The process of conducting a literature review

Independent of what approach will be used to conduct the literature review, a number of steps that must be taken and decisions made to create a review that meets the requirements for publication (for specific considerations in relationship to each step. See Table 3). In the fol­lowing, the basics steps and important choices involved in conducting a literature review will be suggested and discussed using four phases; (1) designing the review, (2) conducting the review, (3) analysis and (4) writing up the review. This process was developed from practical ex­perience and is a synthesis of and influenced by various standards and guidelines suggested for literature reviews (e.g., Liberati et al., 2009; Tranfield et al., 2003; Wong et al., 2013).

1. Phase 1: designing the review

The first question that should be asked is why this review should be conducted. Is there really a need for a literature review in this area? If so, what type of literature review would be the most helpful and would make the greatest contribution? Of consideration should also be what audience will most likely be interested in the review when deciding on the topic. This is a relevant question because it determines the like­lihood of the review being published and the impact it will have on the research community. Conducting a literature review is hard work, so the topic must be one that is of interest to both the author and reader. Therefore, it is a good idea to scan the area as a first step to account for other literature reviews that already exist, to assess the number of re­search studies that must be assessed, and to help formulate and clearly define the purpose, scope, and specific research question the review will address. These are important actions because they will help to identify which approach is most appropriate. For example, if the review aims to summarize or evaluate a large field of research or even several research areas, a strict systematic review approach may not be suitable or even possible. Instead, a narrative or integrative review approach would be preferable. In the same way, if the purpose of the review is to in­vestigate and synthesize evidence of the effect of a specific factor, an integrative review is not trustworthy; instead, a systematic review ap­proach should be used. The stated purpose should then guide the rest of the review.

Once the research question has been identified and an overall review approach considered, a search strategy for identifying relevant literature must be developed. This includes selecting search terms and appropriate databases and deciding on inclusion and exclusion criteria. Here, a number of important decisions must be made that are crucial and will eventually determine the quality and rigor of the review. Search terms can be words or phrases used to access appropriate arti­cles, books, and reports. These terms should be based on words and concepts that are directly related to the research question. Depending on the aim of the review and the research question, these search terms can be broad or narrow. Importantly, it could be worthwhile to consider including additional limitations.

As almost all initial literature searches yield many articles, a strategy is needed to identify which are actually relevant. Inclusion criteria for the review should be guided by the selected research question. Criteria that can be considered and are commonly used are, for example, year of publication, language of the article, type of article (such as conceptual, randomized controlled trail, etc.), and journal. In terms of research quality, deciding on inclusion and exclusion criteria is one of the most important steps when conducting your review. However, important to note is the need to provide reasoning and transparency concerning all choices made; there must be logical and valid motives. This is important, as, independent of the type of ap­proach, the quality of the literature is dependent on, among other as­pects, what literature is included and how it was selected (Tranfield et al., 2003; Wong et al., 2013). Depending on these decisions, a study can end up with very different answers and conclusions to the same research questions. For example, by only selecting some specific jour­nals, years, or even search terms to try to limit your search, you can end up with a very flawed or skewed sample and missing studies that would have been relevant to your case or even contradict other studies. You can also come to the wrong conclusion about gaps in the literature, or perhaps more serious, provide false evidence of a specific effect. A practical approach is to write all decisions down to enable transpar­ency, as the authors must be clear in a way that enables the reader to understand how the literature was identified, analyzed, synthesized, and reported. This should be done carefully and prior to actually con­ducting the review.

2. Phase 2: conducting the review

After deciding on the purpose, specific research questions, and type of approach, it is time to start conducting the actual review. When conducting the review, a pilot test of the review process and protocol is appropriate. By testing the search terms and inclusion criteria on a smaller sample, the process can be adjusted before performing the main review. It is common to adjust the process a number of times before actually selecting the final sample. Importantly, it should be noted that it is preferred to use two reviewers to select articles to ensure the quality and reliability of the search protocol.

