1 Introduction

Modeling research in the COVID-19 vaccination drive requires mapping the hypothesized relationship between different identified elements (Romate et al., 2022). In the past, modeling research has been aided by various methodologies, including structural equation modeling (Hair et al., 2019; Singh, Sharma, & Dhir, 2021; Dixit et al., 2021; Singh et al., 2022), interpretive structural modeling (Warfield, 1974) (ISM), system dynamics (Sterman, 2000), and total interpretive structural modeling (Sushil, 2012; Sushil, 2017a; Sushil, 2018a; Sushil & Dinesh, 2022) (TISM). TISM and ISM transform ill-structured factors into well-structured mental models and offer explanations of what and how. Furthermore, TISM also explains why components and defines the reason for the relationship between two identified elements. Previously, a few advancements in the TISM modeling process were suggested, including m-TISM (Sushil, 2017a), argumentation-based TISM (Sushil & Dinesh, 2022), and TISM-P (Sushil, 2018b). m-TISM is the modified TISM, where a simultaneous transitivity check was performed. Thus, it eliminated various steps for performing transitivity. In TISM-P, polarity in the relationships was identified. It provided the nature of the relationship (positive/negative) between two elements. Earlier, these relationships were identified based on interviews or defined relationships in the existing literature. However, there is another dimension to building an understanding of TISM, i.e., through evidence data (Sushil, 2017b). This dimension explains the relationship and nature of elements based on available evidence data. Therefore, the relationships are simultaneously validated. Numerous evidence data points can also lead to the generalization of results as compared to other methods with limited sample sizes.

The researchers have suggested using evidence data for developing hierarchical models that can aid the decision-making process in organizations (Sushil, 2017b; Chan & Moses, 2016; Nyawa et al., 2022; Rathore & Ilavarasan, 2020). Furthermore, acquiring data from end-users can facilitate the success of new product development (Piller & Walcher, 2006). Social media platforms can enable sharing views and opinions from end-users (Luo et al., 2022; Zhang et al., 2022a; Alkouz et al., 2022; Alwabel & Zeng, 2021; Ullah et al., 2021). Various researchers have studied evidence data for improving the existing process or developing a new product (Rathore & Ilavarasan, 2020; Singh et al., 2020; Zhang et al., 2022b). Further, scholars have studied the effect of the COVID-19 pandemic on the global operations disruptions (Dubey et al., 2022; Masudin et al., 2021; Meier & Pinto, 20222022; Momeni et al., 2022; Sarker et al., 2021) and communication (Tam et al., 2021; Paramita et al., 2021; Warrier et al., 2021). In addition, previous studies have used TISM to develop a hierarchical framework to address various issues. For instance, Sushil and Anbarasan (2021) developed a framework for sustainable operational complexity in organizations. Dwivedi et al. (2021) suggested a framework in value chain flexibility to address the issue of sustainable initiatives. However, very limited studies are conducted to understand the end-users’ views about the “COVID-19 vaccination drive success”. Therefore, this illustrates the applicability of evidence data to understand the “COVID-19 vaccination drive success”.

This study aims to propose an enhanced methodology of TISM-P based on evidence data (TISM-E) to aid the modeling for the COVID-19 vaccination drive success. Further, the study interprets the results based on the evidence data and contributes to the existing literature on TISM, social media analytics, and the COVID-19 vaccination literature.

The study first presents a selective review of the TISM methodology. In Sect. 3, the study enhances the TISM-P method based on evidence data (TISM-E). Section 4 discusses the steps of TISM-E in the context of modeling the elements for the success of the COVID-19 vaccination drive. At last, the study presents discussions, limitations, conclusions, and offers future research directions.

