Introduction

Mentoring has always been applied as a personal empowerment as well as a developmental tool that addresses the main concerns of the mentees through the provision of knowledge and advises that is critical in boosting the existing competency and morale [2]. The difference between the e-mentoring and the traditional mentoring strategies is that the two strategies apply different approaches to the mentor-mentee relationships. For instance, E- mentoring applies the computer-based and mutually beneficial mechanisms that deliver advisory, learning, modelling, promotional, and encouraging programs for the mentees [2]. The mentors and mentees have online interaction platforms in e-mentoring, unlike in the traditional means whereby the mentors and mentees are physically located at one place [1]. Internet and the computers pay major role in facilitating the online mentoring process, and the mentor-mentee relationship is virtually maintained within the online platform.

It can be noted that, with the advancement of the Open Source Software Project (OSSP), there has been a great deal of importance of the e-mentoring system software that would enable the application of the e-mentoring programs on online communities. However, there have been very few empirical studies on the e-mentoring process [3]. It is quite helpful for one to determine how the e-mentoring would be rolled-out within online Communities through the use of internet. For instance, this would necessitate prior training of the community members, ensure effective source of Internet, identification of a number of setbacks that works against the set program and support activities that would necessitate pro-active intervention.

In this paper, the aim is to understand mentoring activities, key issues and structure of mentoring in Stack Overflow. This study aimed at exploring the following research questions:

RQ1: What are the mentoring activities in Stack Overflow? RQ2: How are mentoring taking place in Stack Overflow?

Mixed-methods approach was used. First, it started by collecting 400 randomly threads in first quarter of 2018. Based on mentoring activities models [4, 17-18], each code was categorized and classified to find out mentoring activities. In this report, the aim was also to understand how mentoring taking place in Stack Overflow from 390 moderators answers using qualitative data analysis.

Furthermore, discuss that this tactic could help us realize the difficulties that mentees and mentors face in this regard and identify the required support type and how they get support

Related Work

Various studies with differing objectives have been conducted on e-mentoring platforms such as Stack Overflow. One such research was done by Ford. The research was about finding out women participation in stack overflow. The study focused on the various barrier’s females face when interacting with online communities. The findings were that, only about 5.8% of the total population that takes part actively in stack overflow were women. The study ranked five barriers higher on women than their male counterparts. It was found that females felt overwhelmed due to the large number of users they encountered. The reason for their low participation were deduced from the findings and it was clear that most of the women fear negative feedback in any interaction. As such, since the stack overflow platforms are open for any kind of feedback they fear that the responses may not be in their favour.

Another reason was their inferior nature. Their inferiority complex makes them feel that they are unable and lack any expertise to do anything on the web. Their limited awareness of the features offered on these sites which greatly inhibited their ability to find a mentor (Ford, 4).

This makes them feel demotivated in e-mentoring activities. Another research by Ford on the deployment of just-in-time mentorship program to Stack Overflow. In this study, novice participants were redirected to an on-site Help Room where experienced mentors helped them redraft their questions before submitting them to the Q&A forum. It was found that answers to the mentored questions increased by more than 50% when compared to the non-mentored questions (Ford, 3). Another study conducted on R software community developers revealed that there were a number of prolific contributors who were responsible for providing most of the answers on the Q&A platform (Zagalsky, 5). This is a group of expert programmers from different fields and handling different programming languages. They post quality reliable answer in the platforms making everything easy foe the moderators.

Further research conducted by Stanford University indicate that there was a marked shift towards answering questions which was community-driven (several people offering answers to the same question) and whose end product would benefit a broad number of audiences instead of primarily focusing on the question asker (Anderson, 4). Furthermore, recent research indicates that the emotional style of a technical question influences the probability of one obtaining a satisfying answer for a question asked on Q&A online communities such as Stack Overflow (Novielli, 6).

Research method

This section describes the different methods used in conducting the study and analysing the results of the study. It also clarifies the characteristics the collected data. Both qualitative analysis and Activities modelling were the approaches used in analysing the data. In order to classify and categorize the data, the whole data set was subjected to qualitative analysis. A qualitative analysis was also conducted to a subset of e-mentoring activities.

