Entry/Exit Normalization Protocol¶
Normalization protocol — instantiates Selection–Transmission Change Attribution
Fixes how entrants and exiters enter the weights so that churn in the population does not masquerade as real change in the weighted mean.
A weighted mean computed over a churning population is only as trustworthy as its convention for who counts. Entry/Exit Normalization Protocol sets that convention: it fixes how entrants and exiters enter the weights across the transition window, so that population turnover is not silently read as substantive change. Its defining act is normalizing the weight column — anchoring to a base weighting, deciding whether joiners come in at full or phased weight, deciding whether leavers are dropped or held — which together define the counterfactual weighting against which any later selection term is measured. It handles weights, not identities and not values: it never matches units to each other and never computes a channel. It is the discipline that makes two states' weighted means comparable in the first place.
Example¶
A country's average hourly wage rose 4% between two years, and an economist suspects the headline flatters reality. The composition of who is employed changed: low-wage workers were laid off during a downturn (exit) and a high-wage tech sector expanded hiring (entry). Before the average can be trusted, the protocol sets the rule for the leavers and joiners. It re-weights both years to a fixed base-year employment structure, so the reported change reflects wage movement within a held-constant mix rather than the disappearance of low earners. Under the naïve headline the wage "rose" 4%; normalized to constant employment shares it rose only 1.5% — the other 2.5% was exit composition, the mean drifting up simply because low-wage jobs left the denominator. Reporting both numbers, with the convention disclosed, keeps a survivorship artifact from being sold as a raise.
How it works¶
Declare the base weighting — which state's shares (or an average of both) anchor the comparison. Set the entrant rule: admit new units at full weight, phase them in, or exclude them. Set the exiter rule: drop them, carry them at their last weight, or counterfactually retain them. Renormalize so both states' weighted means are computed on a common, comparable basis. Each choice is really a statement of the counterfactual — "what the mean would be if the population mix had not turned over" — and the protocol's job is to make that statement explicit and reproducible rather than an accident of whatever rows happened to be present.
Tuning parameters¶
- Base-weighting anchor — whether the comparison is fixed to starting-state shares, comparison-state shares, or an average. Anchoring to the base isolates within-mix movement; anchoring to the end lets the new mix speak.
- Entrant admission — full weight, phased entry, or exclusion for new units. Full admission captures entry effects immediately but lets a large cohort of joiners swamp the comparison.
- Exiter treatment — dropping leavers, carrying them at last weight, or retaining them counterfactually. Dropping is simplest but is exactly where survivorship bias creeps in.
- Renormalization frequency — renormalizing once across the whole window versus at each sub-period. Frequent renormalization tracks a fast-churning population but can obscure the cumulative composition drift.
When it helps, and when it misleads¶
Its strength is that it prevents raw population churn from contaminating the weighted mean, making the two states genuinely comparable and giving every downstream term a stable weighting to build on.
Its failure mode is that the normalization convention is itself a modeling choice that can hide or manufacture change. Freezing base-year weights understates the real growth of genuinely new sectors; dropping exiters without a thought is survivorship bias, the mean flattered by the quiet disappearance of the units that would have dragged it down.[n1] The classic misuse is choosing the convention that produces the desired headline and never disclosing it. The guarding discipline is to report both the raw and normalized means, to state the entrant and exiter rules on the face of the result, and to treat a large gap between raw and normalized as a signal that composition, not substance, is moving the number.
How it implements the components¶
unit_weight_measure— it defines and normalizes the weight column so each state's weighted mean is interpretable and the two are comparable.counterfactual_weight_or_value_baseline— the normalization convention is the counterfactual weighting ("what if the mix had not turned over") the comparison is read against.state_pair_or_transition_window— it reconciles weights specifically across the declared two-state window, joiner to leaver.
It matches no units to each other (unit_correspondence_map — that is the Lineage or Panel Correspondence Matrix) and computes no channel (selection_term_definition, transmission_term_definition). Its nearest twin is that matrix: this protocol fixes the weights of entrants and exiters so the mean stays comparable, while the matrix fixes their identity correspondence — the separating component is the weight normalization it owns versus unit_correspondence_map.
Related¶
- Instantiates: Selection–Transmission Change Attribution — it supplies the comparable, normalized weights every decomposition depends on.
- Sibling mechanisms: Price Equation Decomposition Table · Covariance Selection-Term Calculation · Within-Unit Change Assay · Lineage or Panel Correspondence Matrix · Composition-vs-Transformation Dashboard · Decomposition Residual Reconciliation Workflow · Selection–Transmission Sensitivity Analysis
Editorial Notes¶
Form Classification¶
Form family: Analysis, Modeling & Optimization
Rationale: The mechanism applies declared entrant, exiter, and base-weight choices and renormalizes two states onto a common counterfactual basis so churn does not distort the mean.
Nearest alternative: Protocol, Workflow & Routine — The choices must be reproducible, but the operative result is a normalized statistical comparison rather than coordination through an ordered workflow.
Review outcome: Adjudicated after independent review; high confidence.
Origin Attribution¶
Primary origin: Statistics & Experimental Design
Origin pattern: Convergent development
Present-day reach: Multi-domain
Rationale: Longitudinal statistics cohered weighting and normalization rules that distinguish population-composition churn from within-unit change.
Related originating lineages:
- Biology & Ecology — Population genetics supplies explicit decomposition of selection, transmission, entry, and exit effects.
- Economics & Finance — Index-number and panel analysis supplies entrant and exiter treatment that preserves interpretable aggregate change.
Review resolution: The current reviewers agree that statistics_experimental_design is primary. For the reported differences (alternate_origin_disagreement, origin_mode_disagreement, encyclopedia_synthesis_disagreement), the evidence supports convergent, multi_domain, and biology_ecology, economics_finance; these choices preserve materially formative origins without conflating later domain reach.
Encyclopedia synthesis: The exact catalogued form synthesizes established practice rather than reproducing a single standard historical label.
Review outcome: Reconciled after independent review; high confidence.
Notes¶
[n1] Survivorship bias is the distortion that arises when units that drop out of a population are silently excluded, leaving the surviving set unrepresentative. In a churning weighted mean it makes the average appear to improve simply because the units that would have lowered it have exited the denominator. ↩