2 Overview of TISM: review and application

Sushil (2012) highlighted the unclear articulated model of ISM and suggested a way of interpreting the directed link of ISM through a more evolved methodology, i.e., TISM. It provided a step-by-step direction for understanding and building a TISM framework. Sushil (2018a) provided detailed guidelines to check the correctness of the TISM model. In TISM, “large number of pair-comparisons are to be made by experts and cumbersome multi-order transitivity checks are to be carried out” (Sushil, 2018b, p. 39). Therefore, this work was enhanced by m-TISM (Sushil, 2017a), which led to simultaneously checking the transitivity among the elements. This enhancement added more efficiency to the TISM methodology. However, a missing dimension in the hierarchical model was polarity of relationships. Sushil (2018b) introduced the concept of polarity in the m-TISM methodology (TISM-P). Polarity helps to identify the nature of the relationships. Sushil & Dinesh (2022) highlighted the importance of argumentation-based discussion to enrich the interpretation of links.

TISM is widely used in supply chain management and sustainability research (Sushil & Anbarasan, 2021; Dwivedi et al., 2021; Yadav et al., 2021; Dohale et al., 2022). For instance, Dubey et al. (2017) studied sustainable supply chain management and analyzed the drivers and their interlinkages. Balaji and Arshinder (2016) reviewed the food supply chain and modeled the causes of food wastage. Mangla et al. (2014) studied the performance variables under the uncertainty and risk of a sustainable supply chain. Further, the application of TISM can also be seen in the area of industry 4.0 (Ajmera & Jain, 2019; Jain & Ajmera, 2020), lean manufacturing (Chaple et al., 2021; Singh et al., 2018; Virmani et al., 2018), flexibility (Yadav & Sagar, 2021; Jain & Raj, 2015; Mangla et al., 2014), higher education (Sravat & Pathranarakul, 2022; Menon et al., 2021), implementation (Chaple et al., 2021; Singh & Dhir, 2022), start-up ecosystem (Sindhwani et al., 2022; Singh et al., 2020; Singh et al., 2019), and information systems (Dwivedi & Madaan, 2020; Singh & Singla, 2021). With the advancement of technologies and fields, TISM is also applied in the area of blockchain (Shardeo et al., 2020; Mathivathanan et al., 2021; Dwivedi et al., 2022), internet of things (Patil & Suresh, 2019; Singh et al., 2020), artificial intelligence (Mir et al., 2020; Jain et al., 2021), and health care research (Logesh & Vinodh, 2022; Priyadarsini et al.,  2020; Patri & Suresh, 2017).

3 Methodology

3.1 Enhanced TISM Methodology with evidence-based on Semantic Network

The evidence-based TISM (TISM-E) is a methodological extension of TISM (Sushil, 2012), m-TISM (Sushil, 2017a), argumentation-based TISM (Sushil & Dinesh, 2022), and TISM-P (Sushil, 2018b). The modified TISM is the methodological extension of TISM and ISM with a simultaneous transitivity check. TISM with polarity was introduced to study the nature of the relationships. It helps to interpret the results with clarity in polarity of the relationship. However, these studies were performed using interview data. It required a greater involvement of experts to define polarity of the relationships. The application of evidence data (such as data available on social media) was majorly ignored to identify linkages between the elements. Further, the application of evidence data can enhance generalizability of the findings. In this study, TISM with polarity (Sushil, 2018b) is taken as a basis to describe the steps of this evidence-based TISM. The steps of this enhanced evidence-based TISM methodology are presented in Fig. 1.

Figure 1 highlights that steps I, II, III, and VII are different from the defined steps of TISM-P. These steps introduce the basic guidelines for building evidence-based TISM. Steps VIII and IX are additional steps in the enhanced process. Steps IV, V, and VI are similar to the existing methodology of TISM. The steps of evidence-based TISM are discussed below:

Step I: Collect the evidence data from any Social Media platform and perform texts cleaning, tokenization, word stemming, and lemmatization.

