The Genealogical Proof Standard is less a checklist than a decision discipline: the hard part is judging when research is reasonably exhaustive, which evidence outweighs which, and how to explain your resolution in writing. This guide teaches the five GPS elements as decisions you make, with worked scenarios, an evidence-classification table, a drafting exercise with a scoring rubric, and an adaptable preparation sequence. Work one problem at a time, classify every source you touch, and write a short proof summary each week so the writing skill develops alongside the research skill rather than after it.
What the Five GPS Elements Require — and What 'Proof' Means Here
The GPS asks for five things: reasonably exhaustive research, complete and accurate citations, thorough analysis and correlation, resolution of conflicting evidence, and a soundly reasoned, coherently written conclusion. It governs the quality of a conclusion, not a guarantee.
In everyday use, 'proof' suggests certainty, and that everyday meaning causes classification errors of its own. Under the GPS, a conclusion is considered proved when all five elements are satisfied and the result is the most logical explanation of the assembled evidence; it remains open to revision if new evidence appears. Practice distinguishing a defensible conclusion from a merely confident one, because assessment questions of this type ask you to judge whether reasoning and documentation support an answer, not whether the answer sounds plausible.
Keep the GPS separate from adjacent concepts. It is not the Certified Genealogist credential, which is a portfolio-based certification of a practitioner's overall competence; the GPS is a standard any researcher can apply to any conclusion. Blurring the two leads to wrong answers about who must use the standard and when. For current administrative details about certification itself, consult the Board for Certification of Genealogists directly rather than relying on secondary summaries.
Reasonably Exhaustive Research: When Is 'Enough' Enough?
'Reasonably exhaustive' means searching the record types an informed researcher would expect to bear on the question, in the places and periods where they exist — then documenting what you searched, including what you did not find.
Take a paper scenario: you want a girl's mother's name, and an 1850 U.S. census household shows a woman of the right age listed above the child. The plausible mistake is stopping there. That census does not state relationships, so the adjacency is indirect evidence at best — an unrelated woman of that age fits the same record. The better decision is to treat the census as a lead and name the record types that would state or imply the relationship directly.
Exhaustiveness is demonstrated, not asserted: identify the plausible sources — a probate file naming heirs, a marriage record, baptismal entries, land transactions — determine which survive for that place and period, and record the searches that came up empty along with the ones that produced items. This matters because a conclusion resting on a single indirect item cannot survive a conflicting record, and the GPS is built around anticipating and resolving exactly that situation.
Classifying Sources, Information, and Evidence Without Mixing the Labels
Classify on three independent axes: source (original versus derivative), information (primary versus secondary informant), and evidence (direct, indirect, or negative). One document can carry items of different quality on each axis at once.
These axes move independently, which is where classification errors begin. A derivative source can carry primary information: a recorded copy of a marriage license is derivative, but the couple's own statements on it are primary. Conversely, an original document can carry secondary information, such as a headstone cut decades after a death. Grade each layer separately instead of assigning a single quality score to the whole document.
Apply the labels in a two-column working note: what the item is (source class) and what it tells you (information quality and evidence type). In the census scenario above, the enumeration is an original source, an age on it is likely secondary information supplied by whoever answered the enumerator, and the implied parentage is indirect evidence. Writing that chain down is precisely the testing of evidence against all sources that the analysis-and-correlation element calls for.
- Original source: the first form of the record — the loose probate packet, the register itself.
- Derivative source: a later copy, transcription, abstract, or index of that record, which can introduce transcription error.
- Primary information: a statement by someone with firsthand knowledge of the fact, such as the bride reporting her own age.
- Secondary information: a statement reported by someone relying on hearsay, memory, or assumption.
| Evidence label | What it does | Paper example | Typical misreading |
|---|---|---|---|
| Direct | Answers the question by itself | A death record naming the decedent's father | Treating the informant's statement as verified fact |
| Indirect | Suggests an answer only when combined with other items | Household adjacency in a pre-1880 census | Reading adjacency as a stated relationship |
| Negative | The absence of an expected record, usable only if the record should exist and be searchable | No marriage entry in a register series known to be complete for those years | Treating any failed search as proof of the opposite |
Resolving Conflicting Evidence Instead of Picking the Majority
A conflict exists when items cannot all be correct. Resolution means analyzing each item's origin and informant, gathering better evidence when needed, and explaining which item you weighted and why — not counting votes.
