Literature Search and Critical Review

Introduction

A dissertation is not simply an extended essay based on sources. It is a reasoned contribution to knowledge, and its credibility depends substantially on how the existing evidence is located, evaluated and used. A literature review demonstrates what is already known, where scholars disagree, which theories and methods have been used, and why a further study is justified.

In Human Resource Management (HRM), literature is especially diverse. Relevant evidence may appear in peer-reviewed journals, academic books, practitioner reports, government publications, professional standards, labour-market statistics and organisational documents. These sources do not have equal authority or purpose. A systematic search therefore requires more than entering a topic into a search engine. It involves defining concepts, selecting appropriate databases, constructing transparent search strings, screening evidence, assessing quality and synthesising findings into an analytical argument.

A critical literature review is also distinct from an annotated bibliography. An annotated bibliography describes sources individually. A critical review compares and connects sources, evaluates their assumptions and evidence, and develops a conclusion about the state of knowledge.

Learning Objectives

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

1. Explain the purposes and characteristics of a critical literature review.
2. Translate a dissertation topic into searchable concepts and keywords.
3. Develop reproducible search strategies for scholarly and professional databases.
4. Apply inclusion, exclusion and screening criteria systematically.
5. Evaluate the credibility, relevance and methodological quality of sources.
6. Compare and synthesise competing findings without misrepresenting evidence.
7. Identify genuine research gaps and distinguish them from merely unexplored topics.
8. Organise a coherent literature review and use reference-management and AI tools responsibly.

1. The Role of the Literature Review in HRM Research

A literature review performs several connected functions:

• Contextualisation: It places the research problem within a field of knowledge and practice.
• Conceptual clarification: It defines central terms such as engagement, psychological safety, turnover intention or inclusive recruitment.
• Theoretical positioning: It identifies theories that explain relationships among variables or organisational processes.
• Evidence evaluation: It assesses how convincingly prior studies support their conclusions.
• Methodological learning: It reveals appropriate designs, measures, samples and analytical approaches.
• Gap identification: It shows what remains uncertain, contradictory, under-theorised or under-researched.
• Justification: It demonstrates why the proposed dissertation is worthwhile.

A strong review answers a logical sequence of questions:

1. What is the research problem?
2. How has it been conceptualised?
3. What theories explain it?
4. What evidence has been produced?
5. Why do findings converge or conflict?
6. What limitations affect existing knowledge?
7. How will the proposed study address a defensible gap?

The review should not become a catalogue of authors. Statements such as “Smith (2020) found…” repeated paragraph after paragraph produce a source summary, not critical scholarship. The unit of organisation should usually be an idea, debate, mechanism, population, method or finding rather than an individual publication.

2. Types of Literature and Review

2.1 Scholarly and professional literature

Scholarly literature is produced for academic research and commonly includes peer-reviewed journal articles, scholarly books, book chapters, conference papers and doctoral theses. Peer review provides quality control, although it does not guarantee that every article is correct.

Professional literature includes reports by organisations such as the CIPD, ILO, OECD, national statistical agencies and consulting firms. It can provide current labour-market evidence and practical insight, but its methods, sponsorship and definitions must be examined carefully.

Other useful sources include legislation, case law, collective agreements, organisational policies and official statistics. These may be authoritative for particular questions even when they are not peer reviewed.

2.2 Common review approaches

A bachelor’s dissertation will often use a structured narrative or integrative review, sometimes informed by systematic-review principles. The chosen approach should fit the question, available time and evidence base. The researcher must state the approach rather than implying a level of comprehensiveness that was not achieved.

3. From Research Topic to Searchable Question

Search quality begins with conceptual precision. A topic such as “employee wellbeing” is too broad to guide an efficient search. It should be narrowed by population, phenomenon, context, outcome and, where relevant, comparison or time period.

A useful planning framework is PICOC:

• Population: Who is being studied? For example, line managers, remote workers or NHS employees.
• Intervention or phenomenon: What practice or experience is examined? For example, flexible working.
• Comparison: Is there a comparison group or alternative practice?
• Outcome: What result is considered? For example, retention, stress or performance.
• Context: In what sector, country, organisational level or employment arrangement?

PICOC is not a rigid formula. Qualitative questions may use a framework such as SPIDER, which considers Sample, Phenomenon of Interest, Design, Evaluation and Research type. The purpose is to expose the main concepts before searching.

