ScientiaHub Original Studies
Research Atelier

Research design / An original study

A research analysis plan you can inspect

Palmer Penguins: a descriptive protocol, a substantive methods edit and a six-slide research presentation

A useful analysis plan makes its decisions visible before a result appears. Follow one decision here: retaining a record with missing sex changes the eligibility rule, the edited methods paragraph and the presentation of denominators. This folio shows the reasoning in full, with editable text and a source ledger you can examine.

Read the analysis planInspect the annotated edit

ScientiaHub Original Study — self-initiated demonstration.

This is an original, prospective methods demonstration. No dataset has been downloaded for analysis, synthesised or analysed. The before and after passages are original illustrations, not client manuscripts. Documentation facts are attributed to their sources. No empirical findings, publication outcomes or staff credentials are claimed.

01 / Research dossier

The decisions behind the summary

A descriptive protocol for recorded penguin body mass. Each decision links to its editorial consequence and its place in the methods presentation.

The question and the exact data source

The proposed question is deliberately narrow: how will we describe recorded body mass within each species in a specified Palmer Penguins archive? The intended product is a transparent account of the available measurements. It would allow a reader to see which records enter each calculation and which observations remain unavailable. This folio completes the planning and communication work for that question. Executing the analysis would be a separately agreed next step.

The maintainers distinguish the simplified penguins table from penguins_raw. Their documentation describes the simplified table as 344 rows and eight variables. These are documentation facts, not counts recomputed for this demonstration. The package site displays version 0.1.1, while its citation example names 0.1.0. Neither statement establishes the identity of a future analysis file. [S02] [S03]

Before execution, the analyst would record the selected release, retrieval location, retrieval date and a cryptographic checksum of the actual data file. The checksum would identify its bytes. The source register would keep the original research article, package documentation and analysis file as separate entries. This is a purposive selection of sources for a bounded protocol, not a systematic literature review. Query Q01 remains open until the analysis file exists.

A record is the working unit

Each imported row would be the working record. A source-row index would preserve its position in the frozen file and allow an analyst to trace a query back to the import. That index would serve as a tracking key. It would not establish a penguin's biological identity or prove that observations are statistically independent. Equal measurements would not, by themselves, justify removing a row.

The simplified dictionary contains no individual identifier. The original field methods describe sampling adult pairs at selected nests, with species occurring in different island sampling arrangements. These details warrant caution about independence and generalisation. They do not establish that a particular row in the simplified file is duplicated or that the same individual appears repeatedly. Those questions require a separate identity audit. [S01] [S02]

The proposed summaries therefore describe archived records. An identity concern would be logged with the affected row indices and investigated without overwriting the import. Any justified exclusion would need a reason and a reproducible rule. The analyst would not borrow exclusions from the paper's other analyses when the simplified file lacks the variables needed to apply them. Query Q02 identifies the additional evidence required if individual-level claims become part of the commission.

Units, validity and numerical precision

The outcome is body_mass_g, documented as an integer measurement in grams. The analysis would retain that unit throughout. The original study reports scale tolerances of ±25 g or ±50 g, depending on the instrument. An integer field therefore does not imply measurement accuracy to one gram. The dictionary and the field method answer different questions: what the file stores and how the observation was obtained. [S01] [S02]

An observed value must parse as a finite, positive measurement before it contributes to a mean. Nonnumeric entries, nonpositive masses and unexpected nonmissing codes would trigger a documented source query. Dependent summaries would remain on hold until the issue was resolved. The analyst would not silently relabel an invalid measurement as an ordinary missing value or trim a plausible extreme value merely because it looks unusual.

The reporting rule is a project decision: preserve full precision during calculations, then display mean mass to the nearest gram and missing percentages to one decimal place. Use round-half-up for these nonnegative quantities and integer counts throughout. This consistent display convention supports comparison across the report and slides. It makes no claim about additional instrument precision.

