gh-k-dense-ai-claude-scientific-skills-scientific-skills/skills/scientific-writing/references/writing_principles.md

Scientific Writing Principles

Overview

Effective scientific writing requires mastering fundamental principles that ensure clarity, precision, and impact. Unlike creative or narrative writing, scientific writing prioritizes accuracy, conciseness, and objectivity. This guide covers the core principles that distinguish good scientific writing from poor writing and provides practical strategies for improvement.

The Three Pillars of Scientific Writing

1. Clarity

Definition: Writing that is immediately understandable to the intended audience without ambiguity or confusion.

Why it matters: Science is complex enough without unclear writing adding confusion. Readers should focus on understanding the science, not deciphering the prose.

Strategies for Clarity

Use precise, unambiguous language:

Poor: "The drug seemed to help quite a few patients."
Better: "The drug reduced symptoms in 68% (32/47) of patients."

Define technical terms at first use:

"We measured brain-derived neurotrophic factor (BDNF), a protein involved in
neuronal survival and plasticity."

Maintain logical flow within and between paragraphs:

• Each paragraph should have one main idea
• Topic sentence introduces the paragraph's focus
• Supporting sentences develop that focus
• Transition sentences connect paragraphs

Use active voice when it improves clarity:

Passive (less clear): "The samples were analyzed by the researchers."
Active (clearer): "Researchers analyzed the samples."

However, passive voice is acceptable and often preferred in Methods when the action is more important than the actor:

"Blood samples were collected at baseline and after 6 weeks."

Break up long, complex sentences:

Poor: "The results of our study, which involved 200 participants recruited from
three hospitals and followed for 12 months with assessments every 4 weeks using
validated questionnaires, showed significant improvements in the intervention
group."

Better: "Our study involved 200 participants recruited from three hospitals.
Participants were followed for 12 months with assessments every 4 weeks using
validated questionnaires. The intervention group showed significant improvements."

Use specific verbs:

Weak: "The study looked at depression in adolescents."
Stronger: "The study examined factors contributing to depression in adolescents."

Common Clarity Problems

Ambiguous pronouns:

Poor: "Group A received the drug and Group B received placebo. They showed
improvement."
(Who is "they"?)

Better: "Group A received the drug and Group B received placebo. The drug-treated
group showed improvement."

Misplaced modifiers:

Poor: "We measured blood pressure in patients using an automated monitor."
(Are the patients using the monitor, or are we?)

Better: "Using an automated monitor, we measured blood pressure in patients."

Unclear referents:

Poor: "The increase in expression was accompanied by decreased proliferation, which
was unexpected."
(What was unexpected—the decrease, the accompaniment, or both?)

Better: "The increase in expression was accompanied by decreased proliferation.
This inverse relationship was unexpected."

2. Conciseness

Definition: Expressing ideas in the fewest words necessary without sacrificing clarity or completeness.

Why it matters: Concise writing respects readers' time. Every unnecessary word is a missed opportunity for clarity and impact. As the principle states: "We value concise writing because we value time."

Strategies for Conciseness

Eliminate redundant words and phrases:

Wordy	Concise
"due to the fact that"	"because"
"in order to"	"to"
"it is important to note that"	[delete]
"a total of 50 participants"	"50 participants"
"completely eliminate"	"eliminate"
"has been shown to be"	"is"
"in the event that"	"if"
"at the present time"	"now" or "currently"
"conduct an investigation into"	"investigate"
"give consideration to"	"consider"

Avoid throat-clearing phrases:

Wordy: "It is interesting to note that the results of our study demonstrate that..."
Concise: "Our results demonstrate that..." or "The results show that..."

Use strong verbs instead of noun+verb combinations:

Wordy	Concise
"make a decision"	"decide"
"perform an analysis"	"analyze"
"conduct a study"	"study" or "studied"
"make an assessment"	"assess"
"provide information about"	"inform"

Eliminate unnecessary intensifiers:

Wordy: "The results were very significant."
Concise: "The results were significant." (p-value conveys the degree)

Avoid repeating information unnecessarily:

Redundant: "The results showed that participants in the intervention group, who
received the treatment intervention, had better outcomes."
Concise: "The intervention group had better outcomes."

