Making Personal Networks with a Computer: Lessons from the Field using Vennmaker

The initial aim of the transdisciplinary group of Vennmaker developers was to create a software tool that “enables users to interactively collect network relationship data from an actor’s point of view and render them comparable and quantitatively analyzable by means of an intuitive graphical user interface” (Kronenwett and Schönhuth, 2011: 9). The software takes its name from the Venn diagram, developed by the mathematician and philosopher John Venn in his Symbolic Logic (1881). These Venn diagrams have been used widely to represent relations between sets. In personal networks, Kahn and Antonucci (1980) introduced social convoys as concentric circles representing degrees of social support. Recent comparative research on network data-collection and visualization has found that research participants prefer concentric circles over other tools like free drawings or funnels, although that does not affect the network’s composition or size (Hollstein et al., 2020).

Indeed, Vennmaker is a user-friendly software for easily building, visualizing, and analyzing networks. Its welcoming environment allows paper-to-laptop solutions and permits using participatory interviews to make networks on the screen. It does this not only after making a structured questionnaire, but also by directly “drawing” on the screen and making an audio recording, among other features. Its applications are therefore numerous, depending on the research objectives. For example, it allows questions and answers to be posed, while the networks themselves can be visualized and qualitatively changed or evaluated while simultaneously reflecting on their structures and characteristics. Hence, interviewees might co-construct their networks, which provides agency to the research participants and gives preeminence to the process, thereby achieving scientific standards in generating and processing social-network analysis while narrowing the quantitative and qualitative gaps in network research.

Vennmaker has been available since 2010, along with other similar software with visual interfaces, such as Egoweb or OpenEddi (Maya-Jariego and Cachia, 2019), while comparable new and modernized software such as Network Canvas is also appearing (Birkett et al., 2021). Nonetheless Vennmaker continues to be used both for those who are beginning personal-network analysis and those who have not found software that surpasses it, or only one whose learning curve or price is too high. And yet, given the pros and cons, the software is free and easy to use, can be self-taught, runs well in both Apple and Microsoft environments, does not need a powerful computer or a lot of battery power, and it is available in English, German, Russian, Spanish, Chinese and French. All these characteristics make this software very well suited for researchers with no resources to invest in hardware or software, nor the funding to travel and pay for specialist courses, and it is available in other languages, which helps those with limited knowledge of English. Our intention in this article is to map out how the software has been used and give some examples from our own respective experiences for those who are assessing software(s) or just wondering about the potential of Vennmaker.

Gamper, Schönhuth and Kronenwett (2012) describe how Vennmaker can be used for research, especially when bringing together qualitative and quantitative data. They argue that using this software makes possible a combination of these two worlds in a clear way: adding data is through visualization, not a matrix, even though everything is at the same time coded as a matrix that can be exported. In relation to data-collection, this combination is unique and useful. The advantages of using this software include the range of possible representations and storing network maps, maps that can be flexibly adjusted to the interview, audio files that can be recorded at the same time as the visual representation, easy collection of time periods, and ease of sending data to collaborators. However, there are also some drawbacks: digital network maps may not be suitable for the less educated, the elderly and those who cannot cope with computers. Moreover, the requirement to use a good-quality computer, possibly involving extra cost and needing to adapt smart-phones or tablet computers, together with the possibility of having to use a computer during fieldwork, are potential disadvantages, as are issues of data security and the protection of sources.

Vennmaker has been recognized by many authors as a legitimate tool for analyzing ego’s network data because it allows the researcher and ego to move alters on the screen and add information about the types and strengths of ego’s relations with their alters (Halgin and Borgatti, 2012). The software seems to be accessible to anyone because Vennmaker does not require users to enter relational data in a matrix; instead, they produce a visualization of their networks, to be exported as a database later (During et. al, 2011). Vennmaker has been used as a software tool for digitalized visualizations of networks after collecting them manually with pen and paper (Herz and Oliver, 2012). In this case, the software is helpful in making a clearer representation accompanied by the categorization and codification of the relations within it. For example, this ease of use has been exploited in historical research to draw networks of families in ancient history and the personal networks of Jews hiding during the Nazi period (Düring et al., 2011). This ability to draw, which is characteristic of Vennmaker, also allows individuals, groups and organizations to be included in the same network, thus facilitating, for example, the study of migrants’ transnational businesses (Sommer and Gamper, 2021), as well as assessing their transnational ties and practices qualitatively (Bilecen, 2021).