The user-generated texts are collected from Twitter using various hashtags. After data collection, text cleaning was performed. It helps to convert all words to small letters and removes punctuations, stop words, and other noise data. Further, tokenization was performed to split the sentence into meaningful parts and identify individual entities in a sentence. In addition, word stemming and lemmatization were also executed. Word stemming helps to convert various forms of one word to its root form. Lemmatization groups together the inflected forms of a word to its base or dictionary form of a word.

Step II. Define elements based on the evidence data.

Based on the evidence data, the elements are identified. These elements can be used to study the hierarchical relationships using the evidence-based TISM.

Step III: Define contextual relationships based on node-link and semantic network as association methods.

The semantic network relationship is formed with the help of node-link data. Further, the contextual relationships can be visualized based on word matrices based on the semantic network graph. UCINET (Borgatti et al., 2002; Yoo & Lim, 2021) or other similar software can be used to visualize the network graph.

Step IV: Prepare digraph for simultaneous transitivity check and prepare transitive reachability matrix.

Sushil (2017a) describes the process of successive comparison of elements. The author(s) need to compare the first two sets of elements (i,j) and (j,k); further, the comparison between (i,k) is checked based on rule of transitivity. The identified relationship among the identified elements can be forward, backward, transitive, or no relationship. Based on the digraph, a binary transitive matrix is prepared, as elaborated by Sushil (2017a).

Step V: Perform hierarchical partitioning of reachability matrix.

Hierarchical partitioning is performed based on the reachability set, antecedent set, and intersection set (Sushil, 2018; Sushil, 2017a). The elements with same reachability set and intersection set are labeled as level I and are placed at the highest level in hierarchy. Further, this element is removed from the set. This process is repeated iteratively till all the elements are categorized into different hierarchical levels.

Step VI: Prepare hierarchical digraph with selected transitive links.

All the elements are arranged in hierarchical levels. The direct and important transitive links are shown in the digraph as per the transitive binary matrix.

Step VII: Interpret the nodes and links along with polarity of links and obtain TISM based on evidence data.

The interpretation of nodes and links are interpreted with + ve or -ve orientation based on evidence data. In the process, the excerpts from the evidence data can be used for interpretation of the relationship and identification of polarity of relationships. Final TISM model based on evidence data is obtained.

Step VIII: Obtain dependence and driving power of all elements.

Based on driving and dependence power with + ve or -ve orientation, autonomous, driver, linkage, and dependent elements are identified.

Step IX: Identify the paths from driving variable to dependent variable via different linkage or intermediate variable with polarity of relationships based on evidence data.

The paths from the dependent elements to driving elements with linkage / intermediate elements are identified along with their polarity.

In the next section, the above-mentioned steps are illustrated with the help of an illustration in the context of the COVID-19 vaccination program success.

4 Results

4.1 Illustration: COVID-19 vaccination program success model

The present evidence-based TISM is illustrated with an example of successful modeling of COVID-19 vaccination program.

Step I: The data from Twitter platform was collected using hashtags including covid19vaccine, coronavaccine, coronavirusvaccine, and covidvaccine. Scholars have used Twitter analysis to understand the existing processes or to analyze perception of new product (Rathore & Ilavarasan, 2020; Singh et al., 2020; Zhang et al., 2022b). The tweets were collected between 11 November 2020 and 20 February 2021. The tweets collected demonstrate pre- and post-vaccination launch period. The period demonstrates launch of vaccination campaigns all over the world. A total of 11,09,367 tweets were collected. After removing duplications, the data cleaning was performed on 2,60,297 tweets. At first, the unnecessary characters, including punctuations, URLs, hashtags, and special characters were removed. Next, the stop words were removed. Further, tokenization was performed. At last, word stemming was done based on the snowball algorithm (Porter, 2001).