Favor shorter constructions:

Wordy: "In spite of the fact that the sample size was small..."
Concise: "Although the sample size was small..."

Acceptable Length vs. Unnecessary Length

Not all long sentences are bad:

This detailed sentence is fine: "We analyzed blood samples using liquid
chromatography-tandem mass spectrometry (LC-MS/MS) with a Waters Acquity UPLC
system coupled to a Xevo TQ-S mass spectrometer (Waters Corporation, Milford, MA)."

Why? Because each element is necessary information.

The key question: Can any word be removed without losing meaning or precision? If yes, remove it.

3. Accuracy

Definition: Precise, correct representation of data, methods, and interpretations.

Why it matters: Scientific credibility depends on accuracy. Inaccurate reporting undermines the entire scientific enterprise.

Strategies for Accuracy

Report exact values with appropriate precision:

Poor: "The mean was about 25."
Better: "The mean was 24.7 ± 3.2 (SD)."

Match precision to measurement capability:

Inappropriate: "Mean age was 45.237 years" (implies false precision)
Appropriate: "Mean age was 45.2 years"

Use consistent terminology throughout:

Inconsistent: Introduction calls it "cognitive function," Methods call it "mental
performance," Results call it "intellectual ability."

Consistent: Use "cognitive function" throughout, or define explicitly: "cognitive
function (also termed mental performance)"

Distinguish observations from interpretations:

Observation: "Mean blood pressure decreased from 145 to 132 mmHg (p=0.003)."
Interpretation: "This suggests the intervention effectively lowers blood pressure."

Be specific about uncertainty:

Vague: "There may be some error in these measurements."
Specific: "Measurements have a standard error of ±2.5 mmHg based on instrument
specifications."

Use correct statistical language:

Incorrect: "The correlation was highly significant (p=0.03)."
Correct: "The correlation was statistically significant (p=0.03)."
(p=0.03 is not "highly" significant; that's reserved for p<0.001)

Verify all numbers:

• Check that numbers in text match tables/figures
• Verify that n values sum correctly
• Confirm percentages are correctly calculated
• Double-check all statistics

Common Accuracy Problems

Overgeneralization:

Poor: "Exercise prevents depression."
Better: "In our sample, participants randomized to the exercise intervention showed
fewer depressive symptoms than controls (mean difference 3.2 points on the BDI-II,
95% CI: 1.5-4.9, p<0.001)."

Unwarranted causal claims:

Poor (from observational study): "Vitamin D supplementation reduces cancer risk."
Better: "Vitamin D levels were inversely associated with cancer incidence in this
cohort (HR=0.82, 95% CI: 0.71-0.95)."

Imprecise numerical descriptions:

Vague: "Many participants dropped out."
Precise: "15/50 (30%) participants withdrew before study completion."

Additional Key Principles

4. Objectivity

Definition: Presenting information impartially without bias, exaggeration, or unsupported opinion.

Strategies:

Present results without bias:

Biased: "As expected, our superior method performed better."
Objective: "Method A showed higher accuracy than Method B (87% vs. 76%, p=0.02)."

Acknowledge conflicting evidence:

"Our findings contrast with Smith et al. (2022), who reported no significant effect.
This discrepancy may result from differences in intervention intensity or population
characteristics."

Avoid emotional or evaluative language:

Subjective: "The results were disappointing and concerning."
Objective: "The intervention did not significantly reduce symptoms (p=0.42)."