1. 1. Dataset:

The dataset used in the study originated from the Stack Overflow platform. Stack Overflow offers a Q&A platform as well as other tools that enable users to acquire and grow excellent programming capabilities. The data was acquired from the data dump that Stack overflow releases every three months. Only the top vote questions and answers with comments were chosen for the study.

The primary source of information (data) was basically obtained from the question and answers features presented during the mentoring session. We randomly sampled and coded the type of support needed from questions, type of activity/support provided from top vote answers and their comments, type of developer who ask questions (mentee) and those who provided top vote answers (mentor). This process was continued until thematic saturation was reached (no new themes were identified and no issues were raised regarding existing categories, which amounted to 400 threads) [20].

Moreover, the simple formula that was used for the determination of the sample size is as follows; n= (384.16p)/ (p+383.16) [7]. The margin of error was at 5% at 95% confidence level. While it is required that the total sample size used to be approximately 384, the current study collected around 400 responses from the mentoring activities that were undertaken in the Stack overflow. We randomly sampled and iteratively coded questions to characterize the type of discussions that occurred. Every question provided at least one answer that was linked to different kinds of diverse mentoring activities. In this regard, the different mentoring activities were quite evident in the answers that were provided, especially the answers provided by professionals [10]. In this case, the main focus was on the top answer given as well as the comments provided. The comments were analysed carefully through keen examination of the kind of support that was required by the users in regard to what was availed to them. In reference to the provisions of the ground theory and activity models, it was possible to categorize the type of mentoring activities within the Stack Overflow platforms.

1. 2. Activity modelling:

Looking at the activities required in the e-mentoring process within the online communities, major focus is placed on the student-centered means of mentoring like the Resource, Activity, Support and Evaluation (RASE), which is a course model design that is characterized by Questions and Answers. This model is made up of 4 notable components as mentioned above (the RASE) [4]. For the purposes of the current study, focus is basically made on the activity section. This section is based on the questions and answers, features that are found in the Stack Overflow. These are mainly student-centered because the main focus is on what the student is expected to do to learn as opposed to what the students will remember. Also, the resources are considered as the tools that are within the student’s possession.

This process also involves teachers who are in the field of programming for the sake of offering technical support [5]. This process also helps the students develop new literacy as they participate in the process. The main activities carried routinely on the stack overflow always include; there is always the initial preliminary inquiry that is always meant to clear all the doubts among the students with regard to the programming process [2]. In this first stage, experts in the field of programming would respond to the inquiries made. This would be followed closely by the problem-solving passé whereby the learners would be challenged by diverse tasks presented to them to perform in line with their areas of interest. In this process, the learners get mentored. The learners would be exposed to varied programming tasks that would equip them with the most relevant programming knowledge, for instance on hypothesis testing and experimental designs.

The other mentoring model is the GROW model that provides some of the simplest plans of structuring coaching and mentoring. The GROW model presents the Goal, the current Reality, existing Options and Will/Way forward. This model brings about a view of how to plan the entire process of mentoring [17]. The model assumes that the coach is not the main expert in the situation presented by the user, and therefore, the coach acts as a facilitator and the person who helps mentee make the best choices. In this model, one needs to establish the goal, which must be SMART (specific, measurable, achievable, realistic and time bound).

Examine the existing realities and explore all the available options. The final step would be to establish the will, which includes the final idea that will achieve the set goal.

The other model is the Collegial coaching/mentoring. This model consists of three basic parts; the pre-conference, observation and post-conference. In the pre-conference, items of observations are clearly mapped and outlined [18]. This helps in making the entire process focused to the main objectives set. The peer mentor then observes the mentor’s sessions.

Finally, there is the post conferencing whereby outcomes are discussed and evaluations made. Instances of collaborative and consultative training would also be quite critical in the performance of the activities in the Stack Overflow platforms [19]. It is also provided that the Stack Overflow is one good example of e-mentoring program that allows for continuous interaction through questions and answers.

3.3. Qualitative analysis

Qualitative analysis of a subset of mentoring activities was performed. The analysis was performed with the following objectives;

1. To identify the mentoring activities for the top voted answers and comments,
2. To find out the kind of support requested based on a question.
3. To identify both the mentor and mentee.