3.1 Building a concept matrix

Suppose the provisional question is: How does hybrid working influence employee engagement among knowledge workers in medium-sized UK organisations?

Synonyms are important because authors use different terminology. However, not every apparent synonym is conceptually identical. “Job satisfaction,” for example, is related to engagement but should not automatically replace it. The researcher must decide whether a term is a synonym, a broader category, a narrower category or merely a related construct.

3.2 Boolean search logic

Most academic databases support Boolean operators:

• OR broadens a search by retrieving synonyms.
• AND narrows a search by requiring multiple concepts.
• NOT excludes terms, but should be used cautiously because it may remove relevant studies.
• Quotation marks search an exact phrase.
• An asterisk often allows truncation, although rules differ by database. For example, engag* may retrieve engage, engagement and engaging.
• Parentheses control the order of operations.

A basic search string might be:

("hybrid work" OR "hybrid working" OR telework OR telecommut) AND ("employee engagement" OR "work engagement") AND (SME OR "medium-sized organisation" OR "medium-sized compan")

Search syntax varies. Scopus, Web of Science, EBSCOhost, ProQuest and Google Scholar do not interpret every command in the same way. The database name, date searched, fields used and exact search string should be recorded in a search log.

3.3 Searching in stages

A practical search normally proceeds through stages:

1. Orientation search: Learn the vocabulary, influential authors and major theories.
2. Focused search: Apply the concept matrix and eligibility criteria.
3. Citation searching: Examine references cited by useful articles (backward searching) and later publications that cite them (forward searching).
4. Update search: Repeat the search shortly before submission to identify recent evidence.

Citation chaining is particularly valuable when terminology is inconsistent. Nevertheless, it can reproduce the biases of an initial article, so it should supplement—not replace—database searching.

4. Selecting Databases and Managing Search Evidence

No single database indexes all HRM knowledge. A reasonable search may combine:

• Business Source Complete, ABI/INFORM or ProQuest: management and HRM journals.
• Scopus or Web of Science: multidisciplinary citation indexing and citation tracking.
• PsycINFO: work psychology, wellbeing, motivation and behaviour.
• Sociological Abstracts: work, employment and inequality research.
• EconLit: labour economics and productivity.
• Google Scholar: broad discovery, including theses and reports, but with less transparent coverage and ranking.
• CIPD, ILO, OECD and government repositories: professional and official evidence.

Database coverage differs by language, discipline, publication type and date. A search that produces many results is not necessarily good; it may simply be poorly focused. Conversely, a small result set may reflect overly restrictive terms.

A search log should include:

Search logs support transparency, reduce duplicated effort and allow the researcher to explain how sources were found. A screening spreadsheet can record title, author, year, country, sample, design, key concepts, findings, limitations and decision status.

5. Screening and Eligibility Decisions

Searching produces candidate records, not automatically eligible evidence. Screening normally occurs in stages:

1. Remove duplicate records.
2. Screen titles and abstracts.
3. Retrieve and assess full texts.
4. Apply final inclusion and exclusion criteria.
5. Search references and citations of included studies.

Criteria should be linked to the research question. For example:

Possible inclusion criteria:

• published between 2015 and 2025;
• addresses hybrid working and employee engagement;
• concerns employees rather than consumers;
• presents empirical data or a clearly argued theoretical analysis;
• relates to organisational or employment contexts.

Possible exclusion criteria:

• news articles without disclosed evidence;
• studies focused only on student learning;
• duplicate versions of the same study;
• sources where the full text cannot be assessed;
• articles using “engagement” solely to mean customer engagement.

Date and language limits require justification. Excluding non-English research may introduce language bias. Excluding older sources may remove foundational theory. A good review distinguishes seminal sources from current evidence rather than relying on publication date alone.

A PRISMA-style flow record can show the number of records identified, duplicates removed, abstracts screened, full texts assessed and sources included. The PRISMA framework was developed for systematic reviews, but its emphasis on transparent reporting can improve many dissertation reviews. Researchers should not claim to have conducted a full systematic review unless the required protocol and reporting standards were followed.

6. Evaluating Source Credibility and Quality

Credibility is multidimensional. A source may be authoritative but irrelevant, recent but methodologically weak, or highly cited but based on an unsuitable population.

