Agricultural research is a data-intensive activity. Agricultural researchers use various channels to share their data. In Tanzania, much research has been conducted on communication channels used to share information, and knowledge with researchers, smallholder farmers, and other agricultural stakeholders (Barakabitze et al., 2015; Mtega, 2021; Mtega & Ngoepe, 2019; Ndimbwa et al., 2019, 2021). It can be noted, however, that most of these studies have focused on channels for sharing agricultural information, and knowledge, while none of them focuses on channels for sharing agricultural research data in particular, despite the large amount of research data that exists in agricultural research institutions (Tanzania Agriculture Research Institute, 2019). While there is consequently much knowledge on the effectiveness of channels of various forms in sharing agricultural information and knowledge, there is undoubtedly a lot of untapped potential in terms of how these channels can be equally used to enhance agricultural research data sharing among agricultural researchers. This study was, therefore, set up to fill this knowledge gap by investigating the existing channels in agricultural institutions, their potential applicability in enhancing agricultural research data sharing, and their extent of usage to gauge whether they may be effective and efficient in sharing agricultural research data, and factors that influence researchers in channel choice in Tanzania.

The main goal of this research was to investigate the communication channels that enhance data sharing among agriculture researchers in Tanzania.

Data is defined as a visual representation of something. Data encompasses observation or information that is obtained from a research project, including experimental samples, technologies, materials, products, and procedures (Kirub, 2016). In this current study, data can be raw or analyzed, quantitative or qualitative, and is collected by researchers and can be shared with other researchers for use or reuse. Research data are data generated through research activities; such data can be in the form of textual, quantitative, and qualitative (Tripathi et al., 2017), and may be in the form of images, recordings, musical compositions, verbal communications, simulations, experimental readings, and code. In agriculture, research data contain crop breeding data, baseline survey data, genomic data, remote sensing data, and geographical information systems (GIS) data (Kirub, 2016). Agricultural research data can be obtained through electronic objects and tools such as sensors, laboratory tools, voice interviews, and online surveys (Boté & Termens, 2019; Zhao & Wang, 2015). Such important data are required to be shared through appropriate channels simply because they are related to agricultural scientific research. Data sharing in this case is defined as a process of making data available to others, although not necessarily by depositing it in repositories (Thoegersen & Borlund, 2021). Moreover, data sharing can take place in three methods; data is deposited in a repository, sent upon special request, and published as supplementary materials on the journal website (Gray et al., 2018; Williams, 2022). Likewise, data can be shared in the form of manuscripts, including pre-print products generated by researchers (Jeng et al., 2016).

Communication channels are media that facilitate the transfer of messages from sender to receiver (Mtega, 2021). In this case, therefore, communication channels are also data-sharing channels. However, not all channels work in all environments, and the proper selection of the channel is therefore important to attain effective data delivery. Previous studies have examined different communication channels used for sharing scientific data. In a study that investigated how communication channels are used to increase awareness among researchers, Zaira (2012) observes that the online channels preferred by researchers included blogs, Wiki, Facebook, Linkedin, and Twitter. The study also reported that the offline channels preferred by researchers in creating awareness among them include printed materials, magazines, and frequent departmental meetings. Researchers from developed nations, in particular, have been sharing and accessing research data through channels that are web-based, such as data repositories. Several meta-analysis and survey studies have revealed that researchers have been sharing research data by depositing it into data repositories (Cooper et al., 2019; Feder et al., 2015; Drakos et al., 2015; Suhr et al., 2020; Thoegersen & Borlund, 2021).

When the data file is shared in repositories, they can be preserved for future use and reuse. The presence of metadata standards and agreements, among other things, facilitated researchers and research institutions in sharing and accessing data from data repositories. In line with this, through reviewing selected publications by Crop Sciences faculty, it was found that data sharing takes place in three methods, namely depositing data into repositories, sending data to researchers upon request, and supplementary materials on journal websites (Williams, 2022). Findings indicate that data sharing in articles appeared in 50 different journals. However, studies from low-income countries reveal that data sharing is not through data repositories; a majority share data through email on request, within published papers, collaborative initiatives, and through research unit servers (Bull et al., 2015; Katabalwa et al., 2021; Koopman & de Jager, 2016). It is clear that the presence of a data-sharing channel is important to enhance data to be accessible for use by others. Most of the existing studies from developed nations have revealed the presence of formal data sharing channels such as data repositories as compared to most developing countries, where there are no existing data repositories to facilitate data sharing. This therefore called for a detailed study to investigate the current data-sharing channels existing and used by agricultural researchers in less developed countries like Tanzania.

