Content validity is the extent to which the content of a test, questionnaire, scale, rubric, or other instrument adequately represents the construct or subject area it is intended to assess. It concerns whether the included items are relevant, whether important content is missing, and whether the intended respondents can understand the instrument as designed.
Introduction
Researchers frequently create instruments to measure knowledge, attitudes, symptoms, competencies, behaviours, experiences, or other constructs. Before interpreting the resulting scores, they must determine whether the instrument contains the right content.
A workplace-communication scale, for example, may include items about speaking clearly but omit listening, written communication, non-verbal communication, and adapting messages to different audiences. Even if its items produce consistent responses, the instrument does not adequately represent the full domain of workplace communication.
This article explains:
- What content validity means.
- Why it is important.
- How to define a content domain.
- How experts and target users contribute to validation.
- How to calculate CVR, I-CVI, S-CVI/Ave, S-CVI/UA, and modified kappa.
- How to interpret and report the findings.
- What content validity does not prove.
- How digital tools and artificial intelligence can support the process.
Key takeaways
- Content validity asks whether an instrument adequately represents its intended content domain.
- Strong content evidence requires more than a high coefficient: the domain must first be defined clearly.
- Relevance, comprehensiveness, and comprehensibility are central considerations.
- CVR measures expert agreement about whether an item is essential; I-CVI usually measures agreement about relevance.
- Content-validity evidence is specific to an instrument’s intended population, purpose, language, and context.
- Content validation should be followed by pilot testing and other forms of psychometric evaluation.
What Is Content Validity?
Content validity refers to the adequacy with which an instrument’s content represents the construct, competency, behaviour, knowledge area, or outcome that the instrument is intended to measure.
An instrument has stronger content validity when:
- Its items are relevant to the intended construct and purpose.
- All important areas of the construct are represented.
- Irrelevant or construct-irrelevant content is minimized.
- Instructions, questions, response options, and scoring tasks are understandable to the intended users.
- The content is suitable for the target population and context of use.
Content validity is sometimes called logical validity or rational validity because it depends substantially on theoretical reasoning, domain analysis, and informed judgment.
Simple example
Suppose a university course teaches descriptive statistics, probability, sampling, confidence intervals, hypothesis testing, correlation, and regression.
A final examination containing only questions about means and standard deviations would have poor content validity for the whole course. The questions might be accurate, but the examination would underrepresent the curriculum.
A more content-representative examination would sample questions from all major learning outcomes in proportions justified by the course objectives and time devoted to each topic.
What Is a Content Domain?
A content domain is the complete set of relevant knowledge, behaviours, dimensions, symptoms, skills, or experiences that an instrument is intended to represent.
The domain should be defined before final items are selected. It may be based on:
- A theory or conceptual framework.
- Curriculum objectives.
- Professional standards.
- A job analysis.
- Clinical guidelines.
- Interviews or focus groups.
- Prior qualitative studies.
- Existing instruments and systematic reviews.
- Laws, regulations, or competency frameworks.
For a scale measuring online-learning engagement, the domain might include behavioural participation, emotional involvement, cognitive effort, interaction with instructors, interaction with peers, and persistence.
The researcher should then decide how many items or tasks will represent each part of that domain. This mapping is often called a content blueprint, test blueprint, or table of specifications.
Three Core Aspects of Content Validity
Contemporary measurement guidance commonly emphasizes three related questions.
1. Relevance
Relevance asks whether each item is appropriate for the construct, target population, and intended use.
An item may concern the general topic but still be unsuitable. For example, an item about climbing stairs might be relevant to physical functioning among ambulatory adults but inappropriate for an instrument designed specifically for people who use wheelchairs.
Relevance also applies to:
- Instructions.
- Response options.
- Recall periods.
- Examples.
- Administration format.
- Scoring rules.
2. Comprehensiveness
Comprehensiveness asks whether all important parts of the content domain are represented.
A panel can agree that every presented item is relevant while failing to notice that an entire dimension is absent. This is why a high I-CVI does not, by itself, prove comprehensive coverage.
Researchers should explicitly ask reviewers and target participants:
- What is missing?
- Which experiences or behaviours are not represented?
- Are any dimensions overrepresented?
- Are too many items focused on one narrow area?
3. Comprehensibility
Comprehensibility asks whether intended respondents understand the instrument as its developers intended.
It includes:
- Vocabulary.
- Sentence structure.
- Cultural meaning.
