Page-only tool · authored hypothesis

Inspect the link
between a gift and a model.

A dollar amount or depth description cannot tell you what a gift achieved. Make one source link explicit, test the calculation and keep its limits beside the result.

A conditional case, with unknowns visible

Charitology studies the signals, channels, work, outcomes and feedback around giving. This desk helps inspect one proposed link in that system. It does not measure delivery or donor causation, judge a charity or treat depth as effectiveness. Relief, repair, prevention, structure and source describe different mechanisms; every level may be needed.

One selected gift can produce a conditional modeled estimate with authored assumptions. Other gifts and the unmapped portion remain unknown. Sample gifts and illustrative sources stay hypothetical. There is no lifetime impact total or combination of outcomes in this first method.

Start with the Giving Dashboard’s native JSON export and one Model Kit selected-version export. A compatible source needs a referenced artifact checksum, nominal USD in the gift year, an exact program period and a positive lower cost stated as exact plain-decimal cents. Unsupported or missing information stays unknown.

One gift. One explicit model.

This desk starts empty. It does not read your saved gift log or Model Kit workspace. Your chosen data stays on this page; the fixed calculation worker loads from Charitology. No payment, message or source publication is made.

Private controls stay disabled until the local desk is ready.

Prepare one conditional estimate

1. Choose the original files

Giving Dashboard native JSON export · at most 5,000,000 bytes and 2,000 rows

The export may cover a filtered year. This format does not declare complete lifetime coverage. Self-logged gifts are not payment or tax evidence.

Deliberately select one recorded gift

Inspect the gift file and selected row

No gift file chosen.

Model Kit selected-bundle JSON · at most 2 MiB

Choose the selected-version export, not a full backup or a spreadsheet. Its exact recorded run must match a fresh calculation here. An imported “matched” flag supplies no authority.

Inspect the complete selected model and reproduction

No selected model chosen.

Original source artifact referenced by this model · at most 8 MiB

Only its fresh checksum, size and source ID enter the estimate. The source body is excluded. Matching bytes establish neither factual accuracy nor scientific review, applicability or rights.

Inspect the fresh source checksum

No fresh source checksum.

2. Author the funding assumptions

Mapped gift portion in USD

Fixed donor-credit percentage · 0–100

The positive mapped portion must fit the logged gift and the source’s positive lower budget, interpreted from its exact plain-decimal cents. It cannot exceed any supported budget draw. This supplies no evidence of marginal funding or room for more funding.

Blank means unknown; 100% is never assumed. Explicit 0% gives a conditional zero only after every source gate passes. This authored allocation is not a measured probability or proof that the donor caused an outcome. The unmapped portion stays unestimated.

Interpretation of the logged gift amount

The native gift format has no explicit currency, money basis or price-year fields. This method requires a model with nominal USD in the same year and an exact cost period containing the gift date. It supplies no exchange-rate or inflation adjustment.

Basis for that money interpretation Why this exact gift purpose, program and population fit this source version Why proportional funding is an acceptable assumption for this portion Source or assumption behind the fixed donor-credit fraction Exclusions, uncertainty and reasons this conditional estimate may not apply I acknowledge the source model’s counterfactual, independence, duration, discount and harms limits, and the added proportional-funding and fixed-credit assumptions. Applicability, additionality, delivery, fees and omitted harms remain uncertain.

3. Review, then save the exact copy

Unknown — no current reviewed conditional estimate.

Each matched source cost-per-outcome draw is transformed using the mapped portion and fixed donor credit, then the results are sorted again. This is not division by a source median. Separate source arrays are never paired. The model already includes its own counterfactual and duration adjustments; they are not applied twice.

Complete outgoing text · includes private source and model writing

I reviewed this exact outgoing JSON, including its private text, source bindings and limits.

Only the selected gift row is included, with the original file’s checksum. The exact selected model is included. Other gift rows and the source artifact body are excluded. A checksum is not an evidence grade or a charity rating.

Reopen a saved estimate

Choose the actual saved estimate JSON · at most 4 MiB

Historical inspection · current estimate stays Unknown

No saved estimate inspected.

Replace the writing and selected model shown here with this packet’s literal writing and exact nested model. No current qualification is restored.

Then deliberately choose the original gift export and source artifact, reproduce the exact model, acknowledge the limits and select Requalify the inspected saved case. A changed file, row, mapping, model or numerical fingerprint cannot pass as the same case. Review a new conditional case prepares a separate case without claiming that the saved result was requalified.

Keep unfinished writing

Nothing here is saved automatically. Leaving, reloading or browser-history restoration clears private choices. Download literal unfinished fields before leaving. This recovery contains no chosen files or current qualification.

Inspect an unfinished-writing JSON copy · at most 4 MiB

Inspect literal unfinished fields before replacement

No unfinished writing inspected.

Replace the writing shown here with these inspected fields. Choose and qualify original sources separately.

What the range means

The fixed source simulation supplies cost-per-outcome draws. This method applies your mapped amount and fixed donor-credit fraction to each draw, then calculates new percentiles. The p5–p95 range describes those entered model assumptions. It is not a confidence interval or a probability that the gift worked. Uncertainty in applicability, additionality, fungibility, credit and omitted harms is not modeled.

Fresh numerical reproduction checks software consistency. Your source description, checksum, applicability and allocation remain authored claims. They do not establish scientific review, source rights, marginal funding or actual impact. This desk neither endorses nor rates a named organization.

Read the complete adjacent arithmetic method · Inspect the open consumer and file contracts · Read Charitology’s Method