Fig. 1
figure 1

Indicative steps of total interpretive structural modeling with polarity based on evidence data (TISM-E)

Full size image

Step II: The keywords are selected after analyzing the word frequency. The keywords are chosen based on their overall relationship with the overall success of the COVID-19 vaccination program. Therefore, six keywords are identified. Thus, belief, communication, distribution, frontline, policies, and rollout are identified. Further, these keywords were interpreted based on the Tweets. Therefore, the identified keywords are further interpreted and defined as the population’s belief, communication effectiveness, distribution efficacy, role of frontline workers, role of government policies, vaccine rollout implementation, and success of the COVID-19 vaccination program. Further, the factors are defined based on the extant literature and tweets.

Belief of the population

The belief of the population can predict health-promoting behavior by addressing the association between health behaviors and health services utilization. The beliefs of the population were related to the risk of getting infected by the COVID-19 and benefits of getting vaccinated.

Communication effectiveness

The tweets on communication were related to clarity on vaccine safety and efficacy, the COVID-19 vaccine challenges, national vaccine communication, and education campaign, etc., to name a few. Majority of the communications were aimed to make the population confident about taking the COVID-19 vaccination. It is believed that honest, open, and transparent communication is essential for vaccination uptake by the population.

Distribution efficacy

The government aimed to optimize the distribution of vaccines to vaccination sites to minimize the period needed to vaccinate individual populations. In places where the vaccine distribution is not utilized properly, communication can play a vital role.

Role of frontline workers

Frontline health workers are the pillar of health ecosystems, especially in delivering healthcare services in rural areas. The frontline healthcare workers play a vital role in responding to and managing the COVID-19 vaccination process. They have worked in extreme conditions and are responsible for managing the population’s fear, anxieties, and hesitancy.

Role of government policies

The government’s goals, decisions, and actions have played an important role in the COVID-19 vaccination process. The policy of vaccine allocation to states and vaccination policy related to different age groups has a major impact on the success of the COVID-19 vaccination program. Other policies such as wearing masks, maintaining proper social distancing, and closing schools are crucial for success.

Vaccine roll-out implementation

The vaccine roll-out implementation focuses on the key aspects such as preparation of a central vaccination system, training of healthcare workers, strategy for distribution, managing supply chains, and preparing communication plans for vaccine uptake and acceptance. The implementation exercises can be done at the central and regional levels. The main purpose of these exercises is that the vaccines are delivered correctly and disseminated as planned.

Success of COVID-19 vaccination program

The success of the COVID-19 vaccination program can be seen in short term and long term. In the short term, the success can be seen via interrupting disease transmission, including fewer hospitalizations, reduced deaths, and fewer cases. In the long run, the success can be measured by institutional capabilities to foster vaccine confidence among diverse communities, enhanced public understanding regarding vaccination’s value to society, and heightened public trust in government.

Step III: The relationship between the factors is based on node-link data. The interrelationships between the factors are analyzed by studying and comparing the data. Figure 2 shows the semantic network relationship between the identified factors. Table 1 presents the interrelationship between the factors. UCINET software is used to visualize the semantic network based on the interrelationship between the factors.

Fig. 2
figure 2

Semantic network of key factors

Full size image
Table 1 Binary matrix
Full size table

Step IV: Digraph for simultaneous transitivity check is prepared (Fig. 3) based on the evidence data as mentioned in Fig. 2; Table 1. The diagraph shows direct and transitive linkages. Based on the digraph (Fig. 3), binary matrix with transitivity is prepared (Table 2). The relationships are entered as 1, 1*, and 0. 1 denotes a direct relationship, 1* denotes a transitive relationship, and 0 denotes no relationship.

Step V: Level partitioning is performed based on the transitive reachability matrix. The iterations for level partitioning are highlighted in Appendix Table A1. The level-wise summarization of factors for success of the COVID-19 vaccination program is presented in Table 3. It is seen that the ‘success of the COVID-19 vaccination program’ is at level I in the hierarchy (highest level), and ‘role of government policies’ is at level IV (lowest level).