If, then, we draw a random network in Vennmaker, we can export that drawing as data categorized in an Excel document, among others. Another alternative is to assist the interviewees to make their own networks in the program as a visualization rather than an adjacency matrix avoiding errors (Cachia and Maya, 2017). This approach has been recognized for being a simpler and faster method of codifying, visualizing and analyzing networks (Gamper, et al., 2012; Bilecen, 2013).

Vennmaker is commonly used as part of a mixed-methods approach. For example, a study of female sex-workers in Mexico collected quantitative network measures, qualitative narratives and network visualizations simultaneously (Wagner et al., 2018). The potential of the program to collect both qualitative and quantitative data while permitting reflection and validation was also highlighted by Borucki (2022) in her study of the local networks of politicians and welfare workers in Germany. Vennmaker allows the data to be exported so that it can be analyzed in statistical software such as SPSS, as is done by Borucki (ibid.). It could also be rendered complementary with another tools, such as Qualitative Data Analysis Software, to analyze reflections or discourses, as shown by Santha’s research among trainee teachers (Santha, 2017). Despite the different perspectives that can be taken by the researcher, this type of software could actually help in improving the quality of the data that has been collected (Hancean, Molina and Lubbers, 2016).

Lippe and Gamper (2016) made a comparison to try and clarify which methods are better for collecting ego-centered networks. They compared using a questionnaire with visualization-based data-collection in two different studies with the same objectives and questions. The second of these used the digital visual method that Vennmaker allows them. Even though the design of both pieces of research were the same, some differences were reported: those who answer in Vennmaker tend to have larger networks, as they do not have so many restrictions other than the limitations of the screen drawing. Respondents in Vennmaker reported more family members, while there was a greater variation in the number of non-kin in the questionnaire. Besides, the questionnaire produces a higher rating of alter-alter closeness and of denser and more strongly interconnected alter-alter relations because in Vennmaker people can create each of these connections with an arrow manually or in a matrix. Indeed, drawing or presenting alter-alter relations directly on the screen might reduce the burden on the respondent (McCarty et al., 2007) and produce more participation and satisfaction than using matrices (Eddens and Fagan, 2018). Making alter-alter ties one by one is quicker, though it also introduces an overall bias in large networks due to the fatigue in listing ties visually. This can cause interviewees to stop listing their ties before they exhaust them or reorientating the software’s visualization, which is modified when ties are added (McCarty et al., 2019). Therefore, the authors concluded that “it might not be a question of ‘right or wrong’ method, but instead that of what we wish to achieve” (Lippe and Gamper, 2016: 12).

Another study conducted by Lelong et al. (2016) comes to a similar conclusion. They made a comparison between two different software products, Net-Map and Vennmaker, concluding that each one works best for different purposes: Net-Map is helpful for exploratory research, while Vennmaker seems to work best for pre-structuring interviews and comparisons between different networks. Besides, they stressed that Vennmaker is especially helpful in avoiding “research fatigue” (Stevenson, 2003: 2). These are the types of reflection that this article tries to make visible and available to any other researchers who, in the future, would like to use this software. Using this article, they can take them into consideration beforehand.

Before describing our case studies, we explain some of the methodological decisions we have made. Both authors of this paper had used Vennmaker as the main software in conducting social network analysis, and both collected personal networks during long ethnographic fieldwork, which is helpful to reflect on data collection, analysis and visualizations of networks. In the next section, each researcher makes an account of their study by focusing on seven main aspects that allow comparison: (1) description of case study and main question; (2) identification of benefits and challenges during research with the program; (3) description of fieldwork and data collection; (4) the influence of using a computer both during and after fieldwork; (5) description of data analysis; (6) visualization and presentation of networks; and (7) research findings. Then, after our two case studies have been presented, we will reflect on our respective experiences to try to disentangle some of the benefits, challenges and lessons of using Vennmaker in studying personal networks before offering some concluding remarks.