Eligibility follows the question

The primary table would assign records to the documented species categories. A record with missing species would enter a separately labelled unassigned audit count and would not enter a named-species estimate. Missing body mass would remain visible within the eligible species count. A missing bill measurement, flipper measurement, sex value or year would not disqualify an otherwise eligible record from this primary description.

The secondary tables would ask more specific questions. One would group each species by recorded sex. The other would group each species by stored year. Within the applicable table, a missing grouping value would receive an explicit 'not recorded' group. This is an audit label, not an additional biological category. Its inclusion would preserve the species totals while showing where a more detailed description lacks information.

Unexpected nonmissing species, sex or year codes would require resolution before dependent tables could be calculated. The protocol does not prescribe imputation. It also makes no assumption that absent values are random or harmless. A variable-specific eligibility rule preserves relevant measurements while making the limitation inspectable. This distinction is the main substantive change in ED-04: the edited passage replaces blanket removal of incomplete rows with explicit rules tied to each question.

Every reported value has a denominator

For a predeclared group, let N be the count of eligible records before excluding missing body mass. Let n be the count with an observed, valid mass. Once any invalid-value queries have been resolved, the missing count is N−n. For N greater than zero, the missing percentage is 100 × (N−n) / N. For n greater than zero, the mean is the sum of those n masses divided by n. Each table would display N and n beside the mean so that its coverage can be read directly.

The empty cases need their own labels. If N equals zero, the missing percentage is 'not applicable'. If n equals zero, the mean is 'not estimable'. Where N is positive and n is zero, all eligible records lack an observed mass, so the missing percentage is 100. A missing or undefined estimate would never appear as a measured zero. In a future machine-readable results file, an undefined number would use null with its reason.

Table shells would enumerate the documented categories and applicable audit groups, allowing zero-record cells to appear explicitly. Such a cell would describe the selected archive only. It would not demonstrate an absence in nature. The planned outputs contain no populated counts or means in this folio.

Pooled and stratified summaries answer different questions

The primary output would contain one summary for each named species. Two secondary tables would retain that species grouping while separately showing recorded sex and stored year. These are descriptive strata, meaning groups used to organise the records. The protocol does not commission a fully crossed species-by-sex-by-year model, an island adjustment or a causal comparison.

Each table would use the same definitions of N, n and missingness. Within a species, the secondary groups, including 'not recorded', must sum to the corresponding primary counts. The analyst would reconcile both eligible records and observed masses. This check makes an accidental filter or omitted grouping category discoverable. Missing-species records would remain in the separate audit count rather than being assigned to a named species.

A species mean would be calculated from all its eligible observed masses. If reconstructed from disjoint strata, it would use each stratum's observed count n as its weight and retain unrounded means internally. An unweighted average of subgroup means could give a small subgroup the same influence as a large one. This protocol therefore fixes the aggregation rule before any values are seen. Side-by-side tables would support inspection of composition without turning an observed difference into a population effect.

Year codes retain their documented meaning

The dictionary lists year as an integer study-year variable with codes 2007, 2008 and 2009. The original article describes austral summer field seasons of 2007/08, 2008/09 and 2009/10. This folio preserves both source descriptions without asserting an independently verified mapping between the simplified codes and those season labels. [S01] [S02]

The proposed secondary table would label its grouping variable 'Recorded study-year code'. It would treat the codes as categories, preserve their stored values and include a 'not recorded' group if needed. It would not average year codes, fit a temporal trend or relabel them as complete calendar-year observation periods. A row's code alone would not establish the exact date of measurement.

If the eventual manuscript required breeding-season labels, the analyst would first document a reliable mapping from the relevant source metadata or underlying records. Query Q03 records that dependency. The protocol, edited paragraph and presentation would then be revised together. Until that mapping is resolved, retaining the source codes keeps the display precise without filling a provenance gap through assumption.

Scope and the execution handover

This plan commissions descriptive summaries of a specified archive. It does not commission significance tests, confidence intervals, population estimates or ecological explanations. Differences between future table entries would remain descriptions of the available records. They would not establish a cause, a temporal change in a population or a successful replication of the original article.