Furthermore, a thematic analysis was performed on the content of all e-mentoring activities derived from selected threads. These activities were then imported into excel and coded manually. The activities were further organized in datasets and the emerging codes were then categorized into themes.

4. Findings

We found a total of 2481 total posts from 31 different categories that we organized into 10 high level mentoring activities as showed below:

1. 1. E-mentoring Activities on Stack overflow:

The findings of each e-monitoring activity are analysed as shown in the tables below.

Table 1.

Add content for enriching answers.

Table 2.

Explain and clarify answers.

Table 3.

Help understand debug and fix them.

Table 4.

Share best practices.

Table 5.

Offer suggestion for improvement.

Table 6.

Help to organize meeting or communication.

Table 7.

Improve quality of communicated information.

Table 8.

Manage communication information.

Table 9.

Motivate and encourage to continue.

Table 10.

Generate critique and feedback.

Qualitative results

Stack Overflow is presented as a good example of an e-mentoring program that allows for online active engagement through the use of questions and answers, whereby the user asks questions related to computer programming and get response from the mentees. The answers received are voted and the best responses would be considered as being the right answer [6]. In this manner, the user of the Stack Overflow model greatly benefits and steers the mentees towards becoming successful within the programming industry.

1. 2. Stack Overflow organization role

Stack Overflow platform offers a vast service through the feature of Q&A platform and the tools availed to the users online to help them grow and acquire exemplary skills in the field of the programming. We have various types of users on the Stack Overflow platform and they include; first, there are those who ask question and they are mostly the students or individual new to the world of programming. Secondly, there are expert who have a vast knowledge about coding or programming in general and they are the one who assist and mentor the new users or students having difficulties relating to programming and lastly there are the moderators who oversee the whole mentoring process to make sure mentors are guiding their mentees in the right direction and asses the progress with time.

The main role of the mentor is to volunteer and establish an enduring personal commitment to a personal relationship with the main aim of nurturing individuals who need their services and impact their lives in a positive way through the online platform [14]. On the other hand, the role of the mentee is to maintain the personal relationship established by the mentor and in a succinct way describe to the mentor the areas on interest to enable the mentor know how to handle and nurture that mentee accordingly. Since these services are offered online, e- mentoring hugely rely on the trust between the mentor and mentee.

According to the Stack Overflow platform, those who take the role of the mentee are the students or individual who want venture to the programming field and they have limited knowledge about it. The role of the mentor is given to the individuals who have viable knowledge in the programming and they are considered as the experts in that field. For instance, based on the previous study [15], Experience and prolific member have the role to make sure the platform stays clean and with viable content only [15]. Other members serve as the moderators and help in bridging between different groups of people and nurturing the community at large.

We also have caretakers in the Stack Overflow community who over watch the entire system and help with answering the question hence motivating other experts to stay active and be devoted to the community [16]. Therefore, the study considered that all moderators on Stack Overflow are a mentor and other members who ask or request support is a mentee.

1. 3. Qualitative Analysis

On analysing the above results, we find that Stack Overflow is more than just a Question and Answer platform. The activities that are carried out on this site can be categorized into 10 high level mentoring activities as indicated above. The ‘general critique and feedback’ category has the highest number of posts totalling to 490. Many posts on this category (81%) were made by users (both mentor and mentee) who wanted to vote on an answer, a question or a comment. This helped the participants to research before criticizing any answer. Such feedback is critical on a Q&A platform like Stack Overflow since it enables the online community to discuss issues in depth and therefore ensuring that the best answer is found. ‘To motivate and encourage users to continue posting’ is the category with the second greatest number of posts at 394. About 97% of the posts on this category were made by users who wanted to provide compliments, offer thanks, opinions and give positive criticism. To the mentee, their relationship with the mentors could grow as a result of the positive feedbacks and compliments. Some opinions could be educative to both However, only a paltry 12 posts were made by users who wanted to put questions on hold in order to encourage improvements or edits on the answers given. Such users are usually mentors and it is through their contribution that mentees (those asking questions) are able to get the best answers possible.

