**3. Method**

We undertook a systematic literature review to capture the state-of-the-art research in the EDI field and classify its effect areas at the micro and meso levels. We aim to address our RQ by examining the literature incorporated within this review. Our focus was on academic peer-reviewed work, and we adhered to the methodical approach proposed by Xiao and Watson [12] to structure and execute our analysis. This approach consists of eight steps: (1) formulate the problem, (2) develop and validate the review protocol, (3) search the literature, (4) screen for inclusion, (5) assess quality, (6) extract data, (7) analyze and synthesize data, and (8) report findings. The subsequent paragraphs elaborate on our application of these eight steps.

Step (1) was undertaken within the Introduction and Theoretical background sections of this article, wherein we presented the RQ that directed this systematic literature review, which was formulated in an iterative process. Our goal was to collate and consolidate pertinent literature on EDI that explicates its effects on the micro and meso levels. As suggested by Xiao and Watson [12], we conducted a quick mapping to identify aspects such as subtopics, keywords, and publication years of the literature, thereby assessing the feasibility of our analysis in terms of the emerging volume of literature.

In step (2), we crafted and validated the review protocol, which contains the review's objective and search strategy that guided our approach. The literature was simultaneously reviewed and coded by two authors. The protocol was made accessible to all team members in a shared file, allowing for collaborative maintenance and updates.

During step (3), we determined the electronic databases to be used for the literature search. Based on our prior experiences and their diversity, we selected AISeL, IEEE Xplore, Emerald Insight, and ACM Digital Library, considering that EDI synthesizes various research strands as an interdisciplinary field of study. Furthermore, we utilized Google Scholar as a complementary data source, examining the top 100 search results sorted by relevance to the search string. From the RQ and the keywords derived from the preliminary mapping, we developed a Boolean search string which was applied across the aforementioned electronic databases: "*('organizational growth' OR 'institutional growth' OR 'organizational development' OR 'organizational change' OR 'institutional development' OR 'institutional change') AND (entrepreneurship OR skills OR training OR competencies OR intrapreneurship OR development OR*

*Unfolding Effect Areas of Employee-Driven Innovation: A Systematic Literature Review DOI: http://dx.doi.org/10.5772/intechopen.112160*

*management) AND ('employee-driven innovation' OR 'employee innovation').*" This string incorporated identified synonyms, assisting us in locating the most pertinent literature to our RQ while maintaining a manageable volume of literature. The total search yielded 980 results (as illustrated in **Figure 2**), from which we discarded 789 deemed irrelevant to the RQ by their title, leaving us with 191 results. Where any doubt was cast on the relevance of the literature, we opted to include these results in our further analysis.

For step (4), we established a set of inclusion and exclusion criteria to facilitate the decision-making process during the content screening of the results, subsequently determining which papers would be included in the final synthesis. On a meta-level, the selected papers had to be written in English, accessible, and peer-reviewed. Additionally, they had to provide specific content that discusses and describes EDI effects to address the RQ appropriately. The information on these effects was extracted and stored within a shared document. We decided to include "conceptual" papers, for example, theoretical and meta-approaches, but to mark them explicitly. Applying the defined inclusion and exclusion criteria, we screened the abstracts of the remaining 191 results and excluded 126 due to unsuitable content and one due to being a duplicate, leaving 64 results.

In step (5), we obtained full-text versions of the remaining results and undertook comprehensive reviews to evaluate their relevance to the RQ. Discrepancies in rating a result's relevance were resolved through team discussions until a consensus was reached. Consequently, 33 results were excluded due to their lack of appropriate content and one due to lack of accessibility, leading to a total of 30. A subsequent forward and backward search of these resulted in the identification of five additional ones. After this step, we had a set of 35 articles from which we extracted and analyzed the data for synthesis.

During step (6), we coded qualitative statements pertinent to our RQ from the final literature dataset to extract all EDI effects. We adopted an inductive approach wherein the codes were directly derived from the data rather than being predetermined. This laid the foundation for further synthesis of the data and

condensing individual statements into overarching effect areas via interrelated themes, concepts, and facilitators [48].

In step (7), we organized and analyzed the coded data using an integrated design that blended both quantitative and qualitative research methods [49]. Upon completion of the coding process, we moved to data synthesis. This involved aggregating the 124 coded individual statements and identifying common effect patterns. These were collectively discussed within the author team to agree on the final EDI effect areas that form the structural basis of the concept matrix.

Finally, step (8) entailed reporting the findings, which is the objective of this article. We present a concept matrix in line with the approach proposed by Webster and Watson [13], which visualizes the curated collection of relevant literature alongside the identified EDI effect areas. Subsequently, each effect area is described, and implications and future research avenues are discussed in the following sections to complete this step.
