International Marketing Research and Intelligence

Introduction

International marketing decisions depend on evidence about customers, competitors, institutions, technologies, and cultures across countries. Yet international evidence is rarely directly comparable. A survey question may have different meanings in different languages; a market-size estimate may use inconsistent definitions; social-media conversations may overrepresent urban, younger, and highly connected consumers. International marketing research therefore involves more than collecting information. It requires designing a defensible process for defining problems, locating evidence, measuring concepts consistently, analysing variation across markets, and converting findings into responsible decisions.

International marketing intelligence is the systematic collection, interpretation, and dissemination of information about external market conditions and actors. Research is usually conducted for a defined decision or problem, whereas intelligence is an ongoing capability. For example, a company may conduct a cross-national study before launching a product, while its intelligence system continuously monitors competitors, regulation, prices, sentiment, and consumer trends.

This chapter explains how to design cross-national research using secondary and primary data, select samples, achieve measurement equivalence, combine qualitative and quantitative methods, evaluate data quality, and use dashboards, social listening, and artificial intelligence responsibly.

Learning Objectives

By the end of this chapter, students should be able to:

1. Distinguish international marketing research from marketing intelligence.
2. Formulate research problems, questions, and hypotheses for multiple countries.
3. Evaluate secondary data and design primary research.
4. Select appropriate sampling and data-collection methods.
5. Explain measurement equivalence and cross-cultural validity.
6. Compare qualitative, quantitative, and mixed-method approaches.
7. Assess reliability, validity, bias, and data quality.
8. Design useful dashboards and interpret social-listening data cautiously.
9. Evaluate appropriate and inappropriate uses of AI in insight generation.
10. Apply ethical, legal, and governance principles to international research.

1. The International Research Problem

A sound research project begins with a management decision, not with a preferred method. The decision problem expresses what a manager must decide, such as whether to enter a market, reposition a brand, alter a price, or localise communications. The research problem translates that decision into information requirements.

For example:

• Decision problem: Should a beverage company launch a low-sugar product in Country A and Country B?
• Research problem: Estimate category demand, identify consumer segments, measure perceived benefits and barriers, compare willingness to purchase, and assess differences in retail access.

A useful research brief specifies the population, markets, decision horizon, constructs, required precision, budget, and ethical constraints. Researchers should distinguish descriptive questions (What is happening?), explanatory questions (Why is it happening?), predictive questions (What is likely to happen?), and evaluative questions (Did an intervention work?).

1.1 A research process for international markets

A practical sequence is:

1. Define the decision and research problem.
2. Review theory and existing evidence.
3. Specify research questions and, where appropriate, hypotheses.
4. Determine the unit of analysis and target populations.
5. Choose a research design and country-comparison strategy.
6. Audit secondary data and establish a sampling plan.
7. Design instruments and test translations.
8. Collect, clean, document, and secure data.
9. Test measurement equivalence and analyse results.
10. Interpret findings in cultural and institutional context.
11. Communicate actionable implications, uncertainty, and limitations.
12. Monitor outcomes and update intelligence.

The order is iterative rather than strictly linear. A secondary-data audit may reveal that a research question must be narrowed; pilot interviews may reveal that a survey construct is poorly understood.

2. Secondary Data and Market Intelligence

Secondary data are information originally collected for another purpose. Sources include national statistical offices, customs records, regulatory filings, trade associations, syndicated research, company reports, academic studies, retailer data, app-store information, search trends, and publicly available digital content. Internal sources include sales, customer-service records, loyalty data, web analytics, and distribution reports.

Secondary data are valuable because they are faster and often less expensive than primary research. They can establish market size, identify trends, benchmark competitors, and help construct a sampling frame. However, availability does not imply suitability.

A secondary-data audit should examine:

A country’s reported e-commerce sales, for instance, may exclude informal sellers or use a different definition of online retail than another country. Analysts should not combine figures until definitions and units have been reconciled. Monetary data must be distinguished from real-volume data, and exchange-rate conversions must state whether nominal, market, or purchasing-power-adjusted rates are used.

2.1 Market potential and triangulation

Market potential estimates commonly combine population, incidence, purchase frequency, and average value. A simple illustrative estimate is:

Here, N is the relevant population, I is the proportion expected to buy during the period, F is purchase frequency, and V is average value per purchase. All variables must refer to the same period, population, product definition, and currency. This is an estimate, not an observed fact; it is sensitive to assumptions and should be tested with alternative scenarios.