Second scenario: a woman's birthplace is Kentucky in three census years, Ohio on her death certificate, and Kentucky in a family Bible. The plausible mistake is a majority vote — Kentucky wins three to two. The better decision grades the informants instead: who supplied each statement, and what did that person have firsthand knowledge of? The death certificate's informant may never have known the mother; census answers sometimes came from neighbors.
The stronger move is to seek an item that outranks both camps, such as the woman's own statement in a marriage application or an obituary prepared from her papers, then state the weighting explicitly in your conclusion. This matters because conflicts are where conclusions break: an unexplained conflict leaves a reader unable to judge the result, and a vote-counting resolution can carry a wrong birthplace forward through every later generation of research.
Writing the Conclusion: Citations, Visible Reasoning, and What to Leave In
The fifth element demands a conclusion that is soundly reasoned and coherently documented: every fact cited, the path from evidence to answer shown, and the handling of any conflict visible to the reader.
Complete and accurate citations do more than attribute: they let a reader find the item, judge its nature as original or derivative, and locate the exact place within it that the information appears. Practice writing citations that answer those three questions for every record in your scenarios. When you cannot say who supplied a fact, that citation gap is also telling you something about the quality of your underlying evidence.
For the reasoning, draft a one-paragraph proof summary: the question, the key items, what each contributes, the conflict and its resolution, and the answer. If a sentence cannot be traced to a cited item, either add the citation or cut the claim. This habit exposes unsupported leaps while the draft is still cheap to fix, and it builds writing practice in the same compact form conclusions are actually read.
A Self-Check Exercise: Score a Draft Against the Five Elements
Pick one ancestor question, work from what you already hold, and draft a 150-word proof argument. Score each GPS element zero to two. Expected first-attempt observations: citation gaps around informant identity and at least one skipped conflict.
Do this in one sitting of roughly ninety minutes: thirty minutes compiling what you already hold, thirty drafting, thirty scoring. Use the rubric below and do not research while drafting, because the point is to see what your current evidence supports. Repeat weekly with a different question so the scores reflect your habits rather than one unusually easy case.
Expected observations on a first attempt: a 1 on citations (facts cited but source nature unstated) and a 1 on conflict (a discrepancy you knew about but skipped). A 2 on research usually signals the question was too narrow; widen it next round. Treat these scores as learning milestones that show which element to drill — they measure your draft, not any assessment outcome.
- Research: 0 = one record type used; 1 = several types, searches undocumented; 2 = plausible types covered and searches documented, including empty ones.
- Citations: 0 = none; 1 = partial, an item not locatable or its nature unstated; 2 = every fact traceable with source class identified.
- Analysis and correlation: 0 = items merely listed; 1 = items compared; 2 = items combined to produce the answer.
- Conflict: 0 = unmentioned; 1 = mentioned but unresolved; 2 = resolved with the weighting explained.
- Conclusion: 0 = a guess; 1 = an answer with implicit logic; 2 = an answer that follows reasoning stated in the text.
An Adaptable Six-Week Preparation Sequence
Weeks 1–2: learn the five elements and classification axes using one record set. Weeks 3–4: two rubric-scored proof-argument drafts. Weeks 5–6: conflict-resolution drills and timed one-paragraph summaries.
In weeks 1 and 2, work through a single probate file or church register and label each item's source class, informant, and evidence type, checking labels against the definitions rather than intuition. Weeks 3 and 4 add drafting: two proof arguments on unrelated questions, each scored on the rubric, with the weakest element driving the focus of the next draft.
Weeks 5 and 6 target the judgment skills: construct three paper conflicts — change a birthplace, a date, and a name across records — and write a weighting explanation for each, then practice one-paragraph summaries under a timer. Adjust the pace to the time you have; the order — classify, draft, resolve, compress — is what matters, not the calendar.
References and further reading
Use these references to explore the concepts and check the latest information from the relevant organizations.