- Instructions.
- Response categories.
- Time frames.
- Examples.
- Layout and navigation.
Subject-matter experts may understand technical wording that ordinary respondents misunderstand. Therefore, comprehensibility should normally be investigated with members of the target population, often through cognitive interviews or think-aloud testing.
Content Validity as Evidence, Not a Permanent Label
It is common to say that a test “has content validity.” A more precise statement is that evidence supports the adequacy of the test content for a particular interpretation and use.
Validity is context-dependent. An instrument may be appropriate:
- For adults but not children.
- For clinical screening but not diagnosis.
- In one language but not a translation.
- For classroom feedback but not high-stakes selection.
- For one cultural group but not another.
- For measuring change but not comparing individuals.
A high content-validity coefficient does not make an instrument universally valid. Researchers should state the population, setting, language, administration method, and intended decisions for which the evidence was collected.
Why Is Content Validity Important?
Poor content representation can produce misleading conclusions even when calculations are performed correctly.
Content validity matters because it helps researchers:
- Align items with the intended construct.
- Detect omitted dimensions.
- Remove irrelevant content.
- Reduce construct underrepresentation.
- Reduce construct-irrelevant variance.
- Improve interpretability.
- Strengthen ethical and practical justification for using an assessment.
- Prepare a more defensible item pool for pilot testing.
- Explain instrument development to supervisors, reviewers, ethics committees, and readers.
Content validation is especially important for high-stakes assessments, clinical outcome measures, licensing tests, employee-selection procedures, educational examinations, and newly developed research scales.
Content Validity Compared With Related Concepts
| Concept | Main question | Typical evidence |
|---|---|---|
| Content validity | Does the content adequately represent the intended domain? | Domain mapping, expert review, target-user input, CVR, CVI |
| Face validity | Does the instrument appear appropriate at a surface level? | Informal review by respondents, colleagues, or stakeholders |
| Construct validity | Do scores behave as expected for the theoretical construct? | Hypothesis testing, convergent and discriminant evidence |
| Criterion-related validity | Are scores associated with a relevant external criterion? | Concurrent or predictive relationships |
| Structural validity | Does the internal dimensional structure match the proposed model? | Exploratory or confirmatory factor analysis |
| Reliability | Are scores sufficiently consistent or precise? | Alpha, omega, test–retest reliability, inter-rater reliability |
Content validity vs. face validity
Face validity is based on appearance: the instrument seems to measure what it claims to measure.
Content validity is more systematic. It requires a defined domain and an evaluation of whether the content adequately represents that domain.
A mathematics test filled with mathematical symbols may have face validity. It has content validity only if its questions represent the appropriate topics, skills, difficulty, and learning objectives.
Content validity vs. construct validity
Content validity examines the instrument’s content before or alongside field testing. Construct validity concerns the broader evidence supporting the interpretation of scores as measures of a theoretical construct.
Good content representation contributes to the overall validity argument but does not replace empirical construct-related evidence.
Content validity vs. reliability
Reliability concerns consistency or measurement precision. A test can produce highly consistent scores while repeatedly measuring only a narrow part of the intended domain.
Reliability is therefore necessary for many interpretations, but reliability alone cannot establish content validity.
When Should Content Validation Be Conducted?
Content validation is appropriate when researchers:
- Develop a new questionnaire, scale, checklist, rubric, or test.
- Create an examination from curriculum objectives.
- Adapt an existing measure for another population.
- Translate an instrument.
- Change response options, instructions, recall periods, or administration mode.
- Shorten a scale.
- Convert a paper instrument to a digital format.
- Use an existing measure for a substantially different purpose.
- Update an instrument after professional standards or clinical definitions change.
A published instrument should not automatically be assumed to remain content-valid after major modification or transfer to a new context.
How to Assess Content Validity: Ten Steps
Step 1: Define the construct and intended use
Write an explicit statement identifying:
- What is being measured.
- What is not being measured.
- The target population.
- The setting.
- The intended use of scores.
- The administration format.
- Relevant time frames.
- Any subdomains.
Ambiguous construct definitions produce ambiguous expert judgments.
Step 2: Develop the content-domain framework
Use theory, prior evidence, standards, qualitative research, and stakeholder input to identify the domain’s major and minor components.