Triangulation compares independent sources or methods. A researcher might compare government imports, retailer sell-through, consumer-reported purchasing, and manufacturer shipments. Convergence increases confidence, but disagreement is informative: it may indicate stockpiling, re-exports, informal trade, double counting, or different definitions.

3. Primary Research Designs

Primary research collects data for the current problem. The main designs are exploratory, descriptive, causal, and longitudinal.

Exploratory research

Exploratory research clarifies meanings, motivations, and possible variables. It includes interviews, focus groups, ethnography, observation, diary studies, and analysis of user-generated content. It is especially important when entering a culturally unfamiliar market. Its purpose is depth and discovery, not statistical estimation.

Descriptive research

Descriptive research measures incidence, attitudes, behaviours, awareness, satisfaction, or segment characteristics at a specified time. Cross-sectional surveys are common, but panels and repeated surveys are better for tracking change.

Causal research

Causal research tests whether a change in an independent variable produces a change in an outcome. Experiments may vary price, message, packaging, or promotion while controlling or randomising exposure. In international settings, researchers must determine whether the treatment is interpreted equivalently across countries. A message that is humorous in one culture may be confusing or offensive in another.

Mixed methods

Mixed methods combine qualitative and quantitative evidence. A sequential exploratory design may use interviews to develop culturally appropriate survey items, followed by a survey to estimate prevalence. A sequential explanatory design may begin with a survey and then use interviews to explain unexpected country differences. Integration should occur in the research design and interpretation, not merely by placing two unrelated studies in the same report.

4. Sampling Across Countries

The target population is the complete set of units to which the conclusions are intended to apply. The sampling frame is the operational list or procedure used to reach those units. The unit of analysis may be an individual, household, organisation, store, transaction, or country. Confusing these levels produces invalid conclusions.

Probability sampling gives units a known, non-zero chance of selection. Simple random, systematic, stratified, cluster, and multistage sampling are common. Non-probability methods include convenience, purposive, quota, snowball, and opt-in online panels. Probability samples support stronger population generalisation, but they can be expensive or impossible where frames are incomplete. Non-probability samples may be appropriate for exploratory work, provided limitations are explicit.

4.1 Sample size and precision

For a simple random sample estimating a population proportion, a commonly used large-population approximation is:

where n₀ is the required sample size before adjustments, z is the standard-normal critical value for the desired confidence level, p is the anticipated proportion, and e is the maximum tolerated margin of sampling error. The formula assumes a probability sample, independent observations, an appropriately specified proportion, and a large population relative to the sample. If no prior estimate of p exists, p = 0.5 gives the largest variance and therefore a conservative sample size.

For a finite population of size N, a finite-population correction can be applied:

These formulas do not correct coverage error, nonresponse bias, poor measurement, or quota-panel bias. A larger biased sample is not automatically better.

Worked example. Suppose a researcher wants a 95% confidence estimate of a proportion, assumes p = 0.5, and accepts e = 0.05. Using z = 1.96:

n₀ = (1.96² × 0.5 × 0.5) / 0.05² ≈ 384.16

The researcher would normally round up to 385 completed responses under the stated ideal assumptions. If the design uses clusters or a complex online panel, the effective precision may be lower and additional justification is required.

Cross-national samples must also address allocation. Equal samples facilitate comparisons, while proportional samples better represent a combined regional population. If country-level estimates are the priority, equal or minimum analytical samples may be appropriate. If a regional estimate is required, country results must be weighted according to the target population and design.

4.2 Sources of sampling error and bias

Important threats include undercoverage, nonresponse, self-selection, interviewer effects, panel conditioning, translation-related exclusion, and unequal internet access. Researchers should report response or completion rates where meaningful, recruitment procedures, screening rules, weighting, exclusions, and the intended inference population.

5. Measurement Equivalence and Cross-Cultural Validity

Measurement equivalence means that a construct is measured in sufficiently comparable ways across groups. Without it, observed country differences may reflect the instrument rather than real differences.

Four levels are often distinguished:

1. Configural equivalence: the same general factor or conceptual structure exists across groups.
2. Metric equivalence: factor loadings or scale-unit relationships are comparable, allowing relationships between constructs to be compared.
3. Scalar equivalence: item intercepts are comparable, supporting comparison of latent means.
4. Strict equivalence: residual variances are comparable; this is a stronger condition and is not always necessary.