Distinguish fact from speculation:

"The observed decrease in cell viability was accompanied by increased caspase-3
activity, suggesting that apoptosis may be the primary mechanism of cell death."
(Uses "suggesting" and "may be" to indicate interpretation)

5. Consistency

Maintain consistency throughout the manuscript:

Terminology:

• Use the same term for the same concept (not synonyms for variety)
• Define abbreviations at first use and use consistently thereafter
• Use standard nomenclature for genes, proteins, chemicals

Notation:

• Statistical notation (p-value format, CI presentation)
• Units of measurement
• Number formatting (decimal places)

Tense:

• Past tense for your specific study actions
• Present tense for established facts
• See detailed tense guide in IMRAD structure reference

Style:

• Follow journal guidelines consistently
• Citation format
• Heading capitalization
• Number vs. word for numerals

6. Logical Organization

Create a clear "red thread" through the manuscript:

Paragraph structure:

1. Topic sentence (main idea)
2. Supporting sentences (evidence, explanation)
3. Concluding/transition sentence (link to next idea)

Section flow:

• Each section builds logically on the previous
• Questions raised in Introduction are answered in Results
• Findings presented in Results are interpreted in Discussion

Signposting:

"First, we examined..."
"Next, we investigated..."
"Finally, we assessed..."

Parallelism:

Not parallel: "Aims were to (1) measure blood pressure, (2) assessment of
cognitive function, and (3) we wanted to evaluate mood."

Parallel: "Aims were to (1) measure blood pressure, (2) assess cognitive
function, and (3) evaluate mood."

Verb Tense in Scientific Writing

General Guidelines

Present tense for:

• Established facts and general truths
  • "DNA is composed of nucleotides."
• Conclusions you are drawing
  • "These findings suggest that..."
• Referring to figures and tables
  • "Figure 1 shows the distribution..."

Past tense for:

• Specific findings from completed research (yours and others')
  • "Smith et al. (2022) found that..."
  • "We observed a significant decrease..."
• Methods you performed
  • "Participants completed questionnaires at baseline."

Present perfect for:

• Recent developments with current relevance
  • "Recent studies have demonstrated..."
• Research area background
  • "Several approaches have been proposed..."

Section-Specific Tense

Section	Primary Tense	Examples
Abstract - Background	Present or present perfect	"Depression affects millions" / "Research has shown..."
Abstract - Methods	Past	"We recruited 100 participants"
Abstract - Results	Past	"The intervention reduced symptoms"
Abstract - Conclusions	Present	"These findings suggest..."
Introduction - Background	Present (facts), present perfect (research)	"Exercise is beneficial" / "Studies have shown..."
Introduction - Gap	Present or present perfect	"However, little is known..."
Introduction - This study	Past or present	"We investigated..." / "This study investigates..."
Methods	Past	"We collected samples..."
Results	Past	"Mean age was 45 years"
Discussion - Your findings	Past	"We found that..."
Discussion - Interpretation	Present	"This suggests..."
Discussion - Prior work	Past or present	"Smith found..." / "Previous work demonstrates..."

Common Writing Pitfalls

1. Jargon Overload

Problem: Excessive use of technical terms without definition

Example:

Poor: "We utilized qRT-PCR to quantify mRNA expression via SYBR-Green-based
fluorescence detection following cDNA synthesis from total RNA using oligo-dT primers."

Better: "We quantified mRNA expression using quantitative reverse transcription PCR
(qRT-PCR). Total RNA was reverse transcribed to complementary DNA (cDNA) using
oligo-dT primers, then amplified with SYBR Green fluorescent detection."

2. Nominalization

Problem: Turning verbs into nouns, making writing heavy and indirect

Examples:

Nominalized	Direct
"give consideration to"	"consider"
"make an assumption"	"assume"
"perform an investigation"	"investigate"
"conduct an examination"	"examine"
"achieve a reduction"	"reduce"

3. Hedging Excessively or Insufficiently

Excessive hedging (sounds uncertain):

"It could perhaps be possible that the intervention might possibly have some effect
on symptoms under certain conditions."

Insufficient hedging (overstates conclusions):

"The intervention cures depression."

Appropriate hedging:

"The intervention significantly reduced depressive symptoms in this sample,
suggesting it may be effective for treating mild to moderate depression."