Create a table showing:
| Domain | Definition | Indicators | Intended weight | Proposed items |
|---|---|---|---|---|
| Domain A | What it includes | Observable indicators | 30% | 6 |
| Domain B | What it includes | Observable indicators | 40% | 8 |
| Domain C | What it includes | Observable indicators | 30% | 6 |
The weights should reflect the intended interpretation and use, not merely the researcher’s convenience.
Step 3: Generate a broad item pool
Initial item generation should normally be inclusive. It is easier to remove redundant items later than to recover important concepts that were never represented.
Items may be generated from:
- Literature.
- Interviews.
- Focus groups.
- Existing instruments.
- Curricula.
- Job tasks.
- Clinical terminology.
- Open-ended surveys.
- Observational research.
Maintain a record connecting every candidate item to its proposed domain.
Step 4: Select qualified and diverse reviewers
Expertise should be defined in relation to the task. Relevant qualifications may include:
- Research or publication experience.
- Professional practice.
- Teaching or assessment experience.
- Knowledge of the target construct.
- Instrument-development expertise.
- Cultural or linguistic expertise.
- Experience with the target population.
A panel should include enough diversity to prevent one professional viewpoint from defining the entire construct. Researchers should justify the panel rather than relying on a universal number.
Peers or supervisors may provide useful feedback, but they should not be described as subject-matter experts unless their qualifications support that description.
Step 5: Prepare a structured rating form
Provide reviewers with:
- The construct definition.
- Domain definitions.
- Intended population and use.
- Administration instructions.
- The complete item pool.
- A rating scale.
- Space for comments.
- A request to identify missing content.
- Clear decision criteria.
A common relevance scale is:
- Not relevant
- Somewhat relevant
- Quite relevant
- Highly relevant
Researchers may assess clarity, simplicity, necessity, cultural appropriateness, and domain assignment separately. These dimensions should not be silently combined into a single score.
Step 6: Collect qualitative feedback
Numerical ratings show where disagreement exists but not why it exists.
Ask reviewers to explain:
- Why an item is irrelevant.
- Which wording is unclear.
- Whether two items are redundant.
- Whether an item belongs to another domain.
- Which content is missing.
- Whether the response options fit the question.
- Whether the item could disadvantage a subgroup.
For self-report measures, conduct cognitive interviews or pilot interviews with target respondents to investigate how they understand and answer each item.
Step 7: Calculate appropriate indices
Choose indices according to the rating question:
- Use CVR when experts classify whether an item is essential.
- Use I-CVI when experts rate item relevance.
- Use S-CVI/Ave or S-CVI/UA to summarize relevance across the instrument.
- Consider modified kappa when chance agreement is a concern.
The method should be selected before examining the results.
Step 8: Revise items using all evidence
Do not delete an item automatically because it falls below a numerical threshold.
First examine:
- Reviewer comments.
- Domain importance.
- Whether the item is the only indicator of a key concept.
- Whether disagreement reflects poor wording.
- Whether experts interpreted the construct differently.
- Whether the panel lacked relevant expertise.
- Whether the item is culturally or linguistically specific.
Possible decisions are:
- Retain without change.
- Retain with minor editing.
- Substantially revise.
- Move to another domain.
- Split into two items.
- Combine with another item.
- Remove.
- Replace.
- Add new content.
Step 9: Conduct another review and field test
Substantially revised or newly added items should be reviewed again. After content validation, pilot the instrument with an appropriate sample.
Later evaluation may include:
- Item analysis.
- Response distributions.
- Missing-data patterns.
- Cognitive interviewing.
- Internal consistency.
- Test–retest reliability.
- Inter-rater reliability.
- Factor analysis.
- Measurement invariance.
- Criterion relationships.
- Known-groups comparisons.
- Responsiveness to change.
Step 10: Document the process transparently
A reproducible report should describe the domain definition, panel, rating instructions, calculations, qualitative revisions, decision rules, and final item set.
Content Validity Ratio
The Content Validity Ratio, introduced by Lawshe, measures the extent to which experts regard an item as essential.
Experts classify each item as:
- Not necessary.
- Useful but not essential.
- Essential.
The formula is:
CVR = (nₑ − N/2) ÷ (N/2)
Where:
- nₑ = number of experts rating the item essential.
- N = total number of experts.
CVR ranges from −1 to +1.
- +1 means every expert rated the item essential.
- 0 means exactly half rated it essential.
- A negative value means fewer than half rated it essential.