These levels should be treated as analytical requirements, not as automatic properties of a translated questionnaire. Full equivalence may fail. Partial equivalence can sometimes support carefully limited comparisons if non-equivalent items are identified and the decision rule is transparent.

5.1 Translation and instrument development

A robust process may include:

• defining constructs in conceptual rather than literal terms;
• preparing a source questionnaire with simple, culturally neutral wording;
• independent forward translations;
• reconciliation by bilingual experts;
• back-translation as a diagnostic, not as proof of equivalence;
• cognitive interviews in each market;
• pilot testing response distributions and completion time;
• review by local subject-matter and ethics experts.

Researchers should consider differences in acquiescence, extreme responding, social desirability, familiarity with rating scales, and the meaning of categories such as “household,” “income,” or “regular purchase.” Anchoring vignettes, behavioural measures, and culturally appropriate examples may improve comparability, but they do not eliminate the need for validation.

For multi-item scales, internal consistency may be assessed with coefficient alpha:

Here, k is the number of items, σᵢ² is the variance of item i, and σₜ² is the variance of the total score. Alpha assumes, among other conditions, a common construct and a suitable covariance structure; a high value does not prove validity or unidimensionality. Researchers should examine item content, dimensionality, and construct validity rather than treating a threshold as a universal rule.

6. Qualitative and Quantitative Analysis

Qualitative analysis involves systematic coding, comparison, memo writing, and interpretation. A codebook should define concepts, inclusion and exclusion rules, examples, and procedures for resolving disagreement. Researchers should retain negative cases rather than selecting only quotations that support the preferred conclusion. Reflexivity is important: the researcher’s nationality, language, status, and assumptions may influence access and interpretation.

Quantitative analysis may include descriptive statistics, cross-tabulation, scale construction, regression, segmentation, conjoint analysis, or experimental comparisons. Country should not be treated merely as a label. It may proxy institutional, economic, linguistic, and cultural conditions; therefore, interpretations should avoid simplistic claims about national character.

A basic regression model can be written as:

where yᵢ is the outcome for observation i, xᵢ is a predictor, β₀ is the intercept, β₁ is the expected change in the outcome for a one-unit change in the predictor under the model, and εᵢ is the unexplained component. The interpretation depends on measurement scales, model specification, sampling, and assumptions such as independence and appropriate functional form. A statistically significant association is not necessarily causal or managerially important.

For country comparisons, analysts may include interaction terms to test whether a relationship differs across countries. However, country-level findings should not be inferred from individual-level associations, a problem known as ecological fallacy. Conversely, individual behaviour cannot always be inferred from country averages, known as atomistic fallacy.

7. Data Quality and Research Governance

Data quality is multidimensional. Accuracy concerns closeness to the relevant reality; completeness concerns missing values and coverage; consistency concerns agreement across records and sources; timeliness concerns usefulness at the decision date; validity concerns whether the measure represents the intended construct; and uniqueness concerns duplicate records.

A quality-control workflow should include range checks, logic checks, duplicate detection, missing-data analysis, duration and straight-line response checks, attention checks used cautiously, interviewer monitoring, audit trails, and reproducible syntax. Cleaning rules must be recorded before results are selectively reviewed. Researchers should distinguish legitimate extreme values from errors.

Validity includes:

• Content validity: coverage of the construct’s relevant dimensions.
• Construct validity: appropriate relationships with related and unrelated concepts.
• Criterion validity: association with a credible external criterion.
• Internal validity: credibility of causal conclusions within the study.
• External validity: transferability to the intended population and settings.

Ethical research requires informed consent, voluntary participation, minimisation of harm, confidentiality, secure storage, controlled access, and a clear retention and deletion policy. Sensitive international topics may create risks that differ by country, including legal exposure, employment consequences, political surveillance, or community stigma. Local law and institutional ethics requirements must be followed, but legal compliance is a minimum rather than a complete ethical standard.

Data protection principles commonly include purpose limitation, data minimisation, accuracy, storage limitation, security, and accountability. Cross-border transfers may require contractual, organisational, or regulatory safeguards. Researchers should avoid collecting personally identifiable information when aggregate data will answer the question.

