Study for the Boston University Genealogical Research Certificate by drilling the Genealogical Proof Standard as a sequence of decisions: classify source, information, and evidence; cite completely; correlate findings; resolve conflicts in writing; and justify conclusions. Work through realistic record scenarios and self-check your analysis against a rubric. For registration and administrative details, check the program pages at Boston University's Center for Professional Education (cpe.bu.edu), which lists Genealogy among its non-credit certificate offerings.
The Genealogical Proof Standard as a Decision Checklist, Not a Slogan
Treat the Genealogical Proof Standard (GPS) as five ordered decisions you make about every research question, not a phrase to memorize for recitation.
The GPS asks whether your search was reasonably exhaustive, whether citations are complete and accurate, whether you analyzed and correlated findings, whether you resolved conflicting evidence, and whether your conclusion is soundly reasoned. Study each element as a separate skill with its own failure mode. A search can be large but shallow; a citation can exist but omit the layer you actually used, such as the database image rather than the original register.
Build fluency by narrating the five steps aloud for a small question, such as identifying a couple's marriage date from a census and a church register. Say what you searched, what you cited, how the items agree or differ, and what your written conclusion claims. If you cannot complete the narration in about a minute, the weak step shows you exactly which element of the GPS to drill next rather than vaguely rereading notes.
Separating Source, Information, and Evidence in Every Record
Classify each record on three independent layers: its source type, the information's informant status, and the evidence's relationship to your question.
A source is the artifact itself: original, derivative (such as a transcript, abstract, or index), or authored narrative. Information is primary if supplied by an eyewitness participant, secondary if reported afterward, or indeterminate when the informant is unknown. Evidence is direct when it answers your question outright, indirect when it only helps you argue the answer, and negative when a searched source would likely have answered it but did not. The layers are independent, so an original document can hold secondary information that is only indirect evidence.
Practice with a death certificate: it is usually an original source, but the birthdate on it is often secondary information supplied by someone who was not present at the birth, and it becomes indirect evidence if your question is the exact birthplace. Compare that with a marriage record where the couple themselves report their ages. Doing this layered reading for ten documents from your own family file, on paper, trains the classification reflex that every later analysis depends on.
The table below shows the three layers side by side for a cemetery gravestone. A useful exercise is to classify a photograph you took of a stone and then classify the published transcription of the same stone, and observe exactly where the layer labels change and why.
| Layer | Question to ask | Categories | Gravestone example |
|---|---|---|---|
| Source | What kind of artifact is this? | Original; derivative (transcript, abstract, index); authored narrative | Original if you photographed the stone yourself; derivative if you used a transcription published in a book |
| Information | Who supplied the fact, and how do they know? | Primary (eyewitness participant); secondary (reported later); indeterminate (informant unknown) | Death year is likely primary from burial timing; birth year is usually indeterminate or secondary, since the carver relied on someone else's report |
| Evidence | Does the fact answer my question? | Direct; indirect; negative | Death year directly answers a death-date question but is only indirect evidence for the birth year |
| Decision | What does this mean for use? | Weight, corroboration needed, citation layer to cite | Weigh lightly against a vital record; cite the image you actually viewed |
Citations That Let a Reviewer Retrace Your Steps
A citation is complete when a stranger can locate the same source, the same image or entry, and see where you viewed it.
Study citations as layered constructions rather than formats to memorize. The first layer points to the original record: its title, the jurisdiction or institution holding it, and the record type with its date. The second layer points to the copy you actually used: the microfilm number, the database, the image path. The third layer records the specific item: page, entry name, and certificate number. Omitting a layer is the recurring weakness, because most online research never touches the original and the citation must admit that.
Drill by reverse-engineering: take five citations from published genealogical writing and mark where each layer begins and ends, then write equivalent citations for records in your own file at three depths, as if the reader had only a subscription database, only a courthouse visit, or only a FamilySearch account. A workable self-check is the retrace test, where a study partner with access to the same database finds your exact page in under two minutes. If they cannot, the missing element names itself, and that element is what to fix, not the whole style.