This first case study was led by the first author of this article, Ignacio Fradejas-García. It forms part of an unfinished PhD thesis entitled “Aid workers: work–life practices in the Syrian humanitarian cross-border operation from Turkey”. This study analyses humanitarian cross-border operations from Turkey to Syria, and specifically the work–life practices and power dynamics of aid-workers. Although the thesis remains unfinished for political and ethical reasons, some research results have been already published (Fradejas-García, 2019; Fradejas-García and Mülli, 2019). In this article, I, as first author, will zoom in on the use of Vennmaker software and consider what lessons can be drawn from the experience I gained from this research.

My fieldwork extended over two years (2015–2016) while I was living in the border city of Gaziantep in Turkey. This twin city of Aleppo, with two million inhabitants, promptly became an important destination for Syrian refugees fleeing the civil war in their country and was transformed into a node for sending cross-border humanitarian aid into Syria. Numerous headquarters of NGOs, UN agencies and international organizations were settled in this ‘duty station’, producing a new transnational space in Aidland in which three categories of aid-worker emerged: internationals, Syrians and Turkish. Participant observation was carried out in various local and international NGOs, as well as in other public and private spaces in which different aid-workers work or spend their free time. In some cases, I was able to volunteer for an NGO, having already been accepted by workers and organizations to conduct my research. Moreover, I performed dozens of informal interviews; few people were interested in being formally interviewed using audio-recordings. I also held a focus-group discussion about humanitarian work and completed sixteen formal qualitative interviews, in which I used Vennmaker as follows.

The qualitative interviews were designed with a strong biographical focus in a flexible model divided into three parts: (1) life-history, (2) personal network(s), and (3) current humanitarian life–work balance. The first part included interviewees’ life-courses and trajectories, with special emphasis on their long-term trajectories, (im)mobilities and working experiences, and it lasted between thirty minutes and one hour. The second part extended to one hour and covered the creation, visualization and on-site analysis of the interviewee’s personal network using Vennmaker. The software enables audio-recordings and drawings to be made and allows the network to be modified on the screen in line with the interviewee’s corrections until an accurate final network emerges. These features allow one to focus on the content and meaning of the interviewee’s network by collecting relevant qualitative information (Ryan et al., 2014).

The third part of the interview was kept open-ended and focused on current aid-workers’ lives and work, the humanitarian operation and future plans. The aim of this chronological structure was to cover the biographical aspects and to revisit some questions if they had been left unclear in previous answers. Also, the breaks between each part allowed the interviewees to rest, which might increase the duration of the interview but also helped them to reflect on or cross-examine previous questions, and even to postpone one part, thereby avoiding interviewee fatigue (in a couple of interviews, the last part of the interview was held in a second session). The total duration of most of the interviews went beyond the two hours initially set aside for them.

The use of Vennmaker in the second part of the interview brought in my own laptop. However, if the laptop had any problem or if the interviewee, who had previously been informed about the interview’s parts and process, did not want to construct the network in the laptop, I also brought with me paper on which I had marked out concentric circles, coloring pencils and little post-it notes. During the design of the personal network interview, I carried out one pilot interview on paper (see Figure 1 left). This helped me spot some minor issues with the interview design in Vennmaker when the data gathered were entered into the software (see Figure 1 right), and to work out how to conduct the interview on paper in case the laptop or the software failed.

Figure 1:

Interviews were carried out in public spaces such as coffee shops, but some were done in interviewees’ homes or in NGO offices after working hours and with permission. The laptop was received positively, as informants’ anonymity and confidentiality were guaranteed through their giving their informed consent and through encrypted data storage. Indeed, Vennmaker allows anonymization using a password, and every network was anonymized in front of the interviewee. Images of their networks, both anonymized and not, were shared with them, which they usually received and welcomed as a gift.

The interview process is as follows. To start the interview, I sit together with the interviewee looking at the screen. I manage the keyboard and open Vennmaker. As a default, the software shows a network map with ego in the middle, which helps to explain how the alters are created and located, and what the network looks like at the end using this software. Second, I open the questionnaire with five questions with name generators designed in accordance with my objectives. These questions elicit the following information about the interviewee’s alters:

Who supports you emotionally?