The selection of proper channels can ensure the message, or in this case, the data reaches the intended audience. The criteria to guide channel selection was studied in detail. In the study that focused on literature review of 36 papers to identify methodologies and knowledge gaps in channel choices by Madsen and Kraemmergaard (2015), the results revealed that the criteria that guide researchers on choices of communication channels include: channel characteristics (interactions, ease of use, usefulness), task characteristics (type of task, adequacy of information), personnel characteristics including social demographics (age, gender, race, education, and income), situational constraints (channel availability, price, and distance to channel) and satisfaction with services offered by the channel. All of these were said to motivate individuals in making their channel selections.

Another study that indicates factors influencing choices of channels is the one by Fichman and Hara (2007), the findings show that channel choice is determined by six criteria: the task at hand, physical proximity, social proximity, sender and receiver accessibility to use the channel, individual preferences about the channel, and the initial channel used. Furthermore, the study that examined the knowledge sharing and channel choice in The New Way of Working (NWOW) by Kok et al. (2016), has reported that employees’ choices of channels for knowledge sharing were based on factors such as usability, experience with the channel, effectiveness, and convenience. For example, some employees preferred to use multiple channels when sharing knowledge of sensitive information, while some other employees were still working in a more formal and traditional non-New Way of Working (NWOW) manner, using fewer channels such as face-to-face communication.

Most published literature on data-sharing channels focuses on research in high-income settings. Most existing studies have revealed that researchers are motivated to deposit their data due to the existing formal data repositories with well-defined metadata standards, file formats, policies, and other agreed arrangements (Cooper et al., 2019; Elsayed, 2018; Federer et al., 2015; Jeng et al., 2016; Thoegersen & Borhund, 2021; Zhao & Wang, 2015). However, although the literature is limited, literature concerning data-sharing channels from low and middle-income settings has exposed several aspects of data sharing (Bull et al., 2015; Katabalwa et al., 2021; Koopman & de Jager, 2016). Earlier literature reveals that the best data-sharing channels that have been operating include the sending of data upon request to researchers who trust each other and work collaboratively, publishing in the papers, and depositing data in repositories. It is interesting to see how Tanzania fits into the general landscape of data sharing in the contexts other developing countries.

This study adopted the Concentric Layered Model. This model in prior studies was used to explain the factors influencing employees in channel choices when sharing knowledge in work teams and in a New Way of Working (NWOW) (Kok et al., 2016; Snyder & Lee–Partridge, 2013). The previous four-level model comprised four concentric circles, each consisting of a group of factors that influence the choice of a particular channel. The first or inner layer (primary layer) includes the type of information, sender efficacy level, and preferences. The secondary level includes richness, ability to keep records, reliability, conveyance of the message, feedback, trust, and preference. The third (tertiary) layer includes embedded practices, the degree of translation and transformation, and the key boundaries. The fourth (quaternary) layer includes the institutional culture, policies, and legal requirements (Snyder & Lee–Partridge, 2013).

In this study, the Concentric Layered Model was modified to take into account the factors influencing the choice of channels in the tertiary and quaternary layers. Based on the modified model (Figure 1 below), the factors that influence researchers’ choice of channel for sharing data include the primary layer (data characteristics), secondary layer (timely delivery of data, convenience, and availability of data sharing platforms), tertiary layer – the current data sharing practice (submission of the data file to institutions or funders for permanent records) and quaternary layer (funding agency policies). Based on Figure 1, the current study findings indicate that some factors concur with the factors in the previous model and the model, therefore, explaining very well the factors influencing the choice of data-sharing channel and thus the model fits for this study.

The conceptual framework in Figure 2 was developed to explain the linkages between independent, intervening, and dependent variables. The intervening variables include researchers or research institutions that generate, and use research data. When researchers want to share their data, they are guided by some factors in the selection of the channels (independent variables). Data sharing will take place when data files are shared through the selected channels.