The handover would separate the frozen input from derived calculations and maintain a working data dictionary. Broman and Woo recommend these organisational practices. Here they support traceability; they do not certify the statistical validity of an unexecuted analysis. The analyst would also preserve a query log, an exclusion log if needed and a record of any departure from this protocol. [S04]

Before results could be reported, the selected file and unresolved queries would need review, the calculations would need execution, and an independent check would need to reproduce the counts and aggregations. The deliverable demonstrated here is the completed specification and its communication across a methods passage and a talk. A researcher can inspect whether the decisions are appropriate, identify the evidence still needed and agree a precise next commission without mistaking a polished presentation for completed empirical work.

On a narrow screen, scroll within the table to read every column.

Working variable dictionary
FieldMeaningProtocol handling
speciesDocumented categories: Adélie, Chinstrap and Gentoo. [S02]Primary grouping. Missing values enter a separate unassigned audit count; unexpected nonmissing codes require resolution.
islandDocumented island categories: Biscoe, Dream and Torgersen. [S02]Retain as sampling context. No island-adjusted estimate is commissioned.
bill_length_mmBill length in millimetres. [S02]Outside the specified outcome. Its missingness does not remove a record from the body-mass summary.
bill_depth_mmBill depth in millimetres. [S02]Outside the specified outcome. Preserve the imported value without using it as an eligibility filter.
flipper_length_mmFlipper length in millimetres. [S02]Outside the specified outcome. Do not require it to be present for this protocol.
body_mass_gBody mass stored as integer grams. [S02]Outcome. Preserve missingness. Query nonfinite, nonnumeric or nonpositive observations before calculating dependent summaries.
sexDocumented recorded values: female and male. [S02]Secondary grouping only. Include a 'not recorded' audit group; query unexpected nonmissing values.
yearRecorded study-year codes: 2007, 2008 and 2009. [S02]Secondary categorical grouping. Retain codes and a 'not recorded' group; do not infer a season mapping.

Decision register

  1. DEC-01

    Dataset identity

    Freeze an identified analysis file before execution.

    A documentation page and a dataset release identify different objects. The file checksum and citation must refer to the actual input.

    [S02] [S03] See the editorial reasoning · See the slide

  2. DEC-02

    Record identity

    Use imported rows as working records without assuming individual independence.

    The simplified dictionary lacks an individual identifier. Equal measurements alone do not establish duplicate records.

    [S01] [S02] See the editorial reasoning · See the slide

  3. DEC-03

    Mass and precision

    Retain grams, resolve invalid values and round only for display.

    Stored units, field measurement tolerance and the chosen display precision have distinct meanings.

    [S01] [S02] See the editorial reasoning · See the slide

  4. DEC-04

    Variable-specific eligibility

    Retain missing body mass in N and do not delete a primary record because another measurement or secondary grouping value is missing.

    Eligibility follows the stated question. Missing grouping values remain visible in their applicable audit groups.

    [S02] See the editorial reasoning · See the slide

  5. DEC-05

    Denominators

    Report eligible N, observed n and missing N−n with explicit empty-case labels.

    Readers need to distinguish the archive covered by a group from the measurements contributing to its mean.

    See the editorial reasoning · See the slide

  6. DEC-06

    Stratification and pooling

    Use separate sex and year tables within species and pool through raw valid records or observed-count weights.

    Unweighted subgroup means can change the contribution of individual records. Reconciled totals reveal unintended filters.

    See the editorial reasoning · See the slide

  7. DEC-07

    Year and season

    Retain recorded year codes until a season mapping is verified.

    The dictionary's codes and the article's field-season descriptions must not be treated as an audited row-level mapping.

    [S01] [S02] See the editorial reasoning · See the slide

  8. DEC-08

    Interpretation limits

    Keep interpretation descriptive and reserve execution for a separately reviewed stage.

