How to Write a Literature Review

The literature review is the chapter where many MS and PhD students get stuck. Often it ends up as a long list of "Author A found this, Author B found that" with no argument. A good review is different: it critically synthesises what is known, shows where the evidence agrees and conflicts, and builds a convincing case for why your study is needed. This guide gives you a step-by-step process, including a small quantitative synthesis you can do by hand.

What a Literature Review Must Do

A literature review should:

• map the key theories, concepts and debates in your area;
• evaluate the quality of existing studies, not just summarise them;
• identify patterns, contradictions and gaps;
• justify your research questions, variables, hypotheses and methods;
• provide the theoretical framework your study tests.

Types of Review

• Narrative (traditional) review: the most common type in theses. It is flexible but can be subjective.
• Systematic review: follows a pre-specified, reproducible search and selection protocol, usually reported with a PRISMA 2020 flow diagram.
• Meta-analysis: statistically combines effect sizes from many studies.
• Scoping review: maps how much research exists on a broad topic and what kind, often to identify gaps.

Many theses use a narrative review with systematic elements: a clearly described search strategy plus a thematic synthesis.

Step 1: Define Your Scope

Start from a focused research question. Decide on your inclusion criteria before you search:

• publication years (for example 2010–2025);
• study types (empirical, peer-reviewed);
• populations (for example university students in South Asia);
• languages.

Frameworks such as PICO (Population, Intervention, Comparison, Outcome) help in health and education research.

Step 2: Search Systematically

Use several databases: Scopus, Web of Science, Google Scholar, PubMed, ERIC, IEEE Xplore and JSTOR, along with the HEC Digital Library available through Pakistani universities. Build search strings with Boolean operators:

("job satisfaction" OR "work satisfaction") AND ("transformational leadership") AND (bank* OR "financial sector") AND (Pakistan OR "South Asia")

• AND narrows the search, OR broadens it with synonyms, and quotation marks search for exact phrases.
• The asterisk (*) is a truncation wildcard, so bank* finds bank, banks and banking.
• Use citation chaining: check the reference lists of key papers (backward) and "cited by" lists (forward).

Record your searches, including the database, the date, the search string and the number of results. Examiners increasingly expect this transparency.

Step 3: Screen and Manage Sources

Screen titles and abstracts first, then full texts, against your criteria. Use a reference manager such as Zotero, Mendeley or EndNote from day one. Prioritise peer-reviewed journals, and check journal quality through indexing (for example Scopus, Web of Science or the HEC Journal Recognition System). Be wary of predatory journals.

Step 4: Build a Synthesis Matrix

This is the single most useful tool for moving from summary to synthesis. Create a spreadsheet with one row per study and columns for the key features:

Now read down the columns, not across the rows. Do findings differ by sector? By method? By year? Those patterns become your themes.

Step 5: Synthesise Quantitatively: A Worked Example

Suppose the three studies above all examined the correlation between transformational leadership and job satisfaction. Their sample sizes were \( n = 53, 83 \) and \( 123 \). How strong is the relationship overall? A simple fixed-effect meta-analysis answers this.

Step 1: Convert each r to Fisher's z, which makes correlations approximately normal:

$$ z = \tfrac{1}{2}\ln\!\left(\frac{1 + r}{1 - r}\right) $$

• \( r = .30 \) gives \( z = 0.3095 \)
• \( r = .45 \) gives \( z = 0.4847 \)
• \( r = .20 \) gives \( z = 0.2027 \)

Step 2: Weight each study by \( w = n - 3 \), giving weights of 50, 80 and 120.

Step 3: Weighted mean z.

$$ \bar{z} = \frac{\sum w z}{\sum w} = \frac{50(0.3095) + 80(0.4847) + 120(0.2027)}{250} = \frac{15.48 + 38.78 + 24.33}{250} = 0.3143 $$

Step 4: Standard error and 95% CI.

$$ SE = \frac{1}{\sqrt{\sum w}} = \frac{1}{\sqrt{250}} = 0.0632 $$

$$ 0.3143 \pm 1.96(0.0632) = [0.1903,\ 0.4383] $$

Step 5: Convert back to r using \( r = \tanh(z) \):

$$ \bar{r} = \tanh(0.3143) = .30, \qquad 95\%\ \text{CI} = [.19,\ .41] $$

Step 6: Check heterogeneity.

$$ Q = \sum w(z - \bar{z})^2 = 3.82, \qquad df = 2, \quad p = .148, \qquad I^2 = \frac{Q - df}{Q} = 48\% $$

Sanity check: the pooled \( \bar{r} = .30 \) lies between the smallest (.20) and largest (.45) study values, and it sits closer to the large nurse study, which carries the most weight.

How to write it: Across three Pakistani studies (N = 259), the pooled correlation between transformational leadership and job satisfaction was moderate, r = .30, 95% CI [.19, .41]. Heterogeneity was moderate, I² = 48%, and the weakest effect appeared in the healthcare sector, suggesting that sector may moderate the relationship. That final clause turns a summary into an argument and can motivate a moderation hypothesis.

With only three studies, Q has very little power, so treat heterogeneity tests cautiously. In R, the metafor package runs these analyses (escalc() and rma()). Comprehensive Meta-Analysis (CMA), Stata's meta command and JASP are alternatives.

r library(metafor) dat <- escalc(measure = "ZCOR", ri = c(.30, .45, .20), ni = c(53, 83, 123)) rma(yi, vi, data = dat, method = "FE") # pooled z = 0.314

Step 6: Organise and Write

Organise your review thematically or conceptually, and only chronologically if the history of ideas is itself important. A typical structure is:

1. Introduction: the scope, search strategy and how the review is organised.
2. Theoretical framework: the main theories, such as social exchange theory or the theory of planned behaviour.
3. Thematic sections: one per variable or relationship, often matching your hypotheses.
4. Synthesis and gap: what is known, what is contested and what is missing.
5. Conceptual framework and hypotheses: a diagram linking your variables.

Write synthesis sentences that connect several sources:

• Weak (summary): "Khan (2019) found a positive relationship. Ali (2021) also found a positive relationship."
• Strong (synthesis): "Survey studies in banking and education consistently report moderate positive associations (Ali, 2021; Khan, 2019), but evidence from healthcare is weaker (Raza, 2023), possibly reflecting shift-based work structures."

Critique as you go: comment on sample sizes, measurement validity, cross-sectional designs and generalisability. End every section by linking back to your study.

Common Mistakes

• Listing studies one by one with no themes or argument.
• Relying on outdated sources, or on websites instead of peer-reviewed research.
• Leaving out contradictory evidence.
• Quoting at length instead of paraphrasing and synthesising.
• Failing to state the research gap explicitly.

Did You Know?

• The PRISMA statement for reporting systematic reviews was first published in 2009 and updated in 2020.
• Gene Glass coined the term "meta-analysis" in 1976.
• The Cochrane Collaboration, founded in 1993, has produced thousands of systematic reviews that guide medical practice worldwide.

Key Takeaways

• A literature review synthesises and critiques evidence to justify your study. It is not a list of summaries.
• Search systematically with Boolean strings across several databases and record your strategy.
• Use a synthesis matrix and read down the columns to find themes, contradictions and gaps.
• Where studies report comparable effect sizes, a simple meta-analysis such as pooling r via Fisher's z adds rigour to your argument.

If your literature review needs structure, synthesis or a meta-analysis, Amathify's subject experts and PhD scholars offer step-by-step guidance and assignment writing support. Reach us on WhatsApp or use the Ask a Question form.