Dissertation Literature Review Example (PhD): Annotated Sample & Writing Guide

This guide walks you through a complete, annotated PhD-level literature review example, with a focus on original contribution and methodological positioning — the two things that separate a dissertation lit review from every other kind.

What Makes a Dissertation Literature Review Different?

Dimension Master's Thesis Lit Review PhD Dissertation Lit Review

Primary goal Show you understand the field Defend your original contribution

Depth of synthesis Thematic grouping Thematic + methodological + theoretical critique

Gap treatment Identify a topic-level gap Identify a methodological, theoretical, or empirical gap you will fill

Methodology engagement Brief mention of methods used in cited studies Deep engagement with why certain methods were chosen and their limitations

Positioning 1 paragraph linking to research questions 2–4 pages justifying your approach, design, and contribution

Typical length 10–20 pages 20–40 pages

Annotated PhD Dissertation Literature Review Example

Topic: The Impact of Algorithmic Content Moderation on Political Discourse in Democracies

Note: This is a condensed example. A full dissertation lit review would expand each section significantly and engage 100–200+ sources.

Section 1: Introduction

Example passage

The proliferation of algorithmic content moderation on social media platforms has transformed how political discourse unfolds in democratic societies. Since 2016, when major platforms began deploying machine learning systems at scale to detect and remove policy-violating content, scholars have debated whether these systems protect democratic deliberation or distort it (Gorwa et al., 2017; Roberts, 2019). Early research emphasized the opacity of these systems and their potential for over-removal of legitimate political speech (Myers West, 2018; Tufekci, 2018). More recent work has shifted toward empirical measurement of moderation effects on specific political outcomes, including electoral participation (Grinberg et al., 2019) and polarization (Bail et al., 2018; Levy, 2021).

This dissertation reviews literature across four interconnected domains: (1) the technical architecture of algorithmic moderation systems, (2) the political and social consequences of automated content removal, (3) the regulatory and legal responses across democratic states, and (4) the methodological challenges of studying platform governance. This review argues that while substantial progress has been made in documenting moderation outcomes, the field lacks a unified methodological framework for cross-platform comparative analysis — a gap this dissertation addresses.

Annotations

Sentence(s) Function

"The proliferation of algorithmic content moderation…" Establishes the phenomenon and stakes

"Since 2016…" Anchors the timeline — shows you know when the field emerged

"Early research emphasized…" Maps the evolution — demonstrates the field has developed over time

"More recent work has shifted toward…" Signals current debates — shows you're current

"This dissertation reviews literature across four domains…" Roadmap — tells the reader exactly what's coming

"This review argues that…" Thesis of the lit review itself — this is the through-line argument, not just a summary

Writing formula

[Phenomenon] has transformed [field/context]. Since [year], scholars have debated [core tension] (Author, Year; Author, Year). Early research emphasized [X] (Author, Year), while more recent work has shifted toward [Y] (Author, Year; Author, Year). This dissertation reviews literature across [N] domains: [1], [2], [3], and [4]. This review argues that [through-line argument that sets up your gap].

Section 2: Technical Architecture of Algorithmic Moderation

Example passage

Algorithmic content moderation relies on a combination of machine learning classifiers, natural language processing, and computer vision systems trained on platform-specific policy guidelines (Gorwa et al., 2017; Roberts, 2019). These systems operate through two primary mechanisms: proactive detection, where content is flagged before user visibility, and reactive detection, where user reports trigger algorithmic review (Myers West, 2018).

A significant body of scholarship has examined the accuracy limitations of these systems. Noble (2018) demonstrated that training data for content classifiers often reflects the biases of their creators, leading to disproportionate removal of content from marginalized communities. Similarly, Benjamin (2019) argued that the technical architecture of moderation algorithms embeds what she terms "discriminatory design," where historical inequalities are encoded into automated decision-making systems.

However, more recent technical scholarship has pushed back against the narrative of inevitable bias. Raji et al. (2020) proposed a framework for algorithmic auditing that enables external researchers to test moderation systems for disparate impact without requiring platform cooperation. Building on this, Sandvig et al. (2021) developed a methodology for "reverse engineering" content moderation decisions through controlled experimentation, providing a model for independent assessment of algorithmic accuracy.

Despite these advances, a persistent methodological limitation remains: most technical studies examine single platforms in isolation. No published research has systematically compared the architecture and outcomes of moderation systems across multiple platforms (Facebook, YouTube, TikTok, X/Twitter) using a unified analytical framework — a gap that has implications for both scholarly understanding and regulatory approaches.

Annotations

Move Example

Defines the domain Explains what algorithmic moderation is technically

Presents the critique Noble (2018) and Benjamin (2019) — systems embed bias

Signals the counter "However, more recent technical scholarship has pushed back…"

Presents the response Raji et al. (2020) and Sandvig et al. (2021) — auditing methodologies

Identifies the gap Single-platform studies only; no cross-platform framework

Key PhD-level move: Notice how this section doesn't just summarize findings — it traces a methodological evolution (from identifying bias → proposing audits → developing reverse engineering → identifying the next methodological frontier). This is what PhD lit reviews must do: show how methods have developed and where they've hit a wall.

Section 3: Political and Social Consequences

Example passage

The political consequences of algorithmic moderation have been studied primarily through two lenses: effects on political speech and effects on political behavior.