Third, at the end of the questionnaire, the software places all alters on to the side bar, and one by one they can be dragged and located into the network, according to their place of living and profile. Also, some attributes are requested and filled in for each person: sex/gender, age, place of origin (nationality), place of residence (duty station); Family, Friends, Co-worker, Others (girl-boyfriend/fiancée/spouse); Expat / Turkish / Syrian / Other; Work organization (just for aid workers); Job position (just for aid workers); and salary. This is the most difficult part of the questionnaire, as filling in each alters’ attributes takes time to answer and is very repetitive.

Then some other alters and their attributes can be added or modified directly onto the map. This is an important feature because, when the network is visualized, it is easier for research participants to spot individuals who were missed in the questionnaire. Indeed, follow-up questions and thinking aloud increase network size and aid the interviewee’s memory by adding important alters previously forgotten (Hollstein et al., 2020). The next step is to ask interviewees about relations with their alters. Thus I asked: Have your “alters” been contacting each other independently of you? Following interviewees’ instructions, lines indicating relations are drawn between alters. There is no need to draw all the lines between ego and his or her alters if it is assumed that ego has a relationship with them. The question of bad relations was also raised, but no one reported any.

Finally, the interviewee was asked to make any changes and reflect on the resulting network, which also validated the results. Overall, in some situations these reflections were very hard for international workers and refugees to express, as it caused them to visualize how their personal networks have “detonated” due to the Syrian war or to their family members living apart from each other. In all cases, interviewees were surprised by the results and the image of their networks. Some took photos of the laptop screen to have an image of their networks, and the rest were also asked if they wanted to have such an image. Most of them agreed and were sent them by email afterwards. The qualitative analysis of personal networks was carried out as material complementary to the interviews and fieldnotes, all of them being codified with RQDA for coding and subjected to qualitative analyses.

My first aim in using Vennmaker was to conduct a qualitative analysis of how working in Aidland shapes aid-workers’ personal networks. However, the number of elicited alters (approximately 300) allowed me to map, albeit incompletely, the social structure of international aid in Gaziantep at the time, a sector that employs only a few thousand people. No such quantitative analysis has been conducted before, which may facilitate a mixed-methods form of analysis. The plan is to conduct the analysis and then repeat the interviews ten years after the first interview, in 2025, to make a longitudinal analysis. I asked interviewees about this possibility, and all were open to doing it, although I would not expect all of them to agree to participate in a second wave. Moreover, as they would now have previous experience of constructing networks in Vennmaker, it is to be expected that they could do this by remotely sharing my screen, which might facilitate remote online interviews.

To provide a true picture of the interview while it is ongoing, the Vennmaker images below (Figure 2) are exactly those that interviewees visualized at the end of the interviews on their personal networks. They have not been edited to add extra information or to make the material more attractive for publication; only their names have been anonymized. As can be seen in the images below, the three categories of aid-worker had very different networks. The first network corresponds to an expat aid-worker who is well connected with friends (green) and co-workers (purple) both locally (inner circle) and internationally (outside circle), although she no longer has any connection with her family after many years working abroad. The second example shows the network of a Turkish aid-worker who lives with his close family and is already well connected locally with co-workers and friends. After three years working in this sector, he is starting to have contacts abroad, which motivates him to pursue a career as an international aid-worker in the future. The third network is of a Syrian aid-worker and refugee who fled Syria and started to work in the aid sector. Most of his family and friends are living in Syria or just fled as refugees to Europe, and he has only one close friend in the city, a co-worker.

Figure 2:

In sum, the use of Vennmaker in qualitative analysis was a key tool for the emergence of research findings, and it clearly revealed how aid-work structures personal networks. For example, most of the Syrian aid-workers had suffered what they called the “explosion” of their kinship and friendship networks due the Syrian war, and Aidland offered few but promising contacts to go ahead and make it their career. However, for international aid-workers, being far from their families and friends is common. Instead they can rapidly construct new relations within expat aid bubbles, which is part of the networking business to get good jobs at the international level, which are best paid, have more status and entail less risk (see more detailed research findings: Fradejas-García, 2019; Fradejas-García and Mülli, 2019).