Postings that aimed at ‘adding content for improving answers’ were third highest in number (368). 76% of these posts were made by users who added code examples to clarify and enrich the answers given. Code example additions are made mostly by mentors who are experts in their fields. Making an improvement on the answers could ensure that the quality of the knowledge acquired from the platforms was high and satisfactory Another mentoring activity with high postings is the activity that aims at ‘improving the quality of communicated information’ with 310 postings. 68% of postings in this category (mentoring activity) edited a question or answer by either rewording it or improving the vocabulary used. The rest of the postings (32%) targeted at correcting typos and coding mistakes. It is expected that most postings in a Q&A platform such as Stack Overflow should be about answering questions or offering improvements to answers provided, however, this is not the case. The analyses of the results indicate that only 210 postings (a paltry8% of the total postings) are about ‘understanding, debugging and fixing code errors’ (answering questions from mentees). Another 186 postings (7% of the total postings) are aimed at ‘explaining and clarifying answers. An even smaller number of postings (55 – about 3% of total postings) are made with aim of ‘suggesting improvements to the answers provided’. The suggested improvements could open up the eyes of the mentees in case they were to consider the suggestions viable.

Other mentoring activities that were carried out on Stack Overflow Q&A platform include posts on ‘how to help organize communications or meetings’ which had 112 postings.95% of these postings were on collaborative discussion about issues such as answers given and positive experiences. ‘Sharing best practices ‘is another mentoring activity that was identified in the research findings with 179 postings. Most of the postings in this activity involved sharing a blog or article. ‘Managing communication information’ was also identified as a major mentoring activity and had 177 postings. 80% of the postings in this activity involved creating a new tag to a question. The activity here could bring the sense of having a plan and incorporating the right tools in the platforms.

An analysis of the results also provides an insight on the characteristics of the mentors and mentees on this Q&A platform. The moderators of the Q&A platform make up the greatest number of mentors with 44%. Full stack developers are second with 33%. These two groups make up the majority of mentors found on Stack Overflow Q&A platform. It is worthy to note that a good number of students (7% of the mentors) are also involved in providing mentorship to users. Contrary to expectations, only a handful of researchers or educators are involved in mentorship activities. The analysis shows their numbers make up only 1% of those offering mentorship on this Q&A platform.

Users with only one answer make up the majority of mentees on Stack Overflow Q&A platform with half of all mentees falling under this category. Beginner software developers make up the category with the second greatest number of mentees with 39 %. These two categories of users make up the majority of the mentees found on this Q&A platform. Contrary to expectations, students make up only about 9% of all the mentees. It is expected that students should make up a majority of the mentees but that is not the case. Furthermore, it is worthy to note that even experienced software developers with 3% and moderators with 2%also make up a significant number of mentees on the platform.

A study conducted on R software community developers brought it clear that that there were a number of prolific members who were responsible for providing most of the answers on the platform (Zagalsky, 5). This is a group of expert programmers handling different programming languages. They post quality reliable answers to questions posted in the platforms making everything easy foe the moderators.

Limitations

One of the limitations of the study is that most data used in analysis was limited to particular answers that were voted the best and those that had comments. Analysis were based on findings from this data without considering the criteria that was used to vote an answer right or wrong. Therefore, the data lacked diversity in the sense that conclusive results were made without considering there could be difference in the data if different approaches of collecting it were used. As such, the data gave a narrow range of understanding to the researcher. This could bring the risk of drawing erroneous conclusions from the limited information drawn from the narrow scope in case data from other sources like use of questionnaire proven to be different. For example, from the data, only 14 posts provided a tip to the mentee, if the same research could be made through other approaches different result could have been availed and a different conclusion about their help to the mentee could have been made another drawback was that data collected from help rooms could have been compromised due to a flaw in their design. Help rooms were designed to aid mentors and mentees converse on the Q&A Stack Overflow platform but a flaw in design made it difficult to easily distinguish the conversations between mentors and mentees. For example, some mentees may have asked clueless questions that are very basic and could not participate in research or giving feedback to get their problem solved. The number of recorded feedbacks therefore could less and some wrong to draw conclusions from. This lender the thematic analysis not trustworthy. (Ragins, 32).