6.1 Questions for evaluating a source

Authority and provenance

• Who conducted or published the work?
• Is the journal reputable and appropriate to the field?
• Are the authors’ affiliations and possible conflicts disclosed?
• Is the source peer reviewed, official or professionally accountable?

Relevance

• Does the population resemble the proposed research context?
• Are the concepts defined in a way relevant to the dissertation?
• Is the setting transferable to the researcher’s context?

Methodological quality

• Is the research question clear?
• Is the design suitable for the question?
• Is the sample adequately described and justified?
• Are measures valid and reliable where quantitative methods are used?
• Are qualitative data-collection and analysis procedures sufficiently transparent?
• Are conclusions supported by the evidence?

Currency and contribution

• Does the publication reflect current technology, law or employment practice?
• Is it a foundational source that remains conceptually important?
• Does it add evidence, theory, method or context?

Integrity and accessibility

• Is the full text available for verification?
• Are data, instruments or analytic procedures reported sufficiently?
• Is the source free from fabricated, plagiarised or misleading content?

Citation count is a discovery aid, not a direct measure of quality. A highly cited article may be influential because it is controversial, old or widely criticised. Journal impact metrics similarly describe journals at an aggregate level and should not be used as a substitute for reading and evaluating an individual article.

6.2 Appraising different designs

The criteria must fit the design. In a cross-sectional survey, examine sampling, response rate, measurement validity and whether causal claims exceed the design. In an interview study, examine recruitment, reflexivity, data sufficiency, coding and interpretation. In an intervention study, examine allocation, comparison conditions, attrition and the plausibility of causal inference. A checklist should support judgement, not replace it.

A useful distinction is between internal validity, external validity, construct validity and credibility or trustworthiness. Internal validity concerns whether an observed relationship is plausibly caused by the proposed explanation. External validity concerns transferability to other populations or settings. Construct validity concerns whether a measure represents the intended concept. In qualitative research, credibility, dependability, confirmability and transferability are often more appropriate evaluative ideas.

7. Reading and Extracting Evidence

Efficient reading is purposeful rather than linear. Begin with the abstract, introduction, methods, results and discussion, then return to sections requiring close attention. Record claims in your own words and separate them from direct quotations.

An evidence matrix may contain:

The matrix is a thinking tool, not a substitute for prose. It helps reveal patterns, missing populations, inconsistent measures and possible explanations for disagreement.

8. Synthesising Competing Findings

Synthesis means developing a higher-level interpretation from multiple sources. It does not mean averaging opinions or selecting the result that supports the researcher’s preferred conclusion.

A useful synthesis process is:

1. Group studies by concept, mechanism, population, method or finding.
2. Compare definitions and measures before comparing results.
3. Identify convergence, divergence and qualified findings.
4. Examine whether differences arise from context, sample, time, design or theory.
5. Develop a reasoned conclusion that represents uncertainty accurately.

For example, research on remote working may report improved autonomy and reduced commuting alongside isolation, visibility concerns and longer working hours. These findings are not necessarily contradictory. Effects may depend on job interdependence, managerial practices, home circumstances, organisational culture and the degree of employee choice.

Theoretical frameworks help organise such complexity. The Job Demands–Resources model distinguishes job demands, which require sustained effort, from job resources, which support motivation and goal achievement. Hybrid work may increase resources such as autonomy while also increasing demands such as coordination complexity and work–home boundary management. The framework does not predict one universal outcome; it suggests mechanisms that should be examined in context.

A review should also distinguish:

• Association from causation: A correlation between flexibility and engagement does not prove that flexibility caused engagement.
• Statistical significance from practical importance: A small effect may be statistically detectable but practically limited.
• Average effects from subgroup effects: A generally positive result may conceal disadvantages for new employees, disabled employees or carers.
• Evidence absence from absence of evidence: Failure to find a relationship may reflect weak measurement or insufficient statistical power rather than proof that no relationship exists.

Useful synthesis language includes “taken together,” “by contrast,” “this difference may reflect,” “the evidence is strongest when,” and “however, interpretation is limited by.” Such language signals relationships among studies rather than a sequence of summaries.