Scenario: Three Conflicting Birth Years in One Research File
When records disagree on a birth year, classify each item's layers, compare informant proximity, and write the reasoning, instead of picking the record that looks official.
Imagine a file with three birth years for one woman: 1874 on her gravestone, 1876 in the 1900 census, and 1872 on her death certificate, completed by her son. The tempting mistake is treating the death certificate as authoritative because it is an official vital record. That skips classification: the certificate is an original source, but its birthdate is secondary information from an informant born decades after the event, while the 1900 census age is likewise self- or household-reported and the stone reflects a carver working from someone's memory.
The stronger decision is to correlate the items against independent anchors, such as her entry in an earlier census taken while she was a child in her parents' household, or a baptismal entry, and to reason in writing about which informant was closest in time and involvement to the birth. The conclusion might favor 1876 or state a range, but it earns that standing through stated reasoning. This matters because a single unsupported pick collapses the moment a reviewer asks why, and genealogical conclusions are only as strong as the argument shown.
Scenario: Two Same-Name Men and a Stopped-Too-Early Search
Identity problems are resolved by correlating associates and multiple record types across the whole lifespan, not by accepting the first matching name, date, and town.
Suppose two men named Michael Doyle of similar age live in one Massachusetts town in the same decade, and you need to know which one married Bridget Finan. The plausible mistake is matching a single census entry by name and approximate age and moving on, which is not a reasonably exhaustive search even though a document was found. Same-name confusions are exactly where the first plausible hit is most dangerous, because a derivative index makes two different men look like one person.
The better decision is to build a mini-profile for each Michael: every census appearance, land and probate touches, church records, naturalization papers, and the associates each appears with, sometimes called tracing the group around a person. Then correlate: one man consistently appears with a brother named Thomas and lots near the mill; the other signs a petition with different neighbors. The marriage that co-occurs with the first cluster belongs to that Michael. This matters because identity conclusions drive everything downstream, and an identity error silently poisons every generation built on it.
Studying Record Groups by Question and Informant, Not by Trivia
Learn each record group by asking what question it answers, who the informants were, and what it cannot tell you, rather than memorizing definitions.
Organize your notes on census, vital, church, probate, land, military, immigration, and newspaper sources around three prompts per record type: what research question it can answer directly, who typically supplied each fact on it, and what the timing of its creation means for accuracy. A probate file, for example, can directly answer death timing and family relationships, has informants close to the events, and was created near the death, which is why it often outweighs a distant informant's later report.
Turn this into a small exercise: pick three record groups and write a one-sentence answer to each prompt for each, then test the sentences against one real document. You should observe that prompt answers written from memory feel confident but fail on the informant column, since who actually reported the age or the birthplace varies by record and era. Discovering that gap on your own notes is cheaper than discovering it mid-analysis, and the corrected sentences become reusable classification aids for the rest of your study.
An Adaptable Preparation Sequence with Readiness Checks
Sequence study in cycles: classify records, write mini-conclusions, run a conflict exercise, and re-check against a rubric each cycle.
A realistic cycle that adapts to any starting point runs in four passes. Pass one: classify ten documents from your own file by source, information, and evidence, on paper. Pass two: write three hundred-word research conclusions with layered citations, one of which resolves a known conflict. Pass three: build a same-name or missing-parent exercise from scratch and log every search as exhaustive or not, with reasons. Pass four: score the cycle against a rubric and repeat with harder records. Adjust the mix based on which rubric line scores lowest.
Use this rubric, scored from one to five per line, as learning milestones rather than pass predictions: classification lines are correctly labeled on all three layers without rereading definitions; citations pass the retrace test with a partner; each conclusion names the conflicting items and the reasoning that decides between them; and the search log justifies every unsearched category as unnecessary or unavailable. When you consistently score four or higher on every line across two cycles, you are ready to spend remaining time on unfamiliar record groups. Self-check scores measure skill growth only and say nothing about any specific assessment outcome.
One practical note: administrative details for the Boston University program, including current enrollment logistics, belong to the issuer, so confirm them on the Center for Professional Education site linked in the sources rather than relying on secondhand summaries.
References and further reading
Use these references to explore the concepts and check the latest information from the relevant organizations.