Another limitation of the study is that different styles of mentors led to some mentees getting more help than others. This necessitated the need to effect measures to control mentorship styles. This could have been made possible by having a uniform style of how the mentees could be helped. However, the effectiveness of these controls cannot be quantified and therefore could have biased the results of the study. Ensuring validity of data (both internal and external) is critical as it determines the validity of any research (Stromei, 42).

Ten activities were actually not enough for the whole platform study. More detailed information like the participation time, number of numbers, their literacy levels and frequency of asking questions on the platform could have been used in the study. This could make a large pool of data from which analysis could be carried and consequently improve the trustworthiness of the final results concerning the platforms

Discussion and Conclusion

1. 1. Key Contributions of the Study

One key contribution of this study is that it identifies and highlights the importance of feedback in e-mentoring activities. Feedback improves the quality of questions asked by mentees hence improving the chances of getting the correct solution. The study also highlights the importance of emotion recognition in social platforms. Q&A platforms such as Stack Overflow will benefit immensely by improving the emotional aspect of their sites and catering for the quality standards of thee information posted as it will help promote the technical aspects the mentees need.

Stack overflow keeps the mentee highly updated on the new ways of handling specific tasks as they get informed through feedbacks and general participation of the members. This has an advantage of being technically knowledgeable in programming. The interaction on the platform creates a virtual relationship between the mentee and the mentor that helps promote their social wellbeing.

Furthermore, the study also underscores how Q&A platforms have evolved in complexity especially on how questions are addressed by multiple experts. This is critical since the best answer will not be based solely on needs of mentees ( the asker of the question). Other considerations such as its lasting value and how many other members of the online programming community will benefit from it are made (Klasen, 63).

The community moderators of the site are able to clean up the information posted on the sites by flagging, editing, closing and even deleting some posts so as to make sure the best is provided to the mentees.

The findings of the study as well as the result based on these findings has helped get an understanding of the Q&A platforms. The findings may be used for further analysis and the results can be compared against results from other mentoring platforms.

1. 2. Opportunities to Further Research

Further research should be conducted on how the use of help rooms can improve e- mentoring activities on Q&A platforms such as Stack Overflow since mentorship programs rely heavily on feedback and this also applies to e-mentorship. This is critical since formative feedback that is derived from help rooms can help mentees improve the quality of their questions hence ensuring they get the best answers. (Burgstahler,70).

Furthermore, additional research should be carried out on the role of emotions on Q&A platforms such as Stack Overflow and their implications. Such research will aid in creating tools with emotional intelligence which help improve user experience and enhance e-mentoring activities (Young, 26). Also the study open up a need of having content clean up tools ad this calls for further research in the programing field to come up with such tool that can delete and edit the unwanted posts in order to improve the information quality and raise the technical aspect of the posts. Moreover, additional research on community processes in question answering is needed. This kind of research will help identify ways in which to improve user experiences and e-mentoring activities on Q&A platforms. For instance, the role played by incentives and competition within online programming communities could be exploration (Ozok, 164).

A research on software’s with artificial intelligence need to be conducted. Such software’s can help reduce the need of having moderators by giving the system generated answers to particular questions. To the mentors, the artificial intelligent software’s can be designed to carry out activities like giving positive critics to the answers so that their quality can be improved. The agents can also be designed to correct the way mentees structure their questions so that wrong answers can be avoided as a result of their design flaw.

A research needs to be conducted on the best tools and software’s that can help analyze the type of data collected in this study. Through the research, a concrete decision of the best data design that can be analyzed through statistical software’s like the R and SPSS software’s can be possible in an easy way.

1. 3. How to Improve Tools/Platforms

Improving the technology that supports e-mentoring activities will greatly aid in ensuring the success of e-mentoring in Stack Overflow. Analysis tools that determine the most popular topics can aid in moderating Stack Overflow website hence improving its efficiency. Once the popular topics are identified then Stack Overflow can create spin-off websites to cater for these topics. Improving emotional interface design will lead to better user engagement and experience. Furthermore, innovative tools with embedded emotional intelligence will also enhance user experience and improve relationships between mentors and mentees (Zachary 60).

Strategic plans need to be set on how, when and where particular questions can be posted and handled. Plans ensure that everything follows order and the result can be predicted. Planning will bring flow in operation and ensure that the Q&A platform becomes a benchmark to other platforms. This has also something to do with creating the best mentor mentee relationship that can help improve the clarity in their conversation to ensure that the mentee’s problems are solved.

