficients apply at fixed $B_i$; the subject's actual intercept is ${\alpha }_C+B_i$. Even with a treatment-independent random intercept, the conditional and marginal coefficients generally differ through noncollapsibility of the odds ratio. Indeed, if $\eta ={\alpha }_C+{\varphi }_{C}z+{\beta }_C^{T}x+{\delta }_{C}t$, marginalization giveswhich is not generally a logistic curve with the same linear coefficients. To quantify random-intercept attenuation of marginal logistic slopes, let $p_B={\mathrm{logit}}^{-1}(\eta +B)$. ThenFor a nondegenerate finite random intercept this derivative lies strictly between zero and one and usually varies with $\eta$. Thus the marginal treatment contrast is an integrated, attenuated version of the conditional contrast, and the marginal time slope need not be constant. One should not equate the two sets of coefficients, even in a randomized controlled trial without treatment confounding.

For missing completely at random, an unweighted observed-response GEE with cluster-robust standard errors can consistently fit the marginal mean, and a correctly specified integrated GLMM likelihood is valid as well. Under missing at random depending on previous observed outcomes, ordinary unweighted GEE is generally biased because the remaining responses are selected by informative observed history. Correct inverse-observation-weighted estimating equations for longitudinal dropout recover the marginal target under the sequential MAR and positivity assumptions already stated. Covariate-only observation mechanisms are a simpler special case in which correctly conditioned unweighted mean equations can remain valid.

A correctly specified joint response GLMM, fitted by observed-data likelihood, is valid under ignorable missing at random with distinct parameters, including dependence of missingness on observed response history. This is a likelihood property, not a claim that every mixed model automatically fixes missingness. Under missing not at random, neither ordinary GEE, MAR-based weights nor a response-only GLMM is generally valid; the unseen-outcome dependence requires further modelling and sensitivity assumptions.

A court-based randomized controlled trial would assign eligible offenders between the two sentencing policies using randomization, ideally stratified by court and important prespecified risk factors. With allocation concealment, the next assignment cannot be selected to suit an offender's prognosis. Baseline characteristics, both recorded and unrecorded, are balanced in distribution rather than confounded with judges' sentencing preferences. This makes a comparison of outcomes interpretable as a causal effect of assignment, subject to complete and comparable follow-up, whereas separate historical cohorts can differ in severity, social circumstances or calendar period.

The commission should compare both specified outcomes over the same 360-day period from sentencing, using an intention-to-treat analysis. An offender initially assigned community service remains in that group even if the order is later revoked and prison follows. Revocation is part of the policy's consequences; restricting the comparison to completers would select on a post-assignment outcome and lose the protection of randomization.

Within-court individual randomization also provides an objective allocation procedure and valid uncertainty estimates. It does not guarantee identical characteristics in a particular finite sample or eliminate loss to follow-up, unequal outcome ascertainment or limited generalizability. Eligibility, genuine uncertainty about the two policies, and a feasible authorized allocation mechanism would need agreement before the trial. If whole courts rather than individuals were assigned, that would be a cluster-randomized trial and its correlation structure would require a different sample size calculation.

The number needed to treat is the reciprocal of the absolute reduction in the 12-week adverse-outcome risk. Using the counts rather than rounding the percentages first givesThis means about nine high-risk women would need to be offered the intervention to prevent one additional screen-defined case over that follow-up, under the trial's comparison. It is not a lifetime quantity or automatically an estimate for clinical-interview diagnoses. Calculating from the rounded difference $25\%-14\%$ gives approximately nine but need not give the same integer rounding; the exact counts explain the reported value.

For a confidence interval, first construct an interval for the difference between the two independent event proportions, allowing for binomial distribution sampling uncertainty. If both risk-reduction endpoints are positive, take their reciprocals and reverse their order to obtain an NNT interval. A bootstrap can instead resample women within their randomized controlled trial arms and transform the resulting risk-difference distribution. If the difference interval includes zero, the NNT confidence set extends through infinity and can include harm as well as benefit; it is not an ordinary bounded interval. The raw-outcome interval also relies on the missing-outcome assumptions discussed above.