9. Identifying and Framing Research Gaps

A research gap is a justified limitation in existing knowledge, not simply a topic that the researcher has not yet studied. Common types include:

• Empirical gap: insufficient evidence for a population, industry or country.
• Conceptual gap: unclear or inconsistent definitions.
• Theoretical gap: an established theory has not been applied or competing explanations remain unresolved.
• Methodological gap: overreliance on one method, measure or research design.
• Contextual gap: findings have not been examined under changed legal, technological or organisational conditions.
• Contradictory evidence: credible studies reach different conclusions.
• Practical gap: research has not adequately addressed an important organisational decision.

A weak gap statement says, “Few studies have examined this topic.” A stronger statement identifies the evidence base and explains the consequence: “Existing studies predominantly use cross-sectional surveys of large organisations, leaving uncertainty about how employees in medium-sized firms experience hybrid work over time.” The proposed dissertation should then show how its population, design or conceptual approach addresses that limitation.

A gap must be feasible. A bachelor’s dissertation cannot solve every limitation in a field. It may make a modest contribution by examining a particular organisation, comparing employee groups, applying an established framework in a new context or providing an in-depth account of an under-researched experience.

10. Constructing the Written Critical Review

A coherent literature review usually moves from broad context to focused argument:

1. Define the research problem and scope.
2. Explain central concepts and theoretical perspectives.
3. Present major themes or debates.
4. Compare evidence and methods within those themes.
5. Discuss contradictions and contextual conditions.
6. Identify the gap.
7. Link the gap to the research question and proposed method.

A paragraph commonly contains:

• a topic sentence stating the analytical point;
• evidence from several sources;
• comparison or evaluation;
• an interpretation of what the evidence means;
• a link to the next point or to the research question.

For example:

Flexible working is often associated with greater autonomy, but its effects on engagement appear conditional rather than universal. Survey studies commonly report positive relationships, yet interview research indicates that autonomy can coexist with intensified monitoring and blurred working hours. These differences may reflect variation in managerial trust, job interdependence and employees’ domestic circumstances. Consequently, the present study treats hybrid working as a set of practices rather than a single uniform intervention.

This paragraph synthesises, evaluates and positions the study. It does not merely list sources.

Referencing and academic integrity

Every substantive claim derived from another source should be cited. Paraphrasing requires genuine reformulation and accurate representation, not merely replacing a few words. Direct quotations should be limited and accompanied by page numbers where required by the referencing style. Researchers must avoid plagiarism, fabricated citations, citation laundering and selective reporting.

Reference-management software such as Zotero, EndNote or Mendeley can store records, insert citations and generate bibliographies. It cannot determine whether a source actually supports a claim. Every automatically generated reference must be checked for author names, dates, titles, journal information, page ranges and DOI details.

11. Technology and Generative Artificial Intelligence

Search platforms increasingly use recommendation systems, semantic search and automated summaries. These tools can help discover terminology and related publications, but ranking algorithms may privilege highly cited, recent or English-language material. Researchers should search more than one platform and retain independent judgement.

Generative AI may assist with brainstorming keywords, organising notes or identifying possible themes. It can also invent references, misstate findings, reproduce bias and obscure the origin of text. Responsible use requires:

• checking every suggested source in the original publication;
• never citing an unverified reference;
• preserving the researcher’s own interpretation and authorship;
• disclosing use where institutional policy requires it;
• not uploading confidential interview data or copyrighted full texts without permission;
• checking outputs for bias, privacy risks and inaccurate summaries.

AI does not remove the researcher’s duty to read the evidence. The final literature review must be traceable to authentic sources and the researcher’s reasoned judgement.

12. Applied Case: Developing a Review on Hybrid Working and Engagement

The following is an illustrative case.

Background

A medium-sized professional-services firm has introduced a three-day office and two-day remote-working arrangement. Senior managers believe the policy improves engagement, while some employees report isolation, unequal visibility and longer working hours. A student proposes the question: How do employees experience the relationship between hybrid working practices and engagement?

Problem

An initial Google search returns thousands of results. Some concern customer engagement, some focus on university students and others treat remote work and hybrid work as identical. The student is at risk of producing an unmanageable and conceptually confused review.

Analysis and approach

The student first defines engagement as a work-related psychological state and distinguishes it from job satisfaction and organisational commitment. The concept matrix includes hybrid work, flexible work, telework, work engagement and employee engagement. Searches are conducted in Business Source Complete, Scopus, PsycINFO and Google Scholar, followed by searches of CIPD and official labour-market reports.