Moreover, providing tools that give instant feedback will increase user engagement and improve communication between mentor and mentee. Constant feedback also helps create a bond between mentor and mentee as the mentor is notified the instant the mentee is stuck on a problem and therefore provide solutions at once (Zachary, 60).

0. Recommendations and Future Work

It is recommended that future studies explore other open source repositories and compare results between them based on the quality of answers and the member activeness in the sites.

The quality of the answers can be determined by executing some programs or editing some according to the answers given to see whether they run in the desired way. More samples can be investigated to find potential trends. For instance, investigating the programming language of choice to determine what is most popular amongst e-mentoring platforms could be done.

Furthermore, data from an extended and larger pool should be analysed to determine the extent of reliability to in e-mentoring activities. This will aid in creating lasting solutions to answer quality issues found in e-mentoring platforms. Quality in Q&A platforms are essential in ensuring the successful implementation of e-mentoring programs (Berg, 45). Furthermore, tools developers can use results analysis of this study to fine tune their applications and tools and also decide which technologies to employ in order to improve e-mentoring activities. Q&A platforms such as Stack Overflow require continuous development and improvement to ensure their continued success (Juan, 87).

Work Cited

• Anderson, Ashton, et al. "Discovering value from community activity on focused question answering sites: a case study of stack overflow." Proceedings of the 18th ACM SIGKDD international conference on Knowledge discovery and data mining. ACM, 2012.
• Andriessen, Jerry, Michael Baker, and Dan D. Suthers, eds. Arguing to learn: Confronting cognitions in computer-supported collaborative learning environments. Vol. 1. Springer Science & Business Media, 2013.
• Barua, Anton, Stephen W. Thomas, and Ahmed E. Hassan. "What are developers talking about? an analysis of topics and trends in stack overflow." Empirical Software Engineering 19.3 (2014): 619-654.
• Berg, Gary A., ed. Cases on Online Tutoring, Mentoring, and Educational Services: Practices and Applications: Practices and Applications. IGI Global, 2009.
• Burgstahler, Sheryl. Creating an e-mentoring community: How DO-IT does it, and how you can do it, too. University of Washington DO-IT, 2006.
• Ford, Denae, et al. "Paradise unplugged: Identifying barriers for female participation on stack overflow." Proceedings of the 2016 24th ACM SIGSOFT International Symposium on Foundations of Software Engineering. ACM, 2016.
• Ford, Denae, et al. "We Don't Do That Here: How Collaborative Editing with Mentors Improves Engagement in Social Q&A Communities." Proceedings of the 2018 CHI Conference on Human Factors in Computing Systems. ACM, 2018.

• Juan, Angel A., ed. Collaborative and Distributed E-Research: Innovations in Technologies, Strategies and Applications: Innovations in Technologies, Strategies and Applications. IGI Global, 2012.
• Kelly, Nick, et al. "Beginning teacher support in Australia: Towards an online community to augment current support." Australian Journal of Teacher Education 39.4 (2014): 68-82.
• Klasen, Nadine, and David Clutterbuck. Implementing mentoring schemes. Routledge, 2012.
• Novielli, Nicole, Fabio Calefato, and Filippo Lanubile. "Towards discovering the role of emotions in stack overflow." Proceedings of the 6th international workshop on social software engineering. ACM, 2014.
• OCSC. Online communities and social computing: third international conference, OCSC 2009, held as part of HCI International 2009, San Diego, CA, USA, July 19-24, 2009; proceedings. Springer, 2009.
• Ragins, Belle Rose, and Kathy E. Kram, eds. The handbook of mentoring at work: Theory, research, and practice. Sage, 2007.
• Rowland, Kimberly Nicole. E-mentoring: Benefits to the workplace. University of Maryland University College, 2011.
• Silberman, Melvin L., and Elaine Biech. Active training: A handbook of techniques, designs, case examples and tips. John Wiley & Sons, 2015.
• Stromei, Linda Kyle, ed. Creating Mentoring and Coaching Programs: Twelve Case Studies from the Real World of Training. American Society for Training and Development, 2001.