    A finished protocol does not establish computed findings, population effects or successful replication.

    [S01] [S04] See the editorial reasoning · See the slide

02 / Annotated edit

A methods paragraph, developed

Original illustrative writing. The initial passage deliberately contains weaknesses for this demonstration. The developed passage changes the underlying specification as well as the English.

Before

Illustrative initial draft. Methodological weaknesses are intentional.

We will use the Palmer Penguins data from the current package website, citing the paper and the package as the source of the analysis file. Every row will represent an independent penguin, and rows with identical measurements will be removed as duplicate observations. The analysis will compare body mass among species and describe differences across the other recorded characteristics. A single cleaned table will provide the basis for the report and presentation so that readers receive a consistent account.

Body mass is recorded in grams, so displaying the mean to the nearest gram will preserve the accuracy of the measurements. Records with any missing value will be excluded before analysis, giving all comparisons the same complete sample. The remaining measurements will be treated as suitable for calculation without a separate record of unusual entries. This approach will keep the methods brief and avoid presenting different eligibility rules for related tables.

We will calculate the percentage of missing masses by dividing the number of missing entries by the number of observed masses. Overall species means will be obtained by averaging the means for the recorded sex groups. The tables will show these means alongside a sample size, with empty numerical cells entered as zero for consistency. Comparisons will use the displayed rounded values, and the same values will appear in the conference slides.

The year values will be relabelled as the corresponding breeding seasons, and changes between these periods will describe temporal patterns. Differences in the summary means will establish which species are heavier in the population and whether those differences vary over time. The final discussion will connect the observed patterns to the ecological account in the original paper. A short methods slide will summarise the cleaning procedure before the audience sees the comparisons.

After

Developed proposed protocol. Outstanding questions remain visible.

Before execution, we will resolve query Q01 by identifying the analysis release and file checksum, while keeping the original article and package documentation as separate references. Imported rows will be working records without assuming that they represent unique individuals or statistically independent observations, and equal measurements will not trigger automatic deletion. The proposed analysis will describe archived body-mass records by species, with separate secondary groupings by recorded sex and year. No analysis has yet been executed.

Mass will remain in grams, with full calculation precision and means rounded half-up to the nearest gram for display, without implying one-gram measurement accuracy. The primary species summary will retain records with missing body mass in its eligible count and will not exclude records because unrelated fields are missing. Missing species will enter an unassigned audit count, while missing sex or year will enter its applicable 'not recorded' group. Unexpected nonmissing codes and invalid masses will require resolution before dependent summaries proceed.

For each group, N will count eligible records, n will count observed valid masses, and missing percentage will equal 100 × (N−n) / N when N exceeds zero. Species means will use valid raw records or equivalent observed-count weighting across disjoint strata, retaining full precision until display. For N=0, missing percentage will be 'not applicable', and for n=0, mean mass will be 'not estimable'. When N is positive and n is zero, missingness will be 100%, and secondary counts will reconcile to their primary species totals.

Stored year codes will remain categorical labels unless source evidence resolves the breeding-season mapping in query Q03. Interpretation will remain descriptive of the selected archive, without significance tests, confidence intervals, population effects or claims of temporal change. Counts will remain integers, and missing percentages will use one decimal place with half-up rounding only at display. The report and talk will retain the protocol's prospective status, unresolved identity query Q02 and explicit distinction between source facts and proposed calculations.

  1. ED-01

    Before B1.1
    After A1.1

    Replace a changing website reference with an identified input and separate bibliographic records.

    A reader needs to distinguish the analysed bytes from documentation and from the original article. The edit leaves file identity prospective rather than implying completed verification.

    Disposition: Protocol wording adopted. Q01 remains open before analysis.

    Related decision: DEC-01. [S02] [S03]

  2. ED-02

    Before B1.2
    After A1.2

    Remove the unsupported independence claim and automatic measurement-based deduplication.

    The simplified table does not supply an individual identifier. A tracking row number cannot establish biological identity, and equal measurements can be legitimate separate records.