Hedging words to use appropriately:

• Suggests, indicates, implies (not proves, demonstrates for correlational data)
• May, might, could (possibilities)
• Appears to, seems to (observations needing confirmation)
• Likely, probably, possibly (degrees of certainty)

4. Anthropomorphism

Problem: Attributing human characteristics to non-human entities

Examples:

Anthropomorphic	Scientific
"The study wanted to examine..."	"We aimed to examine..." or "The study examined..."
"The data suggest they want..."	"The data suggest that..."
"This paper will prove..."	"This paper demonstrates..."
"Table 1 tells us..."	"Table 1 shows..."

5. Abbreviation Abuse

Problems:

• Too many abbreviations burden the reader
• Abbreviating terms used only once or twice
• Not defining abbreviations at first use

Guidelines:

• Only abbreviate terms used ≥3-4 times
• Define at first use in abstract (if used in abstract)
• Define at first use in main text
• Don't abbreviate in title
• Limit to 3-4 new abbreviations per paper when possible
• Use standard abbreviations (DNA, RNA, HIV, etc.) without definition

Example:

Poor: "We measured Brain-Derived Neurotrophic Factor (BDNF) at baseline. BDNF
levels were elevated."
(Only used twice, abbreviation unnecessary)

Better: "We measured brain-derived neurotrophic factor at baseline. Levels were
elevated."

Specific Sentence-Level Issues

Dangling Modifiers

Problem:

"After incubating for 2 hours, we measured absorbance."
(The sentence suggests "we" were incubated)

Better: "After incubating samples for 2 hours, we measured absorbance."
Or: "After 2-hour incubation, we measured absorbance."

Misplaced Commas

Common errors:

Between subject and verb:

Wrong: "The participants in the intervention group, showed improvement."
Right: "The participants in the intervention group showed improvement."

In compound predicates:

Wrong: "We measured blood pressure, and recorded symptoms."
Right: "We measured blood pressure and recorded symptoms."
(No comma before "and" when it doesn't join independent clauses)

Pronoun Agreement

Wrong: "Each participant completed their questionnaire."
Right: "Each participant completed his or her questionnaire."
Or better: "Participants completed their questionnaires."

Subject-Verb Agreement

Wrong: "The group of participants were heterogeneous."
Right: "The group of participants was heterogeneous."
(Subject is "group" [singular], not "participants")

But: "The participants were heterogeneous." (Plural subject)

Word Choice

Commonly Confused Words in Scientific Writing

Often Misused	Correct Usage
affect / effect	Affect (verb): influence; Effect (noun): result; Effect (verb): bring about
among / between	Among: three or more; Between: two
continual / continuous	Continual: repeated; Continuous: uninterrupted
data is / data are	Data are (plural); datum is (singular)
fewer / less	Fewer: countable items; Less: continuous quantities
i.e. / e.g.	i.e. (that is): restatement; e.g. (for example): examples
imply / infer	Imply: suggest; Infer: deduce
parameter / variable	Parameter: population value; Variable: measured characteristic
principal / principle	Principal: main; Principle: rule or concept
significant	Reserve for statistical significance, not importance
that / which	That: restrictive clause; Which: nonrestrictive clause

Words to Avoid or Use Carefully

Avoid informal language:

• "a lot of" → "many" or "substantial"
• "got" → "obtained" or "became"
• "showed up" → "appeared" or "was evident"

Avoid vague quantifiers:

• "some" → specify how many
• "often" → specify frequency
• "recently" → specify timeframe

Avoid unnecessary modifiers:

• "very significant" → "significant" (p-value shows degree)
• "quite large" → "large" or specify size
• "rather interesting" → delete or explain why

Numbers and Units

When to Use Numerals vs. Words

Use numerals for:

• All numbers ≥10
• Numbers with units (5 mg, 3 mL)
• Statistical values (p=0.03, t=2.14)
• Ages, dates, times
• Scores and scales
• Percentages (15%)

Use words for:

• Numbers <10 when not connected to units (five participants)
• Numbers beginning a sentence (spell out or restructure)