The student records exact strings and applies criteria: English-language studies from 2015 onward, organisational employees, empirical or clearly theoretical work, and direct relevance to hybrid or remote working. Titles and abstracts are screened, duplicates removed and full texts assessed. Citation chaining identifies two foundational articles missed by the keyword search.

The evidence matrix shows three patterns. First, surveys often associate autonomy and flexibility with engagement. Second, qualitative studies identify boundary strain, social disconnection and concerns about fairness. Third, results vary by occupation, managerial practice and employee choice. The student therefore adopts the Job Demands–Resources model as a sensitising framework rather than a claim that all hybrid work has the same effect.

Outcome

The final review argues that the literature supports a conditional relationship: hybrid work may increase engagement when autonomy, communication and managerial support are adequate, but may reduce engagement when demands, isolation or surveillance intensify. The gap is not simply that “more research is needed.” It is that medium-sized professional-services organisations and employee experiences of competing resources and demands are underrepresented. The proposed qualitative interviews are justified as a way to explore mechanisms that cross-sectional surveys cannot adequately explain.

Learning points

• Define constructs before searching.
• Use multiple databases and synonyms.
• Treat professional reports as useful but methodologically distinct from journal articles.
• Explain disagreement instead of hiding it.
• Link the identified gap directly to the research design.

13. Challenges, Limitations and Best Practice

Literature searching is affected by database coverage, paywalls, language restrictions, publication bias and inconsistent terminology. Publication bias may make positive or novel results easier to find than null findings. Researchers should acknowledge these limits rather than claiming complete knowledge of a field.

Best practice includes:

1. Start with a focused question and concept matrix.
2. Search at least two appropriate scholarly databases.
3. Add professional and official sources when the question concerns practice or policy.
4. Keep a dated search and screening record.
5. Use explicit inclusion and exclusion criteria.
6. Read methods and limitations, not only abstracts and conclusions.
7. Compare definitions, samples and designs before comparing findings.
8. Distinguish evidence from interpretation.
9. Use quotations sparingly and paraphrase accurately.
10. Update searches and check every reference before submission.
11. Protect confidential data and follow institutional research-integrity requirements.
12. Make the review serve the research question rather than attempting to include everything.

Chapter Summary

A critical literature review is an analytical account of existing knowledge that defines concepts, evaluates evidence, compares competing explanations and justifies a new study. Effective searching begins by converting a research topic into concepts, synonyms and eligibility criteria. Boolean operators, database selection, citation chaining and a search log improve coverage and transparency. Sources must be evaluated for authority, relevance, methodological quality, currency and integrity; citation counts alone are insufficient.

Synthesis requires grouping studies by themes or mechanisms and explaining convergence and disagreement in relation to context, theory, measurement and design. A defensible research gap may be empirical, conceptual, theoretical, methodological, contextual, contradictory or practical. The written review should progress from context and concepts to debates, evaluation, gap and research justification. Reference software and AI can support organisation and discovery, but the researcher remains responsible for verification, ethical use and scholarly judgement.

Key Terms

Literature review: A structured and analytical account of existing scholarship on a research problem.

Critical literature review: A review that evaluates evidence, assumptions, methods and implications rather than merely describing sources.

Peer review: Expert evaluation of scholarly work before publication.

Scholarly source: Academic publication produced for research, such as a journal article or scholarly book.

Professional literature: Evidence produced by professional bodies, consultancies, employers or policy organisations.

Boolean operator: A search command such as AND, OR or NOT used to combine terms.

Truncation: A search technique using a symbol to retrieve multiple word endings.

Concept matrix: A planning table linking research concepts to synonyms and related terms.

Inclusion criterion: A condition that a source must meet to enter a review.

Exclusion criterion: A condition used to remove an otherwise identified source.

Citation chaining: Finding sources through the references of an article or later works that cite it.

Credibility: The extent to which a source is trustworthy and supported by appropriate evidence.

Construct validity: The degree to which a measure represents the concept it claims to measure.

Internal validity: The plausibility that a study’s observed relationship reflects the proposed explanation.

Synthesis: Combining evidence from multiple sources to develop an integrated interpretation.

Research gap: A justified limitation, uncertainty or absence in existing knowledge.