Both qualitative and quantitative data were collected and analysed. The present study used content analysis to analyse qualitative data obtained from open-ended questions in the questionnaire, key informant interviews, and FGDs. The quantitative analysis was accomplished by using Statistical Package for the Social Sciences (SPSS) version 22. In the present study, the descriptive variables for the channels and factors were summarized using a table with the frequencies and percentages. Regarding ethical considerations, the researcher avoided any risk of harm during data collection and ensured the right to privacy and informed consent before engaging any respondent in the study (Mwinami et al., in press).

The demographic characteristics of respondents are presented in Figure 4. Results indicate that a significant percentage (67%) of the researchers who participated in this study were male. This implies that the majority of researchers in the surveyed research institutions were male. This trend suggests that females were in a small proportion (33%) of those employed and therefore participating in agricultural research in the surveyed institutions. The findings further show that most (69%) of the respondents were aged between 31 to 50; the remainder (27%) was aged 50 and above. This implies that significant proportions of researchers were at an active age and engaged in agricultural research. Furthermore, the results also indicate that a large percentage of the respondent (82%) are researchers who had already received their Master’s degree and above, while a few (18%) had only completed a Bachelor’s degree. This implies that most of the researchers in the surveyed area had a higher level of education that enabled them to conduct research. The results further indicate that a large percentage of researchers (95.3%) had work experience of more than (6) years, while only 10% had work experience of between 1 and 5 years. This implies that the majority of the researchers had a reasonable amount of work experience in conducting research activities and the junior researchers would be gaining experience from seniors (Mwinami et al., in press).

The findings in Figure 5 indicate the researchers’ academic field of specialization. From the findings, the largest group (45%) of researchers specialized in botany, crop sciences, and horticulture. An additional 7.8% of the respondents specialized in Animal Science, Aquaculture, and Range Management. This implies that researchers who participated in this study specialized in various research fields (Figure 5).

These findings in Table 1 below reveal that respondents made the heaviest use (77.9% and above) of non-mediated channels, including face-to-face dialogue (meetings, discussions, and seminars) and print media, such as technical reports, etc. These results imply that researchers preferred to use channels that give the chance for face-to-face dialogue, and interaction, and provide timely feedback among researchers. In addition to that, the study established that researchers mostly preferred channels that could deliver data promptly. These findings, concurring with the study by Shen (2017), show clearly that interpersonal discussion was used to help in providing data context, declaring assumptions, clarifying jargon, elaborating data structures, and identifying points of interest for effective data sharing. Similarly, findings are related to previous studies which reveal that face-to-face meetings and conversations are the best for research data sharing among researchers (Howland et al., 2015; Shen, 2017). It has also been reported that researchers in universities and research institutions have used formal seminars, workshops, mobile phones, and open discussions as the most preferred and available channels to exchange knowledge and ideas with colleagues (Delserone & Dinkelman, 2016; Kigatiira et al., 2018).

The findings in Table 1 show that respondents indicated the strongest preferences for mediated channels such as email, and publication on online journal platforms (74.5% and 72.5%, respectively). These findings imply that preference for such channels is a result of the timely delivery of data due to the available data in electronic format. Researchers had the habit of sharing their data files with fellow researchers through their emails. Researchers’ preference to share data files through online journal platforms was motivated by the opportunity they got to increase their data visibility. Moreover, in agriculture research institutions/universities (in this case, at SUA), publishing articles with data in the journal avenue allows for advancement and promotions in the academic ranks. These findings are in line with a previous study by Zhang et al. (2016), which informed that emails were among the best channels used by researchers in sharing data. Concerning online journal publications, other previous studies have also reported the use of this channel (Shen, 2017; Tedersoo et al., 2021).

The study results show that some mediated channels were least used by researchers. These included websites, social media platforms, institutional repositories, and funding agency databases. This underutilization of mediated data sharing channels might have been caused by the absence of data repositories among research institutions, and a lack of awareness among researchers of using such channels. Most of the research institutions in Tanzania do not have well-developed ICT infrastructures, especially data repositories; this might have hindered researchers sharing their agriculture research data through data repositories.