Examples:

"Five participants withdrew" OR "There were 5 withdrawals"
(NOT: "5 participants withdrew")

"We tested 15 samples at 3 time points"
"Mean age was 45 years"

Units and Formatting

Guidelines:

• Space between number and unit (5 mg, not 5mg)
• No period after units (mg not mg.)
• Use SI units unless field convention differs
• Be consistent in decimal places
• Use commas for thousands in text (12,500 not 12500)

Ranges:

• Use en-dash (–) for ranges: 15–20 mg
• Include unit only after second number: 15–20 mg (not 15 mg–20 mg)

Paragraph Structure

Ideal Paragraph Length

Guidelines:

• 3-7 sentences typically
• One main idea per paragraph
• Too short (<2 sentences): may indicate idea needs development or combining
• Too long (>10 sentences): may need splitting

Paragraph Coherence

Techniques:

1. Topic sentence:

"Exercise training improves cardiovascular function through multiple mechanisms.
[Following sentences explain these mechanisms]"

2. Transitional phrases:

• First, second, third, finally
• Furthermore, moreover, in addition
• However, nevertheless, conversely
• Therefore, thus, consequently
• For example, specifically, particularly

3. Repetition of key terms:

"...this mechanism of action. This mechanism may explain..."
(Not: "...this mechanism. This process may explain...")

4. Parallel structure:

"Group A received the drug. Group B received placebo. Group C received no treatment."
(Not: "Group A received the drug. Placebo was given to Group B. No treatment was
provided to the third group.")

Revision Checklist

Content Level

• Does every sentence add value?
• Are claims supported by data?
• Is the logic clear and sound?
• Are interpretations warranted by results?

Paragraph Level

• Does each paragraph have one main idea?
• Are paragraphs in logical order?
• Are transitions smooth?
• Is there a clear "red thread"?

Sentence Level

• Are sentences clear and concise?
• Is sentence structure varied?
• Are there no dangling modifiers?
• Do subjects and verbs agree?

Word Level

• Is word choice precise?
• Are technical terms defined?
• Is terminology consistent?
• Are abbreviations necessary and defined?
• Are numbers formatted correctly?

Grammar and Mechanics

• Is verb tense correct and consistent?
• Are commas used correctly?
• Do pronouns agree with antecedents?
• Is punctuation correct?
• Is spelling correct (including technical terms)?

Tools for Improving Writing

Grammar and Style Checkers

• Grammarly: Grammar, style, clarity
• ProWritingAid: In-depth writing analysis
• Hemingway Editor: Readability, simplification
• LanguageTool: Open-source grammar checker

Caution: These tools don't understand scientific writing conventions. Use them as a starting point, not final arbiter.

Readability Metrics

Flesch Reading Ease:

• 60-70: acceptable for scientific papers
• <60: may be too complex

Caution: Don't sacrifice precision for readability scores designed for general audiences.

Peer Review

Most valuable tool:

• Ask colleagues to read and provide feedback
• Identify unclear passages
• Check logical flow
• Verify interpretations are warranted

Additional Resources

Books on Scientific Writing

• The Elements of Style by Strunk & White (classic on clear writing)
• On Writing Well by William Zinsser
• Scientific Writing: A Reader and Writer's Guide by Jean-Luc Lebrun
• How to Write a Scientific Paper by George M. Whitesides
• Style: Lessons in Clarity and Grace by Joseph Williams

Online Resources

• Academic Phrasebank (University of Manchester): Common academic phrases
• Purdue OWL: Grammar, punctuation, style
• Nature Masterclasses: Scientific writing courses
• WritingCenters: Many universities provide free online resources

University Writing Centers

Most research universities offer:

• Individual consultations
• Workshops on scientific writing
• Online resources and handouts
• Support for non-native English speakers

Venue-Specific Writing Styles

Four Major Writing Style Categories

1. Broad-audience accessible (Nature, Science, PNAS)
2. Clinical-professional (NEJM, Lancet, JAMA)
3. Technical-specialist (field-specific journals)
4. ML conference (NeurIPS, ICML, ICLR, CVPR)