On speech, research has documented both over-removal and under-removal problems. Over-removal occurs when legitimate political content is mistakenly flagged, disproportionately affecting opposition voices in competitive democracies (Huang & Knight, 2020; Wahutu, 2020). Under-removal, conversely, refers to the failure to detect coordinated inauthentic behavior, as seen in the 2016 U.S. election (Howard & Kollanyi, 2016) and the 2020 U.S. election (Grinberg et al., 2019). These dual failures have led scholars to question whether platforms can simultaneously achieve both accuracy and political neutrality (Klonick, 2017; Suzor, 2019).

On behavior, Bail et al. (2018) conducted a large-scale experiment demonstrating that exposure to cross-cutting political content on social media can increase polarization rather than reduce it — a finding that complicates the assumption that more speech is always better for democracy. Levy (2021), however, found that algorithmic curation of political content had minimal effects on polarization when controlling for pre-existing partisan identity, suggesting the relationship between algorithmic moderation and political outcomes is conditional rather than deterministic.

These contradictory findings point to a broader methodological challenge: studies of moderation effects on political behavior have used widely varying experimental designs, making cross-study comparison difficult. This dissertation addresses that challenge by proposing a standardized comparative framework.

Annotations

Move Example

Establishes sub-structure Two lenses: speech and behavior

Presents both sides of a problem Over-removal vs. under-removal

Contrasts contradictory findings Bail et al. (2018) vs. Levy (2021)

Names the methodological challenge Inconsistent experimental designs

Positions the dissertation "This dissertation addresses that challenge…"

Section 4: Gap Identification & Positioning

Example passage

Three methodological gaps emerge from this review. First, while individual platforms have been studied extensively, no research has compared moderation architectures across platforms using a unified framework. Second, the political consequences literature suffers from design heterogeneity that prevents cumulative knowledge-building. Third, and most significantly for this dissertation, no study has integrated technical architecture analysis with political outcome measurement in a single comparative design.

This dissertation addresses the third gap through a multi-platform comparative case study of content moderation systems on Facebook, YouTube, and TikTok during the 2024 electoral cycles in the United States, United Kingdom, and European Union. By combining technical audits (following Raji et al., 2020) with political outcome surveys (following Levy, 2021), this study provides the first integrated assessment of how moderation architecture shapes political discourse across platforms and democracies. The contribution is both methodological — a unified comparative framework — and substantive: evidence on whether algorithmic moderation strengthens or undermines democratic deliberation under different institutional conditions.

Annotations

Move Example

Three gaps, escalating importance Gap 1 (scope) → Gap 2 (method) → Gap 3 (integration — your entry point)

Names the specific design Multi-platform comparative case study

Names the methods being combined Technical audits + political outcome surveys

States the dual contribution Methodological (framework) + substantive (evidence)

Writing formula for PhD positioning

[N] gaps emerge from this review. First, [Gap 1]. Second, [Gap 2]. Third, and most significantly for this dissertation, [Gap 3 — the integration gap].

This dissertation addresses [which gap] through [specific research design]. By combining [Method A] with [Method B], this study provides the first [type of contribution]. The contribution is both methodological — [methodological output] — and substantive: [substantive finding/output].

Common PhD Literature Review Mistakes

1. Describing studies instead of arguing across them. Every paragraph should build toward your gap, not just report what each author found.

2. Weak methodological engagement. At the PhD level, you must discuss not just what studies found but how they studied it and why those choices matter.

3. The "laundry list" structure. If your headings are author names ("Smith's Study," "Jones's Findings"), restructure around themes and debates.

4. Overclaiming the gap. "No one has ever studied social media" is false and easily disproven. Be precise: "No study has compared X across Y using Z method."

5. Under-citing. A PhD lit review typically engages 100–200+ sources. If you have 30 citations, you're not there yet.

How Acade.ai Supports PhD Literature Reviews

A dissertation literature review is an extended research task. Acade can support connected stages of the workflow while the researcher retains responsibility for verification and interpretation:

• Source discovery: Build Boolean strategies and search relevant academic sources. Coverage requirements depend on the discipline and institution.

• Candidate gap analysis: Organize verified sources to identify possible methodological or empirical limitations, then confirm any gap through a current search and expert judgment.

• Synthesis support: Organize sources by theme, debate, or method and draft comparison-focused paragraphs for human review.

• Citation support: Format references in common styles where the selected plan provides citation features. Verify every output against the original source and current style guide.

• Writing workflow: Move from notes and an outline to thematic sections, a candidate gap statement, and a bounded positioning paragraph.

Tell Acade your doctoral research question, source boundaries, theoretical framework, and methodological constraints. Ask it to help organize the evidence and build a positioning structure for your review. Verify every source and novelty claim before use.

Plan My PhD Literature Review

Continue with the Complete Literature Review Guide

For the full workflow—from introduction and thematic synthesis to gap identification and positioning—see the Literature Review Example for Dissertation & Thesis: Complete Annotated Guide.

Acade is an academic research and writing assistant, not an author or substitute for a supervisor, methods expert, librarian, statistician, clinician, or ethics committee. Human researchers remain responsible for source verification, interpretation, original contribution, disclosure, and submission. Product features, credits, and plan availability may change; check the current product information before use.