{"id":23978,"date":"2025-06-01T07:00:00","date_gmt":"2025-06-01T11:00:00","guid":{"rendered":"https:\/\/cehhs.utk.edu\/elps\/?p=23978"},"modified":"2025-05-14T08:27:14","modified_gmt":"2025-05-14T12:27:14","slug":"my-journey-in-writing-a-bibliometric-analysis-paper","status":"publish","type":"post","link":"https:\/\/cehhs.utk.edu\/elps\/my-journey-in-writing-a-bibliometric-analysis-paper\/","title":{"rendered":"My Journey In Writing A Bibliometric Analysis Paper"},"content":{"rendered":"\n
<\/div>\n\n\n\n
\"\"<\/figure>
\n

As a third-year doctoral student in Evaluation, Statistics, and Methodology at the University of Tennessee, Knoxville, I recently completed a bibliometric analysis paper for my capstone project on Data Visualization and Communication in Evaluation.<\/em> Bibliometrics offers a powerful way to quantify research trends, map scholarly networks, and identify gaps in literature. It is an invaluable research method for evaluators and researchers alike. Hello everyone! I am Richard D. Amoako.<\/strong> <\/p>\n\n\n\n

Learning bibliometrics isn\u2019t always straightforward. Between choosing the right database, wrangling APIs, and figuring out which R or Python packages won\u2019t crash your laptop, there\u2019s a steep learning curve. That\u2019s why I\u2019m writing this: to share the lessons, tools, and occasional frustrations I\u2019ve picked up along the way. Whether you\u2019re an evaluator looking to map trends in your field or a researcher venturing into bibliometrics for the first time, I hope this post saves you time, sanity, and a few coding headaches. Let\u2019s explore the methodology, applications, and resources that shaped my project. <\/p>\n<\/div><\/div>\n\n\n\n

<\/div>\n\n\n\n

Understanding Bibliometric Analysis<\/strong> <\/p>\n\n\n\n

Bibliometric analysis is the systematic study of academic publications through quantitative methods- examining citations, authorship patterns, and keyword frequencies to reveal research trends. Bibliometric analysis differs from traditional literature reviews by delivering data-driven insights into knowledge evolution within a field. Common applications include identifying influential papers, mapping collaboration networks, and assessing journal impact (Donthu, et al., 2021; Van Raan, et a., 2018; Zupic & \u010cater, 2015). <\/p>\n\n\n\n

For evaluators, this approach is particularly valuable. It helps track the adoption of evaluation frameworks, measure scholarly influence, and detect emerging themes, such as how data visualization has gained traction in recent years. My interest in bibliometrics began while reviewing literature for my capstone project. Faced with hundreds of papers, I needed a way to objectively analyze trends rather than rely on subjective selection. Bibliometrics provide that structure, turning scattered research into actionable insights. <\/p>\n\n\n\n

Key Steps in Writing a Bibliometric Paper<\/strong> <\/p>\n\n\n\n

Defining Research Objectives<\/strong> 
The foundation of any successful bibliometric study lies in crafting a precise research question. For my capstone on data visualization in evaluation literature, I focused on: “How has the application of data visualization techniques evolved in program evaluation research from 2010-2025?” This specificity helped me avoid irrelevant data while maintaining analytical depth. Before finalizing my question, I reviewed existing systematic reviews to identify underexplored areas – a crucial step that prevented duplication of prior work. When brainstorming and refining your thoughts, utilize productive technologies such as generative AI tools (such as ChatGPT, Claude, Perplexity, Google Gemini, Microsoft Copilot, DeepSeek, etc.)  to enhance and clarify your ideas.   <\/p>\n\n\n\n

Database Selection and Data Collection<\/strong> 
Choosing the right database significantly impacts study quality. After comparing options, I selected Scopus<\/a> for its comprehensive coverage of social science literature and robust citation metrics. While Web of Science<\/a> (WoS) offers stronger impact metrics, its limited coverage of evaluation journals made it less suitable. Nonetheless, I examined the potential applications of using WoS. Google Scholar’s expansive but uncurated collection proved too noisy for systematic analysis. Scopus’s ability to export 2,000 records at once and include meta-data such as author affiliation, country proved invaluable for my collaboration mapping. <\/p>\n\n\n\n

Data Extraction and Automation<\/strong> 
To efficiently handle large datasets, I leveraged R’s Bibliometrix package. Use this R script<\/a> to automate your data extraction with the Scopus API<\/a> (Application Programming Interface). APIs enable software systems to communicate with each other. Researchers can use APIs to automate access to database records (like Scopus, WoS) without manual downloading. To access the Scopus database, request access via Elsevier\u2019s Developer Portal<\/a>. <\/p>\n\n\n\n

Pros:<\/em>\u202fGood for large-scale scraping.\u202fCons:<\/em>\u202fRequires API key approval (can take days or weeks).  <\/p>\n\n\n\n

For targeted bibliometric searches, carefully construct your keyword strings using Boolean operators (AND\/OR\/NOT) and field tags like TITLE-ABS-KEY() to balance recall and precision \u2013 for example, my search TITLE-ABS-KEY(“data visualization” AND “evaluation”) retrieved 37% more relevant papers than a simple keyword search by excluding off-topic mentions in references. <\/p>\n\n\n\n

After exporting Scopus results to CSV, a simple script converted and analyzed the data (Aria & Cuccurullo, 2017<\/a>): <\/p>\n\n\n\n

library(bibliometrix)<\/em> <\/p>\n\n\n\n

M <- convert2df(“scopus.csv”, dbsource = “scopus”, format = “csv”)<\/em> <\/p>\n\n\n\n

results <- biblioAnalysis(M)<\/em> <\/p>\n\n\n\n

This approach provided immediate insights into citation patterns and author networks.  <\/p>\n\n\n\n

Data Screening and Cleaning<\/strong> 
The initial search may return many papers; my search returned over 2,000. To narrow down the most relevant articles, you can apply filters such as: <\/p>\n\n\n\n

    \n
  1. Removing duplicates via DOI matching [use R code, M <- M[!duplicated(M$DO), ]<\/em> #Remove by DOI. Duplicates are common in multidatabase studies.\u00a0\u00a0<\/li>\n<\/ol>\n\n\n\n
      \n
    1. Excluding non-journal articles\u00a0<\/li>\n<\/ol>\n\n\n\n
        \n
      1. Excluding irrelevant articles that do not match your research questions or inclusion criteria\u00a0<\/li>\n<\/ol>\n\n\n\n
          \n
        1. Manual review of random samples to verify relevance\u00a0<\/li>\n<\/ol>\n\n\n\n

          Additional data cleaning may be required. I use R’s tidyverse<\/em>, janitor or dplyr<\/em> packages for these tasks.  <\/p>\n\n\n\n

          The screening process can be overwhelming and time-consuming if performed manually. Fortunately, several tools and websites are available to assist with this task. Notable examples include abstrackr<\/a>, convidence.org<\/a>, rayyan.ai<\/a>, AsReview<\/a>, Loonlens.com<\/a>, and nested-knowledge<\/a>. These tools require well-defined inclusion and exclusion criteria. It is essential to have thoroughly considered criteria in place. Among these tools, my preferred choice is Loonlens.com, which automates the screening process based on the specified criteria and generates a CSV file with decisions and reasons upon completion. <\/p>\n\n\n\n

          Analysis and Visualization<\/strong>  <\/p>\n\n\n\n

          Key analytical approaches included (refer to the appendices for R codes and this guideline<\/a>): <\/p>\n\n\n\n