CVR example
Suppose 9 of 10 experts rate an item as essential:
CVR = (9 − 10/2) ÷ (10/2)
CVR = (9 − 5) ÷ 5
CVR = 0.80
The result should be compared with a critical value appropriate for the panel size and chosen statistical procedure. Researchers should identify which critical-value table or recalculation they used because published tables do not always agree.
CVR evaluates necessity. It does not, by itself, show whether the instrument covers every important domain.
Item-Level Content Validity Index
The Item-Level Content Validity Index, or I-CVI, is usually calculated from relevance ratings.
Ratings of 3 or 4 on a four-point scale are coded as relevant. Ratings of 1 or 2 are coded as not relevant.
I-CVI = A ÷ N
Where:
- A = number of experts rating the item 3 or 4.
- N = total number of experts.
If 7 of 8 experts rate an item as relevant:
I-CVI = 7 ÷ 8 = 0.875
The value is normally reported as 0.88.
Scale-Level Content Validity Index
Two scale-level methods are common and should not be confused.
S-CVI/Ave
The average method calculates the mean of all I-CVI values:
S-CVI/Ave = Σ(I-CVI) ÷ number of items
This method indicates the average proportion of experts endorsing the items as relevant.
S-CVI/UA
The universal-agreement method calculates the proportion of items that every expert rated as relevant:
S-CVI/UA = number of items with universal agreement ÷ total number of items
S-CVI/UA is stricter and is usually lower than S-CVI/Ave.
A report stating only “the CVI was 0.91” is incomplete. Readers need to know whether the value is an I-CVI, S-CVI/Ave, S-CVI/UA, or another statistic.
Modified Kappa
I-CVI does not directly adjust for the possibility that experts agreed by chance. A modified kappa can be calculated as:
k = (I-CVI − Pᶜ) ÷ (1 − Pᶜ)*
The probability of chance agreement is:
Pᶜ = [N! ÷ A!(N − A)!] × 0.5ᴺ
Where:
- N = number of experts.
- A = number rating the item relevant.
- ! indicates a factorial.
Modified kappa is most useful when researchers want an item-level chance-adjusted indicator. It should supplement, not replace, the qualitative review and domain analysis.
Worked CVI Example
Eight experts rate six items for relevance.
| Item | Experts rating 3 or 4 | I-CVI | Approximate modified kappa | Initial decision |
|---|---|---|---|---|
| 1 | 8 | 1.00 | 1.00 | Retain |
| 2 | 7 | 0.875 | 0.87 | Retain |
| 3 | 6 | 0.75 | 0.72 | Review or revise |
| 4 | 8 | 1.00 | 1.00 | Retain |
| 5 | 7 | 0.875 | 0.87 | Retain |
| 6 | 5 | 0.625 | 0.52 | Substantial review |
Calculate S-CVI/Ave
S-CVI/Ave = (1.00 + 0.875 + 0.75 + 1.00 + 0.875 + 0.625) ÷ 6
S-CVI/Ave = 5.125 ÷ 6 = 0.854
Rounded to two decimal places, the S-CVI/Ave is 0.85.
Calculate S-CVI/UA
Only Items 1 and 4 received universal agreement:
S-CVI/UA = 2 ÷ 6 = 0.333
Rounded to two decimal places, the S-CVI/UA is 0.33.
Interpretation
The average relevance rating is reasonably high but below the commonly cited 0.90 benchmark for excellent scale-level content relevance. Universal agreement is much lower because it requires every reviewer to endorse an item.
The researcher should examine comments on Items 3 and 6, revise them if appropriate, check whether they represent important domains, and obtain another review rather than deleting them automatically.
What Is an Acceptable CVI?
A frequently cited item-level guideline is I-CVI ≥ 0.78. Because I-CVI values are discrete, its practical meaning depends on panel size.
For example:
- With 3 experts, unanimous agreement produces 1.00.
- With 4 experts, unanimous agreement produces 1.00, while 3 of 4 produces 0.75.
- With 5 experts, 4 of 5 produces 0.80.
- With 8 experts, 6 of 8 produces 0.75 and 7 of 8 produces 0.875.
A commonly cited scale-level target is S-CVI/Ave ≥ 0.90. These values are guidelines, not universal natural laws.
Interpretation should also consider:
- Expert qualifications.
- Panel size.
- Rating scale.
- Domain complexity.
- Intended stakes.
- Qualitative feedback.
- Whether respondents were involved.
- Whether items were revised and reassessed.
- Whether the statistic adjusts for chance.
- Whether the instrument is multidimensional.