    Disposition: Protocol wording adopted. Q02 remains open for any later individual-level analysis.

    Related decision: DEC-02. [S01] [S02]

  3. ED-03

    Before B2.1
    After A2.1

    Separate measurement units from display precision and state the rounding rule.

    Reporting an integer mean does not preserve or improve instrument accuracy. Calculations retain full precision, with a consistent project-specific display convention.

    Disposition: Adopted as a reporting decision. No instrument-level accuracy claim retained.

    Related decision: DEC-03. [S01] [S02]

  4. ED-04

    Before B2.2
    After A2.2

    Replace blanket complete-case deletion with eligibility for the stated outcome.

    A missing sex value need not discard an observed mass from a species summary. The following sentence in the edited passage separately accounts for missing species and missing secondary grouping values.

    Disposition: Adopted for this descriptive protocol. No claim that missingness is random or unbiased.

    Related decision: DEC-04. [S02]

  5. ED-05

    Before B3.1
    After A3.1

    Use eligible records as the missingness denominator and distinguish N from n.

    Dividing missing entries by observed entries answers a different question and can misrepresent the missing share. The accompanying sentences state the zero-denominator cases explicitly.

    Disposition: Adopted as the calculation specification. No numerical result computed.

    Related decision: DEC-05.

  6. ED-06

    Before B3.2
    After A3.2

    Replace an unweighted mean of subgroup means with equivalent record-weighted aggregation.

    Each observed mass must contribute once to its species mean. Full-precision aggregation and count reconciliation prevent display rounding or omitted audit groups from altering the calculation.

    Disposition: Adopted as an original aggregation rule, not attributed as a published finding.

    Related decision: DEC-06.

  7. ED-07

    Before B4.1
    After A4.1

    Retain source year codes and make season relabelling conditional on evidence.

    The dictionary and article describe time at different levels. Similar labels do not establish a verified mapping for each record, nor a population trend.

    Disposition: Conditional wording adopted. Q03 remains open.

    Related decision: DEC-07. [S01] [S02]

  8. ED-08

    Before B4.2
    After A4.2

    Replace population and temporal conclusions with the commissioned descriptive scope.

    The proposed tables concern an archived sample. Neither the planned calculations nor this unexecuted demonstration establish inferential or ecological conclusions.

    Disposition: Adopted. Any inferential extension would require a new question, design review and analysis plan.

    Related decision: DEC-08. [S01]

03 / Methods presentation

A six-slide research story

The same methodological choices, edited for a talk intended to take approximately eight minutes. All slide text and speaker notes remain editable. This is an HTML presentation, not a PowerPoint download or recorded talk.

ScientiaHub / Methods study01 / 06

A protocol for recorded body mass

How will we describe the available measurements within each species?

  • Primary description: archived records grouped by species
  • Secondary checks: recorded sex and study-year code
  • Prospective protocol. No results computed.
01 / Research question

Speaker notes

This talk presents the decisions that would govern a descriptive analysis of the selected archive. The question concerns the available body-mass records within each species. Separate tables would show the composition by recorded sex and year. We have not downloaded a dataset for analysis or calculated results. The contribution is a complete specification that a researcher can inspect before commissioning execution. DEC-01 identifies the input still to be fixed, while DEC-08 limits the interpretation. The following slides show how those commitments shape the calculations and the methods text.

Decision trail: DEC-01, DEC-08.

ScientiaHub / Methods study02 / 06

Source and record identity

The source register separates the article, the documentation and the future analysis file.

  • File identity: release, retrieval details and checksum
  • Working unit: imported record with a tracking index
  • Individual identity remains an unresolved question
02 / Provenance[S01] [S02] [S03]

Speaker notes

The maintainers provide a simplified table and a fuller raw table. The simplified dictionary contains no individual identifier. The original methods also describe sampling adult pairs at nests. These facts require careful wording about the working record and its possible relationships. [S01] [S02] [S03] Under DEC-01, the analyst would identify the actual input before calculation. Under DEC-02, equal measurements would not trigger automatic deletion. A source-row index would support traceability while leaving biological identity unresolved. The present commission stops at this protocol, so Q01 and Q02 remain explicit.