Table 2 Transitive reachability matrix
Full size table
Table 3 List of elements and their levels in TISM
Full size table
Fig. 3
figure 3

Digraph for transitivity check

Full size image

Step VI: All the elements F1, F2, F3, F4, F5, F6, and F7 are arranged in the hierarchical levels. All the direct links and selected transitive links are shown in the digraph as per the transitive reachability matrix (Fig. 4).

Step VII: Based on data collected from Twitter, the nodes and links, along with the polarity of links, are interpreted. Table 4 lists the direct links, the excerpts from Twitter, and the interpretation of the linkages. Further, the final TISM model based on evidence data is drawn, showing the polarity of the relationships (Fig. 5).

Table 4 Links, sample excerpts, polarity, and their interpretations
Full size table
Fig. 4
figure 4

Digraph after hierarchical partitioning

Full size image

Step VIII: Based on the TISM model, transitive reachability matrix with the polarity of the relationship is drawn. Further, dependence and driving power with + ve or -ve orientation is calculated (Table 5). Additionally, the elements are categorized into autonomous, driver, linkage, and dependent factors. Figure 6 shows the graphical classification for the identified elements of ‘success of COVID-19 vaccination program’. Further, the role of frontline workers (F4), role of government policies (F5), and vaccine rollout implementation (F6) are classified as driving factors. The elements belief of the population (F1), communication effectiveness (F2), and distribution efficacy (F3) are linkage factors. The success of the COVID-19 vaccination program (F7) is a dependent factor. The orientation of the elements is presented in Table 5 as + ve/-ve driving power and + ve/-ve dependence. The elements have more + ve dependence and driving power, so the identified elements can be treated as + ve. For instance, element F5, i.e., role of frontline workers (driving factor), has more + ve driving power than -ve. Therefore, it can be treated as + ve driving factor.

Table 5 Modified binary transitive matrix based on evidence data with dependence and driving power
Full size table
Fig. 5
figure 5

Total interpretive structural modell based on evidence data for COVID-19 vaccination program success

Full size image
Fig. 6
figure 6

Classification of elements for COVID-19 vaccination program success

Full size image

Step IX: Thirteen key paths are identified that can lead towards success of the COVID-19 vaccination program based on driving, linkage, and dependent factors. Further, three factors are categorized as driving factors. Therefore, the driving factors are taken as the starting point to trace the paths and their influence on final dependent variable (Table 6). It can be observed from Table 6 that 8 paths are + ve and 5 paths can have both + ve and -ve impact on the dependent factor. This kind of analysis describes the success of COVID-19 vaccination program adopted in any country or region.

Table 6 Nature of paths
Full size table

5 Discussion

This study provides an enhanced version of TISM as evidence-based TISM or TISM-E (applicable to ISM also as ISM-E) based on evidence data generated from social media platforms. The proposed methodology (as presented in Fig. 1) is different from the existing process of TISM or ISM in collecting data from social media platforms, identification of elements, defining the relationship between the elements, and interpreting the relationship with polarity. Further, it can be noted that a few steps related to preparing digraph for transitivity check, hierarchical partitioning, and hierarchal digraph are similar to the existing methodology of TISM. The enhanced version of TISM-E is primarily based on collecting, analyzing, and interpreting evidence data to identify and define the relationship between the elements. The additional insights by incorporating the evidence data will integrate this TISM method with social media analytics to understand the interrelationships between the identified elements. Further, this method will have practical applications to information management, change management, success model development, and operations management.

In the mentioned context of COVID-19 vaccination program success, the study identifies six elements by collecting and analyzing data from Twitter. Further, based on the node-link data semantic network was drawn, and the relationships between the elements were identified. In addition, the links were interpreted, and the polarity of the relationships were also identified based on tweets. Additionally, in this study, most of the relationships were positive, and a few were both positive and negative. For instance, the relationship between belief of the population and the success of the COVID-19 vaccination program is both positive and negative. The belief and trust in vaccination can increase the chances of success of the COVID-19 vaccination program. In contrast, any disbelief or mistrust in vaccination can decrease the chances of success for the COVID-19 vaccination program. The identified relationships conceptualize a model for the success of the COVID-19 vaccination program and can be validated using empirical measures.