How Many Experts Are Needed?
There is no single panel size that is correct for every content-validation study.
A smaller panel of highly qualified and diverse reviewers may be more informative than a larger panel with limited expertise. However, very small panels produce coarse indices: one reviewer’s rating can substantially change the result.
Panel planning should consider:
- Breadth of the construct.
- Number of professional perspectives needed.
- Cultural and linguistic diversity.
- Availability of genuine experts.
- Whether separate panels are needed for content and methodology.
- Planned statistical interpretation.
- Expected non-response.
- Conflicts of interest.
Researchers should state how expertise was defined, how reviewers were recruited, and why the final panel was adequate.
Content-Validation Examples
Educational assessment
A biology examination is mapped to curriculum outcomes. Teachers and curriculum specialists judge the importance, cognitive level, difficulty, and representativeness of each question. A test blueprint ensures that genetics, ecology, cell biology, evolution, and physiology receive appropriate coverage.
Health questionnaire
A fatigue questionnaire is developed from patient interviews and clinical literature. Patients identify relevant experiences and missing symptoms. Clinicians review clinical relevance, while target respondents complete cognitive interviews to assess wording, response options, and recall periods.
Employee-selection test
A job analysis identifies critical duties and competencies. Experienced job holders and supervisors judge whether test tasks represent important work activities. The test should reflect the actual job rather than convenient but irrelevant knowledge.
Research scale
A researcher develops a digital-learning self-efficacy scale. Literature and student interviews define the domains. Educational-technology specialists rate item relevance, students evaluate comprehensibility, and the revised instrument is later examined with factor analysis and reliability estimates.
Advantages of Content Validation
Content validation:
- Can be conducted before a large field study.
- Connects items to theory and intended use.
- Identifies irrelevant, ambiguous, and missing content.
- Integrates qualitative and quantitative evidence.
- Improves transparency in instrument development.
- Supports defensible revision decisions.
- Can involve multiple stakeholder perspectives.
- Helps prevent expensive field testing of a poorly defined measure.
Limitations
Content validation also has limitations:
- Expert judgments remain partly subjective.
- Results depend on how the domain is defined.
- Panels can share the same blind spots.
- Numerical agreement can hide missing content.
- Cutoffs may encourage mechanical decisions.
- Reviewers may interpret rating criteria differently.
- A high CVI does not establish reliability or structural validity.
- Content evidence may not transfer to another population or language.
- Experts may understand wording that target respondents do not.
- Poor documentation can make the process impossible to reproduce.
Common Mistakes
Treating a coefficient as complete proof
A high CVI supports expert agreement about the items presented. It does not prove that the content domain was complete or that scores will behave as expected in a field sample.
Confusing CVR with CVI
CVR usually concerns essentiality. I-CVI usually concerns relevance. They use different rating questions and should be named accurately.
Reporting only “overall CVI”
Researchers must identify whether they calculated S-CVI/Ave, S-CVI/UA, a mean CVR, or another coefficient.
Using unqualified reviewers
Convenient reviewers should not be presented as experts without evidence of relevant knowledge or experience.
Omitting target users
Experts cannot fully substitute for the people expected to understand and answer the instrument.
Automatically deleting low-scoring items
A low rating may indicate unclear wording, a misunderstood domain, or an underrepresented but important concept. Comments should be examined before a decision is made.
Using factor analysis as the only content-validity test
Factor analysis examines relationships among item responses. It cannot identify an important concept for which no item was written.
Reusing evidence after major adaptation
Translation, shortening, changing the response scale, or transferring an instrument to a substantially different population can alter its content relevance and comprehensibility.
Multidimensional Instruments
When an instrument has several subscales, content validity should be examined at both the item and domain levels.
A high overall S-CVI/Ave can conceal a weak subscale. Researchers should therefore report:
- I-CVI for each item.
- Domain-level averages where appropriate.
- The number of items per domain.
- Missing or overrepresented dimensions.
- Reviewer comments by domain.
- Overall scale-level indices.
Averaging results across unrelated dimensions without reporting domain-specific findings can create a misleading impression of adequacy.
Translated and Culturally Adapted Instruments
Literal translation is not sufficient. Adaptation should examine:
- Conceptual equivalence.
- Cultural relevance.
- Idioms and examples.
- Reading level.
- Response-category meaning.
- Recall periods.
- Sensitive or inappropriate wording.
- Differences in service, educational, or professional contexts.