The actual selection of the sample can be done in a number of ways, depending on the nature and scope of the specific review. Depending on how many articles are yielded, different approaches will be appro­priate. For example, reviewers may read each piece of literature that appears in the search in full; this is a highly useful, but time-consuming approach. Another option could be to focus on the research method or findings, and a third option is to conduct the review in stages by reading abstracts first and making selections and then reading full-text articles later, before making the final selection. Once this is done and the initial articles (or other relevant literature) have been collected, the texts should be screened in full to ensure they meet the inclusion criteria. As an additional strategy, references in the selected articles can be scanned to identify other articles that may potentially be relevant (however, this is not appropriate when using the systematic review method as this requires a more strict protocol). During this time, the process of in­cluding and excluding specific articles should be documented carefully.

3. Phase 3: analysis

After conducting the literature review and deciding on a final sample, it is important to consider how the articles will be used to conduct an appropriate analysis. That is, after selecting a final sample, a standardized means of abstracting appropriate information from each article should be used. Data abstracted can be in the form of descriptive information, such as authors, years published, topic, or type of study, or in the form of effects and findings. It can also take the form of con­ceptualizations of a certain idea or theoretical perspective. Importantly, this should be done in concordance with the purpose and research question of the specific review, and the form will vary. In this step, it is important to consider training the reviewers to avoid any differences in coding and abstraction (if there is more than one) and monitoring the data abstraction carefully during the review process to ensure quality and reliability. Often, if the aim is to publish in an academic journal, this will require a detailed description of the process or a measure of reliability between reviewers. Sometimes this is easy, if the information of interest is, for example, population, effect size, or sample size. However, it becomes harder when the information of interest is themes in the literature, perspectives, or providing an historical timeline.

Depending on the review, different analysis methods can be used and are more or less appropriate (please see above for different con­tributions from different approaches). Nevertheless, independent of the method of analysis, it is important to ensure that it is appropriate to answer the selected research question. For example, if the purpose is to evaluate evidence of the effect of loyalty programs, the use of a meta­analysis is most appropriate. On the other hand, if the purpose was to develop a theoretical model or framework for customer experience, a strict meta-analysis would be a poor choice; rather, an analysis tech­nique suitable for integrative reviews should be used.

4. Phase 4: writing the review

First, when writing the review, the motivation and need for the review must be clearly communicated. Depending on the approach, the final review article can be structured in different ways, and it will re­quire different types of information and different levels of detail. A number of standards and guidelines explicitly address how literature reviews should be reported and structured, including PRISMA, devel­oped for systematic literature reviews and meta-analyses (see Liberati et al., 2009); RAMSES, developed for systematic narrative reviews (see Wong et al., 2013); and guidelines for integrative reviews (Torraco, 2005). Although review articles can be organized in various ways, some generalizations can be made. All authors are expected to follow ac­cepted conventions for reporting on how the study was undertaken. It is necessary to describe transparently the process of designing the review and the method for collecting literature, that is, how the literature was identified, analyzed, synthesized, and reported by the author. Doing so properly gives the reader the chance to assess the quality and trust­worthiness of the findings. The contribution of the specific literature review can take a number of forms, and it should be judged in re­lationship to the field to which it wants to contribute. Depending on a number of factors, such as the maturity of the field or state of knowl­edge, different contributions could be valuable. For example, literature reviews can result in a historical analysis of the development within a research field (e.g. Carlborg, Kindstrom, & Kowalkowski, 2014), an agenda for further research (e.g., McColl-Kennedy et al., 2017), a conceptual model or categorization (e.g., Snyder et al., 2016; Witell et al., 2016), or evidence of an effect (e.g., Verlegh & Steenkamp, 1999).

Source: Hannah Snyder (2019), “Literature review as a research methodology: An overview and guidelines”, Journal of Business Research, Volume 104, Pages 333-339, https://doi.org/10.1016/j.jbusres.2019.07.039