8. Dashboards, Social Listening, and AI-Assisted Insights

A dashboard is a decision-support interface, not merely a collection of charts. It should connect each metric to a decision, define the population and time period, show targets or benchmarks, identify data freshness, and allow appropriate filtering. Useful components include market growth, share, distribution, conversion, customer satisfaction, complaint themes, competitor activity, and confidence or uncertainty indicators.

Good dashboard practice includes:

• one clear question per visualisation;
• consistent definitions across countries;
• visible denominators and units;
• separation of counts, rates, and percentages;
• annotation of campaigns, shocks, and data revisions;
• access controls and an audit trail;
• warnings when sample sizes or data quality are inadequate.

Social listening analyses public digital conversations, search behaviour, reviews, and other online signals. It can identify emerging issues, language used by consumers, competitor complaints, and crisis indicators. It does not measure the opinions of an entire population automatically. Bots, coordinated activity, duplicate posts, platform demographics, algorithmic ranking, sarcasm, code-switching, and cultural differences in posting behaviour can distort results. Sentiment classifiers may perform differently across languages and dialects. Social listening should therefore be triangulated with surveys, sales, interviews, or customer-service data.

AI-assisted insight generation can support transcription, translation drafts, coding suggestions, anomaly detection, summarisation, and hypothesis generation. It must not be treated as an autonomous researcher. Risks include fabricated citations or summaries, hidden training-data bias, privacy leakage, overconfident conclusions, automation bias, and inadequate handling of minority languages or sensitive content.

A responsible AI workflow is:

1. Define the permitted purpose and data boundaries.
2. Remove or protect unnecessary personal information.
3. Select a documented and approved tool.
4. Test outputs against human-coded or verified examples.
5. Check performance by country, language, demographic group, and content type.
6. Require human review for consequential interpretations.
7. Preserve prompts, versions, data sources, transformations, and decisions.
8. Report uncertainty and known failure modes.
9. Provide a route for correction, appeal, or withdrawal where people may be affected.

AI may propose themes; researchers remain responsible for evidence, interpretation, and decisions. Synthetic data can assist software testing or method development, but it should not be presented as evidence of actual consumer behaviour unless its limitations are explicitly established.

9. Applied Case: FreshBite’s Cross-National Launch

This is an illustrative case. FreshBite, a fictional European food company, is considering a ready-to-eat breakfast product in Country X and Country Y. Management believes demand will be similar because both countries have comparable GDP per capita and urbanisation.

Background and problem

Initial secondary research shows that category sales are growing in both countries, but the sources use different definitions: Country X reports retail scanner sales, while Country Y includes food-service purchases. Online reviews also suggest that consumers discuss “healthy” in different ways. FreshBite therefore needs to estimate demand and determine whether the same positioning can be used in both markets.

Analysis and approach

The research team first reconciles market definitions and separates retail from food-service data. It conducts 20 semi-structured interviews in each country, including regular buyers, non-buyers, retailers, and nutrition professionals. Cognitive interviews reveal that the translated term for “convenient” implies “fast but low quality” in Country Y. The team revises the questionnaire and pilots it with 80 respondents per country.

The main survey uses stratified quotas for age, gender, and region, with 500 completed responses per country. Because the sample is an online panel rather than a probability sample, the report avoids claiming a precise population margin of error. Weights are applied to align selected demographics with known population distributions, but the report states that weighting does not remove all coverage bias.

The team measures perceived nutrition, convenience, trust, purchase intention, prior behaviour, and price sensitivity. Multi-item scales are translated, cognitively tested, and assessed for factor structure. Rather than comparing raw mean scores immediately, the researchers first examine whether the constructs have comparable meaning. They also analyse open-ended responses and retailer interviews.

Findings and solution

The evidence indicates that Country X consumers associate health with low sugar and transparent nutrition labels. In Country Y, consumers place greater emphasis on freshness, family suitability, and locally recognisable ingredients. Purchase intention is similar after controlling for prior category use, but the barriers differ. FreshBite therefore develops a common brand promise—practical nutritious breakfast—while localising proof points, packaging examples, and retail partnerships.

A dashboard combines reconciled category sales, distribution, repeat purchase, complaint themes, and campaign measures. Social listening is used for early issue detection, not market-size estimation. An AI tool helps cluster anonymised review text, but bilingual analysts verify the clusters and review minority-language content.