Decision trail: DEC-01, DEC-02. [S01] [S02] [S03]

ScientiaHub / Methods study03 / 06

Eligibility and counts

N counts eligible records. n counts observed valid masses.

N
All eligible records in the group, including missing body mass
n
Eligible records with a valid observed body mass

Missing body mass: N − n. Missing fields unrelated to the summary do not remove a record.

  • N=0: missing percentage not applicable
  • n=0: mean not estimable (100% missing when N>0)
03 / Calculation rules

Speaker notes

DEC-04 defines eligibility before missing mass is removed from a mean. A record with a named species and missing mass remains in N. A missing sex value does not remove an otherwise eligible record from the primary table. DEC-05 then separates coverage from the observed values available for calculation. When N is positive and n is zero, the missing share is 100 percent. An empty or undefined mean receives a text label, never a numerical zero. Invalid measurements require resolution before this missingness accounting can proceed.

Decision trail: DEC-04, DEC-05.

ScientiaHub / Methods study04 / 06

The planned summary tables

One species table, with separate recorded-sex and study-year tables.

Planned summaries only; no results calculated
ItemRule
Missing %100 × (N − n) / N, for N > 0
Mean massSum of observed mass / n, for n > 0
  • Secondary counts include 'not recorded' groups and reconcile to species totals
  • Aggregate at full precision and round only for display
04 / Reporting specification[S02]

Speaker notes

Each table uses the same denominator definitions. Under DEC-06, an overall species mean comes from all valid contributing records, or an exactly equivalent calculation weighted by observed subgroup counts. The analyst would reconcile both N and n across the secondary groups. DEC-03 retains full precision internally, with means displayed to one gram and missing percentages to one decimal place using half-up rounding. DEC-07 keeps the dictionary's year codes until the season mapping is resolved. [S02] These are reporting choices for an unexecuted analysis, so the table contains no invented findings.

Decision trail: DEC-03, DEC-06, DEC-07. [S02]

ScientiaHub / Methods study05 / 06

One edit changes the eligible sample

The revised sentence ties exclusion to the research question.

  • Illustrative before: exclude records with any missing value
  • Revised rule: missing unrelated fields do not remove a primary record
  • Consequence: missing sex stays visible in its secondary audit group
05 / Substantive editing

Speaker notes

ED-04 changes the method rather than polishing a phrase. The illustrative draft proposes deleting every incomplete row. The revised protocol keeps a record in the primary species description when an unrelated field is absent. It accounts for missing sex separately in the applicable secondary table. This choice does not establish that missingness is random, and it does not supply an unobserved mass. It preserves relevant measurements while exposing the denominator. The linked annotation, DEC-04 and the edited paragraph let a reader inspect the same decision at different levels of detail.

Decision trail: DEC-04, DEC-05.

ScientiaHub / Methods study06 / 06

Scope and next steps

Execution begins with the identified file and the unresolved source queries.

  • Proposed outputs describe the archive without population or causal claims
  • Next stage: resolve applicable queries, execute and independently reproduce the calculations
  • This folio supplies the protocol, edit and methods story
06 / Handover[S04]

Speaker notes

DEC-08 limits the proposed account to descriptive summaries of the selected records. A population estimate, temporal effect or ecological explanation would need a different commission and an appropriate design. The execution handover would preserve the input separately from calculations and maintain a dictionary, consistent with Broman and Woo's organisational guidance. [S04] A future analyst would resolve applicable source queries before running the plan. An independent check would then reproduce its counts and aggregations. No such result exists in this folio. The researcher can now assess the specification and agree the scope of further work.