Bilingual experts can assess semantic and conceptual equivalence, but members of the target population should also evaluate comprehensibility and relevance.
Digital Tools for Content Validation
Researchers can manage content validation with:
- Spreadsheets.
- Online survey platforms.
- Statistical software.
- Qualitative-analysis software.
- R or Python scripts.
- Version-control systems.
- Collaborative annotation tools.
A practical spreadsheet may contain:
| Item ID | Domain | Expert | Relevance | Essentiality | Clarity | Comments | Decision | Revision |
|---|
Automated formulas can calculate I-CVI, S-CVI/Ave, S-CVI/UA, CVR, and modified kappa. Researchers should still inspect the raw ratings and comments and retain an audit trail of every revision.
Can Artificial Intelligence Assess Content Validity?
Artificial intelligence can support content validation, but it should not be treated as an independent substitute for qualified human judgment or target-population research.
AI can help:
- Organize literature-derived domains.
- Detect duplicate or highly similar items.
- Flag complicated language.
- Identify possible double-barrelled questions.
- Suggest alternative wording.
- Classify comments.
- Check calculations.
- Compare item coverage against a supplied framework.
- Maintain a revision log.
AI cannot independently guarantee:
- That the theoretical framework is correct.
- That important lived experiences are represented.
- That culturally specific wording is appropriate.
- That respondents interpret an item as intended.
- That the selected evidence is unbiased.
- That high-stakes decisions are ethically justified.
A recent exploratory preprint compared several AI and human reviewers for a specific English-language test and reported similar aggregate rating patterns in that limited setting. Such preliminary findings do not establish that AI can replace experts across constructs, languages, populations, or high-stakes applications.
A defensible modern workflow uses AI as a review assistant while preserving human accountability, expert diversity, target-user participation, and transparent documentation.
How to Report Content Validity
A methods section should report:
- The construct and intended use.
- The target population and setting.
- How the content domain was developed.
- The item-generation process.
- Number and qualifications of reviewers.
- Reviewer-selection criteria.
- Rating scales and instructions.
- Whether reviewers worked independently.
- Qualitative feedback collected.
- Statistical indices and formulas.
- Critical-value or cutoff sources.
- Handling of missing ratings.
- Item-retention and revision rules.
- Target-population testing.
- Number of review rounds.
- Final changes to the instrument.
Example methods wording
The preliminary item pool was mapped to the predefined domains of the construct. Eight reviewers with relevant research, professional, and instrument-development experience independently rated each item’s relevance on a four-point scale. Ratings of 3 or 4 were classified as relevant. I-CVI, S-CVI/Ave, S-CVI/UA, and modified kappa were calculated. Reviewers also commented on clarity, redundancy, domain assignment, and missing content. Items below the prespecified item-level criterion were reviewed qualitatively and revised or removed according to their domain importance and reviewer feedback.
Example results wording
Item-level CVI values ranged from 0.63 to 1.00. The S-CVI/Ave was 0.85, whereas the stricter S-CVI/UA was 0.33. Two items received universal agreement. One item was removed because reviewers judged it irrelevant and redundant, while two items were revised because they represented important content but contained ambiguous wording. Three new items were generated to address a domain identified as underrepresented. The revised instrument underwent a second expert review and cognitive testing with members of the target population.
Content-Validity Reporting Checklist
Before submitting the study, verify that the report answers these questions:
- Is the construct defined?
- Is the intended use stated?
- Is the target population identified?
- Is the content-domain framework shown?
- Can each item be traced to a domain?
- Are expert qualifications described?
- Were target users involved where appropriate?
- Are rating scales reproduced?
- Are CVR and CVI named correctly?
- Is the S-CVI method identified?
- Are qualitative comments incorporated?
- Are low-scoring items explained?
- Are missing domains discussed?
- Are revision rounds documented?
- Are later psychometric analyses clearly separated from content validation?
Conclusion
Content validity concerns whether an instrument contains relevant, comprehensive, and understandable content for its intended construct, population, and use. It begins with a well-defined domain and is strengthened through systematic expert review, target-user input, transparent qualitative analysis, and appropriately labelled indices such as CVR, I-CVI, S-CVI/Ave, S-CVI/UA, and modified kappa.
No single coefficient proves that an instrument is valid. Strong content evidence provides a defensible foundation that must be combined with pilot testing, reliability analysis, response-process evidence, structural evaluation, and other evidence appropriate to the intended interpretation of scores.