Outcome and learning points

FreshBite launches a controlled pilot rather than a full rollout. It tracks trial, repeat purchase, stock availability, and qualitative feedback separately by country. The case demonstrates that apparently similar markets may differ in construct meanings, category definitions, retail systems, and evidence quality. It also shows that standardisation and adaptation are not alternatives in absolute terms: a company can standardise the research logic while adapting instruments, stimuli, and execution.

Challenges and Best Practices

International research faces practical constraints: inconsistent statistics, limited sampling frames, multilingual fieldwork, political sensitivity, low response rates, unequal digital access, currency volatility, and pressure to deliver rapid answers. Researchers should resist false precision and document what the data cannot support.

Best practices include:

1. Start with the decision and define the inference population.
2. Establish a country-comparison framework before collecting data.
3. Audit secondary-data definitions, sponsors, dates, and coverage.
4. Use local expertise without surrendering methodological consistency.
5. Pilot translations and test cultural interpretation.
6. Match sample design to the intended inference, not merely to budget.
7. Test measurement equivalence before comparing means or relationships.
8. Triangulate sources and investigate disagreement.
9. Display denominators, uncertainty, and data freshness in dashboards.
10. Treat social and AI-generated signals as evidence requiring validation.
11. Protect participants and minimise personal-data collection.
12. Make limitations visible in executive recommendations.

Chapter Summary

International marketing research transforms a managerial decision into a systematic inquiry across markets, while marketing intelligence provides an ongoing capability for monitoring the external environment. Effective research combines carefully audited secondary data with suitable primary designs. Sampling decisions must identify the target population, sampling frame, unit of analysis, allocation strategy, and likely sources of bias. Sample-size formulas describe precision under specific probability-sampling assumptions; they do not eliminate nonresponse or measurement error.

Cross-national comparison requires measurement equivalence. Translation, cognitive testing, pilot studies, and analysis of configural, metric, scalar, and—where necessary—strict equivalence help distinguish genuine market differences from instrument artefacts. Qualitative methods provide cultural depth and discovery; quantitative methods support estimation and structured comparison; mixed methods connect both forms of evidence.

Data quality includes accuracy, validity, completeness, consistency, timeliness, and coverage. Dashboards should support decisions through transparent definitions and uncertainty. Social listening can reveal emerging themes but is not automatically representative. AI can accelerate coding and synthesis, but human accountability, privacy protection, bias testing, documentation, and verification remain essential. The strongest international research is methodologically rigorous, culturally informed, ethically governed, and explicit about its limits.

Key Terms

International marketing research: Systematic research designed to support marketing decisions across countries or markets.

Marketing intelligence: Ongoing collection and interpretation of information about markets, competitors, customers, and the external environment.

Secondary data: Existing information collected originally for another purpose.

Primary data: Information collected specifically for the current research problem.

Target population: The complete set of units to which findings are intended to apply.

Sampling frame: The operational list or procedure used to access the target population.

Probability sampling: Sampling in which each unit has a known, non-zero selection probability.

Non-probability sampling: Sampling without known selection probabilities, such as quota or convenience sampling.

Measurement equivalence: Comparability of a construct and its measurement across groups or countries.

Configural equivalence: Evidence that the same broad construct structure exists across groups.

Metric equivalence: Comparable relationships between items and an underlying construct.

Scalar equivalence: Comparable item intercepts that support comparison of latent means.

Triangulation: Use of multiple sources, methods, or analysts to strengthen interpretation.

Reliability: Consistency or dependability of a measurement procedure.

Validity: Extent to which evidence supports the intended interpretation of a measure.

Coverage error: Bias caused when the sampling frame excludes part of the target population.

Social listening: Analysis of digital conversations and online signals to identify themes or changes.

Dashboard: A visual information interface designed to support monitoring and decisions.

AI-assisted insight generation: Use of artificial intelligence to support tasks such as coding, summarisation, or anomaly detection.

Ecological fallacy: Incorrectly inferring individual-level relationships from aggregate data.

Review Questions

Short Answer Questions

1. How does marketing intelligence differ from a single international marketing research project?
2. What are two limitations that should be examined when evaluating secondary data?
3. Distinguish a target population from a sampling frame.
4. What is measurement equivalence, and why is it important?
5. Why should social-listening results not automatically be treated as population estimates?