Decision trail: DEC-08. [S04]

Terms used in this folio

Record
One imported row used as the working unit in this protocol.
Individual
A biological penguin whose identity requires source evidence beyond a tracking row index.
Available-case summary
A summary using the observed values needed for its particular calculation, with eligibility and missingness reported explicitly.
Denominator
The count by which another quantity is divided. Here N supports the missing percentage and n supports the mean.
Missing value
An absent observation. An invalid observed value needs investigation and is not silently treated as ordinary missingness.
Stratum
A declared subgroup used to organise the records, such as recorded sex within a species.
Source fact
A statement attributed to an inspected external source, distinguished from a project decision or an unexecuted calculation.
Proposed analysis
A specified future calculation. Its presence in a polished report or slide does not mean it has been performed.
Checksum
A value calculated from a file's bytes to support exact file identification. It does not establish the truth or quality of the data.
Source notes

A visible source foundation

These sources support the facts and context cited in the folio. The analysis decisions and illustrative writing are original contributions. The accompanying ledger distinguishes the two.

  1. S01

    Gorman, Williams and Fraser: Ecological Sexual Dimorphism and Environmental Variability within a Community of Antarctic Penguins

    PLOS ONE, 2014, 9(3), e90081. DOI: 10.1371/journal.pone.0090081. Accessed 2026-09-11.

    Materials and Methods: Field methods, Statistical methods and Data management.

    Original article cited through concise attributed paraphrase. No article figure or substantial passage is reproduced. Citation does not imply endorsement.

  2. S02

    Palmer Penguins: Size measurements for adult foraging penguins near Palmer Station

    Maintainers' penguins reference page. Site header displays package version 0.1.1; this does not identify a downloaded analysis file. Accessed 2026-09-11.

    Description, Format and Source, including all eight variable definitions.

    Variable definitions and format facts attributed to the maintainers. No dataset is redistributed or analysed in this demonstration.

  3. S03

    Horst, Hill and Gorman: palmerpenguins R data package

    Project page displays 0.1.1. Its Citation example names package version 0.1.0; these statements are retained separately. Accessed 2026-09-11.

    About the data, License, Citation and Additional data use information.

    Maintainers describe the data as CC0 and also provide scholarly-use guidance. This folio cites documentation without redistributing data or reusing the penguin artwork. No author affiliation or endorsement is implied.

  4. S04

    Broman and Woo: Data organization in spreadsheets

    Author-hosted manuscript as accessed. The web copy is cited without asserting a separately pinned document release. Accessed 2026-09-11.

    Be consistent; Create a data dictionary; No calculations in the raw data files.

    Concise attributed discussion of organisational recommendations. No table, figure or substantial manuscript passage is reproduced.

Keep the working files

Editable, inspectable source

The complete text, sentence locators, annotations and slide notes are included. The JSON ledger links source facts to their evidence and distinguishes proposed decisions.

The complete folio

Download Markdown source (38,327 bytes)

UTF-8 text. Edit paragraphs, table cells and speaker notes in a text editor. No Word tracked changes are claimed.

Claims and decisions

Download the JSON ledger (22,340 bytes)

Source locators, decision references, illustrative passages and unresolved queries. No results dataset is included.

Discuss a project

Discuss your project

Use this folio to identify the kind of work your project needs. We can discuss a research-design dossier, a substantive academic edit or a presentation built around your own evidence. The scope, deliverables, review arrangements and price are agreed after the brief is understood. The examples here show what to inspect when making that decision.

  • Your research question, discipline and current stage of work
  • The material available for review and any unresolved data or source questions
  • The intended institution, journal or conference, with its relevant requirements
  • The requested output format and language convention
  • Your deadline with time zone, priorities and preferred review arrangements
Discuss your projectScope and quotation arrangements

This folio and the website interface are in English. Commissioned service outputs are available in English, German, French, Italian, Russian, Chinese, Portuguese, Spanish and Japanese. Confirm the required language, regional convention and subject terminology in your brief. Project coordination is in English by default; the language list does not represent translated website pages or guaranteed support in every language.