Publication bias: The tendency for published research to differ systematically from completed but unpublished research.

Evidence matrix: A structured table used to record and compare source characteristics and findings.

Research integrity: Commitment to honest, transparent, accurate and responsible research practice.

Review Questions

Short Answer Questions

1. What is the difference between a literature review and an annotated bibliography?
2. Explain the functions of AND and OR in a Boolean search.
3. Identify four dimensions that can be used to evaluate source credibility.
4. What is meant by citation chaining?
5. Why is a research gap more than an unexplored topic?

Long Answer Questions

1. Explain how a broad HRM topic can be transformed into a focused, searchable research question.
2. Compare narrative, systematic, scoping and integrative literature reviews, indicating when each may be appropriate.
3. Discuss how methodological differences can explain conflicting findings in HRM research.
4. Describe a transparent process for searching, screening and recording literature for a dissertation.
5. Explain how a critical literature review should lead from existing evidence to a justified research design.

Analytical / Application Questions

1. Construct a concept matrix and Boolean search string for a study of flexible working and employee retention among early-career employees.
2. Two studies report opposite conclusions about remote work and engagement. Develop an analytical framework for determining whether the disagreement results from differences in sample, measurement, theory or research design.
3. A student uses generative AI to produce ten apparently relevant references, but three cannot be found and two misrepresent their findings. Evaluate the academic-integrity risks and propose a responsible workflow.

Multiple Choice Questions with Answers

Question 1

What is the primary purpose of a critical literature review?

A. To list every publication found on a topic

B. To evaluate and synthesise existing evidence in order to justify a study

C. To replace the methods section of a dissertation

D. To report only findings that support the researcher’s hypothesis

Correct Answer: B. To evaluate and synthesise existing evidence in order to justify a study

Question 2

Which Boolean operator normally broadens a database search by combining synonyms?

A. AND

B. NOT

C. OR

D. NEAR

Correct Answer: C. OR

Question 3

Which source is generally most appropriate for current official labour-market statistics?

A. An anonymous blog

B. A government statistical agency

C. An undated social-media post

D. A student essay

Correct Answer: B. A government statistical agency

Question 4

What does citation chaining involve?

A. Counting the citations in a journal’s impact report

B. Finding sources through an article’s references and later citing publications

C. Replacing all database searches with Google Scholar

D. Removing articles with few citations

Correct Answer: B. Finding sources through an article’s references and later citing publications

Question 5

Which statement best describes a research gap?

A. Any topic that the researcher finds interesting

B. A claim that no previous research exists

C. A justified limitation or uncertainty in existing knowledge

D. A list of articles published in the last year

Correct Answer: C. A justified limitation or uncertainty in existing knowledge

Question 6

Why should citation count not be treated as a direct measure of article quality?

A. Citation counts are always unavailable

B. Highly cited articles may be old, controversial or widely criticised

C. Only professional reports can be cited

D. Peer-reviewed articles cannot receive citations

Correct Answer: B. Highly cited articles may be old, controversial or widely criticised

Question 7

Which practice most directly improves the transparency of a literature search?

A. Using only the first page of search results

B. Keeping a record of databases, dates, search strings and filters

C. Selecting sources based only on title length

D. Deleting all excluded records

Correct Answer: B. Keeping a record of databases, dates, search strings and filters

Question 8

A cross-sectional survey finds that flexible working is associated with engagement. What conclusion is most defensible?

A. Flexible working definitely causes engagement

B. Engagement definitely causes flexible working

C. The study demonstrates an association, but causation requires further justification

D. The finding has no possible value

Correct Answer: C. The study demonstrates an association, but causation requires further justification

Question 9

Which is an example of a methodological research gap?

A. A theory has never been named

B. Existing studies rely almost exclusively on cross-sectional surveys

C. A topic is mentioned in a newspaper

D. A researcher prefers interviews

Correct Answer: B. Existing studies rely almost exclusively on cross-sectional surveys

Question 10

What is the most responsible use of generative AI in literature searching?

A. Citing every reference it produces without checking

B. Using it to suggest keywords, then verifying sources and findings in originals

C. Uploading confidential participant data for summarisation

D. Allowing it to write the review without disclosure

Correct Answer: B. Using it to suggest keywords, then verifying sources and findings in originals