Long Answer Questions

1. Explain the stages of an international marketing research process and show how a management decision is translated into research questions.
2. Compare exploratory, descriptive, causal, and mixed-method research designs in an international launch decision.
3. Discuss probability and non-probability sampling, including the implications of coverage and nonresponse bias.
4. Explain configural, metric, scalar, and strict equivalence and describe how translation and cognitive testing support validity.
5. Discuss the ethical and governance requirements for using customer data, social listening, and AI-assisted analysis across borders.

Analytical / Application Questions

1. A firm finds that brand-trust scores are higher in Country A than Country B. Design an analysis plan that determines whether the difference reflects genuine attitudes or measurement non-equivalence.
2. A dashboard reports a sharp increase in negative online sentiment after a product launch. Identify at least five checks required before recommending a product withdrawal.
3. Design a mixed-method research programme for a company deciding whether to standardise or adapt a new product’s positioning in two culturally different markets. Justify the sample, methods, sequencing, and quality controls.

Multiple Choice Questions with Answers

Question 1

Which statement best distinguishes marketing intelligence from a single marketing research project?

A. Marketing intelligence is an ongoing system, whereas research is usually tied to a defined problem or decision.

B. Marketing intelligence uses only qualitative data, whereas research uses only quantitative data.

C. Marketing intelligence is always internal, whereas research is always purchased externally.

D. Marketing intelligence does not require evaluation of data quality.

Correct Answer: A. Marketing intelligence is an ongoing system, whereas research is usually tied to a defined problem or decision.

Question 2

In the sample-size formula n₀ = z²p(1 − p)/e², what does e represent?

A. The population size

B. The anticipated population proportion

C. The tolerated margin of sampling error

D. The number of questionnaire items

Correct Answer: C. The tolerated margin of sampling error

Question 3

Which level of measurement equivalence is generally required for comparing latent means across countries?

A. Configural equivalence only

B. Metric equivalence only

C. Scalar equivalence

D. Strict equivalence in every study

Correct Answer: C. Scalar equivalence

Question 4

What is the principal purpose of cognitive interviewing during questionnaire development?

A. To increase the sample size

B. To discover how respondents interpret questions and response options

C. To calculate the final market share

D. To replace all quantitative research

Correct Answer: B. To discover how respondents interpret questions and response options

Question 5

Which situation is an example of coverage error?

A. A respondent changes an answer after reading the question again.

B. A survey contains a duplicate record.

C. An online sampling frame excludes consumers without internet access.

D. A researcher reports results in the wrong currency.

Correct Answer: C. An online sampling frame excludes consumers without internet access.

Question 6

A researcher assumes p = 0.5 when estimating a proportion without prior information. Why is this commonly done?

A. It produces the smallest possible sample.

B. It gives the largest variance and therefore a conservative sample-size estimate.

C. It guarantees a representative sample.

D. It removes nonresponse bias.

Correct Answer: B. It gives the largest variance and therefore a conservative sample-size estimate.

Question 7

Which use of social listening is most defensible?

A. Treating the number of posts as the exact size of the target market

B. Using online sentiment alone to infer the opinions of all citizens

C. Identifying emerging complaints and triangulating them with other evidence

D. Ignoring platform demographics because all users have equal posting behaviour

Correct Answer: C. Identifying emerging complaints and triangulating them with other evidence

Question 8

Which statement about a high coefficient alpha is correct?

A. It proves that a scale measures the intended construct.

B. It proves that the scale is unidimensional.

C. It indicates internal consistency but does not by itself establish validity.

D. It demonstrates causal validity.

Correct Answer: C. It indicates internal consistency but does not by itself establish validity.

Question 9

Which practice best supports responsible AI-assisted insight generation?

A. Accepting generated summaries without checking source data

B. Using personal data in an unapproved tool to improve convenience

C. Testing outputs across languages and requiring human verification

D. Suppressing uncertainty to make recommendations more decisive

Correct Answer: C. Testing outputs across languages and requiring human verification

Question 10

Why can equal sample sizes across countries be useful?

A. They automatically make each sample probability-based.

B. They facilitate comparisons between countries, although they may not represent the combined regional population proportionally.

C. They eliminate translation problems.

D. They guarantee measurement equivalence.

Correct Answer: B. They facilitate comparisons between countries, although they may not represent the combined regional population proportionally.