Writing Style Comparison

Aspect	Nature/Science	Medical	Specialized	ML Conference
Sentence length	15-20 words	12-18 words	18-25 words	12-20 words
Vocabulary	Minimal jargon	Clinical terms	Field-specific	Technical + math
Tone	Engaging, significant	Conservative	Formal	Direct, contribution-focused
Key phrases	"Here we show"	"We conducted"	"To elucidate"	"We propose", "Our contributions"

ML Conference Style:

Characteristics:

• Direct, technical language with mathematical notation
• Contribution-focused (numbered lists common)
• Assumes ML expertise (CNNs, transformers, SGD, etc.)
• Emphasizes novelty and performance gains
• Pseudocode and equations expected

Example opening (NeurIPS style):

Vision transformers have achieved state-of-the-art performance on image classification,
but their quadratic complexity limits applicability to high-resolution images. We propose
Efficient-ViT, which reduces complexity to O(n log n) while maintaining accuracy. Our
contributions are: (1) a novel sparse attention mechanism, (2) theoretical analysis
showing preserved expressive power, and (3) empirical validation on ImageNet showing
15% speedup with comparable accuracy.

• Problem stated with technical context
• Solution previewed
• Numbered contributions
• Quantitative claims

Key Writing Differences

Aspect	Nature/Science	Medical	Specialized	ML Conference
Paragraph length	3-5 sentences	5-7 sentences	6-10 sentences	4-6 sentences
Math/equations	Minimize	Rare	Moderate	Essential
Active voice	Preferred	Mixed	Passive OK	Preferred
Hedging	Moderate	Conservative	Detailed	Minimal (claim gains)
Figure integration	Tight	Systematic	Detailed	Dense, in-page

Evaluation Focus by Venue

Venue	Key Evaluation Criteria
Nature/Science	Accessible to non-specialists? Broad significance clear? Compelling story?
Medical	Clinical relevance apparent? Professional tone? Methods adequate?
Specialized	Technical precision? Field expertise shown? Methods detailed?
ML conferences	Clear contributions? Claims supported by experiments? Reproducible?

Common rejection reasons:

• Poor writing quality/unclear prose
• Inappropriate style for venue
• Overstated claims
• Methods insufficient for reproduction
• Missing key details (baselines, ablations for ML; statistics for journals)

Quick Style Adaptation Guide

From → To	Key Changes
Journal → ML conference	Add numbered contributions; include equations/pseudocode; emphasize quantitative gains; condense prose
ML conference → Journal	Remove contribution numbering; expand motivation; separate Results/Discussion; reduce equations in main text
Specialist → Broad	Simplify language; emphasize broad implications; explain technical concepts; add context for non-experts
Broad → Specialist	Add technical detail; use field terminology freely; expand mechanistic discussion; cite field literature
Basic science → Clinical	Add patient/clinical context; use clinical language; emphasize outcomes/implications; cite clinical evidence

Pre-Submission Style Checklist

All venues:

• Writing style matches 3-5 recent papers from venue
• Sentence length appropriate
• Technical vocabulary level correct
• Tone consistent with venue
• No overstated claims

ML conferences add:

• Contributions clearly numbered in intro
• Mathematical notation correct and consistent
• Pseudocode/algorithms included where appropriate
• Claims quantified (e.g., "15% faster", "2.3% accuracy gain")
• Limitations acknowledged

Final Thoughts

Effective scientific writing is a skill developed through practice. Key principles:

1. Clarity trumps complexity
2. Conciseness respects readers' time
3. Accuracy builds credibility
4. Objectivity maintains scientific integrity
5. Consistency aids comprehension
6. Logical organization guides readers
7. Journal-specific adaptation maximizes publication success

Remember: The goal is not to impress readers with vocabulary or complexity, but to communicate your science clearly and precisely so readers can understand, evaluate, and build upon your work. Adapt your writing style to match your target journal's expectations and audience.