Findings in Table 1, showing multiple responses, indicate that both mediated and non-mediated channels were used in data sharing at varying levels. In support of this argument, one of the key interviewees at the TARI Selian centre commented:

Another key interviewee at TARI Katrin centre commented that:

Furthermore, another key interviewee on channels for data sharing at SUA commented that:

In the same context, through focus group discussions held at Ilonga and Mikocheni, respondents reported the following:

The second objective of this study was to determine the extent of channel usage to gauge whether they may be effective and efficient in sharing agricultural research data. Through cross-tabulation, the usage of each channel is shown for each of the research institutions. The findings in Table 2 below show that SUA had reported the use of all the existing data-sharing channels, and the extent of usage is higher than at the other research institutions investigated. Of all the data sharing channels that exist within the studied institutions, meetings, conferences, and the organizations’ accessible website had a high statistical significance difference at a 1% level of significance (p = 0.005 and 0.001, respectively). Moreover, workshops, telephone-mediated conversations, and print media such as institutional publications, including technical reports that carry data had a statistically significant difference at a 5% level of significance (p = 0.049, 0.045, 0.045, and 0.025 respectively). These findings imply that these channels or avenues were preferred over other channels in the studied area. Agricultural researchers at SUA and TARI centres always work in teams; this explains the role of meetings, conferences, telephone conversations, project reports, and workshops for data sharing. Findings show that there were variations in the existence of channels and the extent to which such channels were used, with SUA showing the primacy in all the channels used for sharing agricultural research data. This might be due to SUA having better ICT infrastructure compared to the other research institutions. SUA also has a large number of researchers from different research fields that would use a multiplicity of channels in sharing data files.

The third objective of this study was to examine factors influencing the choices of data-sharing channels as reflected in the Concentric Layered Model in Figure 1. Findings in Table 3 indicate that more than 50% of respondents shown strongly agree that timely delivery, the cost of the channel, and the convenience of the channel were among the factors influencing researchers’ choice of channels for data sharing. Timely delivery of data is among the motivating factors for researchers to use avenues such as face-to-face dialogue that provide timely delivery of data. For example, researchers preferred to use face-to-face dialogue (meetings, seminars, and discussions) because they were among the avenues which provide a chance for the timely exchange of data files, technologies, and innovations. With the advancement of technology, the weather forecast has been improved through the use of the mobile phone in data sharing. The use of CAMALIOT Android apps that uses a cell phone to support scientific research has been used in weather data collection and exchange. The app facilitates citizen and researchers to access and collect raw GPS satellite data from individuals with cell phones thus a cell phone act as a channel for transferring data to servers for permanent preservation for future use (See et al., 2022).

The convenience of the channel was observed to encourage researchers to choose a channel. The meetings convened in agricultural research institutions are convenient for researchers. Regular technical meetings for project evaluation (seminars and discussions held in research institutions) are convenient to most researchers to the extent that they motivate researchers in sharing data related to technologies or innovations. Likewise, almost every researcher owns a cell phone which makes it easy to call fellow researchers participating on the same project team or collaborators who are found in another institution just for sharing experiences in particular research projects. It is during such conversations that research data files are shared and hence exchanged using such channels. Elia (2018) recommended the use of mobile phone devices in meteorological stations to facilitate the timely collection and recording of weather raw data. The phone is a channel that can help to make data to be timely shared. In line with this, a study by Msemo et al. (2021) reveals that cell phone has been very useful to disseminate and receive weather data and warning information from weather stations located in different regions in Tanzania.

Data is characterised by its need for permanent records, availability of ICT infrastructures, and project funders’ policies; yet these are also among the factors indicating less than 50% slightly in motivating researchers in their choice of channel. The availability of ICT infrastructure such as databases, and repositories that are connected by the internet motivates researchers to share their data by depositing their data files in funders’ databases at the end of the project. Other data files have been published as supplementary materials in journal e-platforms. The need for permanent records for data obtained from research activities is among the factors that motivate researchers to choose a channel or avenue to share data. For example, by sending their data files to the funder’s databases and research institutions, the researchers’ main intention is to have permanent records for the data obtained during research projects. Therefore, the possession of data repositories or databases by the project funder motivates researchers to share their data through the available avenue. These findings are in line with the prior studies which have revealed that the availability of funding agencies’ databases was observed to facilitate researchers to select funder’s databases as their channel to share their data (Bezuidenhout & Chakauya, 2018; Dodgson et al., 2021; Suhr et al., 2020).