This second study was led by the second author of this article, Francisca Ortiz Ruiz. It was conducted for her PhD thesis, and the main focus was older people’s social-support networks. The main objective was to understand how the elderly in Santiago, Chile, deal with their pensions through their social-support networks. Some of the results have been already published (Ortiz, 2021; Ortiz and Bellotti, 2021). In this paper, I, as second author, will concentrate only on those stages in the research where Vennmaker was used.

As already noted, the study was conducted in Santiago, Chile, South America. The research used a mixed-methods approach involving different tools of data collection and analysis. The fieldwork was conducted between October 2019 and January 2020. During that time, the life histories of thirty elderly people and their social-support networks were collected. The sample was divided between fifteen older women and fifteen older men, who were also divided by socioeconomic level: ten were from a higher-class district, twenty from a middle-class district, and ten from a lower-class district.

After the life history of each person had been constructed, another interview was held, sometimes two more, to complete the name generator. In the subsequent interview(s), Vennmaker was used. During each of these conversations, I as the researcher have the computer with me. Sometimes people make jokes about the fact that I could actually enter all their answers in a computer, which makes them curious. Some ask me to show them the program. I show them the questions, the program and the resulting network. Sometimes they are interested, which declines a little with the passing of time, especially after all the questions have been asked about their alters’ characteristics. Using the computer, I ask all the questions that refer to their networks. I ask them to nominate people who give them seven dimensions of social support, then to nominate people to whom they give those same dimensions. Finally, I ask many questions about the characteristics of those relationships and their alters.

After each interview, I come back to them with a visualization of the network to ask their opinions about it. Also, I give them an opportunity to give me their thoughts, say whether they agree or not, and point out what they would change to make it clear. Of the thirty people I interviewed, 21 agreed with the visualization and did not want to make any alterations. First, they seem surprised at the results, commenting on the colors and expressing curiosity about their meanings. Secondly, they seem satisfied by the results. Of the other nine people, they made suggestions that changed the organization of the network. Most of these suggestions can be divided into three groups: moving one alter to a position closer to ego, adding more dimensions given by the partner, and adding a link between alters that was not mentioned before. Having an opportunity to talk to the interviewees about the visualization of their networks benefits the research greatly, allowing a more detailed and accurate representation of their social-support networks. This part of the research serves to validate the data collected.

During the fieldwork, using the computer was not always easy. In interviews conducted in a lower-class district, carrying a computer around entailed some risk, especially for me, as it brought me unnecessary attention, though after entering the house of the interviewee that changed. Also, taking the computer out of the bag tended to create a gulf (a feeling of distance) between the interviewee and myself. The solution to that was a lot of eye contact with the informant and paying them a lot of attention. Another tactic that helped break the ice was conducting the life-history interview(s) before constructing the networks. In these ways, the interviewee already feels confident with me, making the gulf easier to bridge.

After the fieldwork, the first task was to start transcribing the life-history interviews. Every time I completed one case, I exported the visualization of the interviewee’s social-support network to the desk’s PC outside Vennmaker (one per direction of the relationship) and the Excel document with all the answers in the interviewee’s survey. Afterwards, I ran all the analyses using UCINET and R Studio. As I asked for the social-support network for each of the most important events in their lives, I had longitudinal data that were interesting from a comparative perspective. The visualization of the networks in Vennmaker allowed me to compare easily all the events and networks in the interviewee’s life. This was possible because the software uses one standardized image, which helps visualize different networks as another layer of the same image. This visualization was also helpful not only for purposes of comparison, but also in making presentations and as an outcome of the research. I used the template with concentric circles, making it more user-friendly, as can be seen in Figure 3.

Figure 3:

The main results of this research show how older people construct their social support networks during their life-course and how the structural aspects intervene through them, thereby increasing some of the basic inequalities that intersected with others, like gender, age, and socioeconomical level. Furthermore, it shows the social support networks as multilevel, multidimensional and longitudinal. Among the main results was the fact that the visualization of each represented network (as in Figure 3) helped the actors’ understanding and enabled them to give more support dimensions to ego, as well as the other way around. These graphs were complemented by the narratives behind each life history.