5.1 Implications for research

The advanced version of TISM, i.e., TISM-E will be useful for various research fronts, including developing the research based on practical insights where the scant literature is available and hypothesizing the relationship between the elements based on semantic network.

The TISM-E will also facilitate the validation of relationships by capturing insights from tweets. It can act as an alternate pathway for validating relationships from experts. The positive and negative relationship can also be captured from tweets. Therefore, it can help in identifying the facilitators or barriers of the research topic. Further, the evidence data can also help in building propositions for theory development and testing. The TISM-E also contributes to graph-theoretic literature. The hierarchical digraph developed can be used as an input to a mathematical model. Based on the relationships and interdependencies, the matrix can be formed for further calculations. In addition, the measures of nodes can be quantified for simulation purposes or assessment of relationships.

5.2 Implications for practice

For practitioners, the evidence-based TISM (TISM-E) can be useful in the identification of various elements along with polarity that can contribute to the success of the project/policy under study. This study will also enable the practitioners to identify the facilitators and inhibitors. Therefore, the practitioners can plan a proper course of action for minimizing the effects of inhibitors. In practice, the enhanced version of TISM can be applied to different areas including supply chain, sustainability, new product launches, policy implications, strategic thinking, etc., to name a few. In the mentioned areas, the evidence data will play a major role in understanding and generalizing the role of facilitators or inhibitors. The COVID-19 effects are long-term. Therefore, by developing a model for successful COVID-19 vaccination program will enable policy makers to vaccinate maximum population and overcome the challenges related to vaccination processes. In addition, the model will lead to distribution efficacy. The distribution efficacy can be achieved through role of government policies, vaccine rollout implementation, and proper communication. Further, in the developed hierarchal model more positive linkages are seen, and there are a few both positive and negative relationships. Therefore, the model also highlights that how an element’s negative impact should be minimized to promote the positive desired outcome. The model developed can also be applied to other vaccination programs for better outreach.

6 Conclusion

This research provides the enhanced version of ISM and TISM based on evidence data from social media platforms. The enhanced version of TISM-E is based on social media analytics. The method provides a mix of quantitative and qualitative analysis along with the identification of polarity. The identification of elements and relationships are based on social media analytics, i.e., node-link data and semantic network. The interpretations are based on social media data gathered through different platforms.

Polarity identification is based on the tweets that depict the relationship between two elements. The proposed method contributes to the existing literature on TISM, graph-theoretic, and modeling literature, in addition with social media analytics literature. Further, the research has important implications for research and practice. In addition, the study also extends the literature on the COVID-19 vaccination and provides a path for success of the COVID-19 vaccination program. The elements such as role of frontline workers, role of government policies, and vaccine rollout implementation are driving factors. In contrast, the elements including communication effectiveness, belief of the population, and distribution efficacy are categorized as linkage factors. The independent and linkage factors lead to the success of COVID-19 vaccination program.

The limitations of this study are twofold. First, the study captures and defines the relationship using social media platform data. Therefore, some factors important for the study could be ignored or missed. In this case of the COVID-19 vaccination program, wastage is an important factor for the success of the overall program. But overall, the relationships and elements were not identified using Twitter analytics. Therefore, to overcome such limitation Twitter data can be combined with existing literature. Second, the linkages and polarity are defined using tweets excerpts. Therefore, subjectivity can be a problem during interpretation. It is important to validate such interpretation and linkages with the help of experts or based on previous literature. The TISM-E is a methodological extension of TISM and ISM with polarity. It can be examined across different research intersections using TISM and social media analytics. Further, the interrelationship of the elements can be tested using the structural equation modeling technique.