The first limitation of this research was disclosure of information in the data collection.
Self-disclosure is revealing personally relevant experiences, thoughts, and feelings to others (Derlega, Metts, Petronio, & Margulis, 1993). The problem lies in whether the information divulged in the interviews is true and reliable. Research has suggested that in computer-mediated environments the chances of self-disclosure is greater (McKenna & Bargh, 2000; Schouten, Valkenburg, & Peter, 2007; Tidwell, & Walther, 2002; Walther, 1996; Tian, 2011).
Since an anonymous environment of Google Docs was used for the synchronous online interviews in this thesis, it may have been of help to gain more reliable data, and so it can only be assumed that the information gathered was true. A bigger sample size would also help reduce this uncertainty.
As is common with other qualitative studies, this research also suffers from low external validity and little ability to generalize the findings. The sample of this study is also not
representative of any specific population and only show the experiences and ideas of the interviewed participants. The size of the sample pool was limited due to the researcher’s time available. Another possibility is the lack of an attractive incentive or reward such as a monetary incentive or gifts. Even though this was aimed towards the best interests of streamers and spectators alike, future research should aim at having an incentive attractive to their study subjects. Furthermore, should a future researcher require having a study generalizable to a greater population of Twitch Affiliate Streamers, the sample should be greater.
In the initial phase of this thesis, focus groups were to be conducted however due to low turnout of volunteers, it was decided to change the methodology and attempt one-on-one online
47
DOI:10.6814/THE.NCCU.IMICS.005.2018.F05
interviews with streamers. Even after changing to interview method, the turnout was still low, possibly due to the lack of physical or monetary incentive.
Also related to sampling, recruiting streamers especially Affiliates and Partners is difficult, not only because of their busy schedule, but unless one already has a network of close Affiliate streamers, the only place to recruit them would be online social media like Facebook or Twitter, or forums such as Reddit.
An important limitation in this thesis is the potential of sample bias. The question that is important to ask is why were there only ten interviewees who volunteered to be interviewed?
There is the possibility that these streamers are volunteering to the interviews with an ulterior motive, such as to gain more recognition. Furthermore, it is important to take notice of the fact that streamers with two or less years of streaming experiences came forward. Due to this, the thesis could not consider the possible motivational differences of streamers with more streaming experiences.
Another limitation related to sample bias is the possibility of cultural differences
affecting motivation differences. Since the sample in this research has a majority being from the United States, there is the probability of a cultural bias. Thus, the results provide a more
‘Western’ perspective, not having the opportunity to analyze streamers from different countries or different continents. A reason for this is not only due to Twitch having a majority of American streamers and viewers, but a language barrier could also exist, with streamers hesitant to reach out and volunteer for the interviews due to inability to speak English.
Unintentionally, eight out of ten of the interviewees were from the United States, with the other two being from Canada and Sweden. This could be a limitation because people from other
48
DOI:10.6814/THE.NCCU.IMICS.005.2018.F05
countries or continents might consider other motivations more significant. Furthermore, this research has left out economic background of the interviewees, which could play a role in conjunction with motivations to stream and could be implemented in future studies.
Qualitative interviews are also impossible to replicate since they rely on social
interaction, location, medium used, and no matter whether face-to-face, online, or phone calls, interviews are time consuming. Furthermore, qualitative gather is subjective and specific to the participants and because of this it is even harder to generalize.
With regards to conducting online text-based interviews, no pilot test interview was conducted to refine the questions and interview guide due to time constraints. This could be an improvement for future studies using this method of data collection. Using the method used in this thesis also resulted in a lot of extra unusable information, which can be expected even in traditional interviews, with interviewees wandering off in unexpected directions (Leech, 2002).
This study is also limited to one Livestreaming platform, the most popular one for gaming, Twitch. Future studies can analyze other livestream platforms that also broadcast audio visually, YouTube and Hitbox just to mention two of them. As YouTube is best known for being the internet’s biggest video repository, it does not share Twitch’s reputation of being dedicated to game streaming as Twitch does (Pires & Simon, 2015). Furthermore, according to Hemmert (2017), Twitch was the market leader in “video gaming video content monetization” with a 43%
“share of advertising, donation and subscription revenues in the segment”. And though it is mostly up to the preference of the streamer, Twitch was shown to be the more desirable streaming site with YouTube’s main focus being uploading recorded and edited videos
49
DOI:10.6814/THE.NCCU.IMICS.005.2018.F05
(Hemmert, 2017). Due to this, the results of this thesis may not also represent the motivations of streamers based on YouTube.
In agreement with King (2004), there is a lack of substantial literature on Template Analysis which as a novice researcher, left doubts on whether the analytic decisions taken were the best ones. This would result in oversimplification or overcomplication of the final templates.
In this case it is probable that the template may be too simple however still did the job to find the results to the research questions of this dissertation.
50
DOI:10.6814/THE.NCCU.IMICS.005.2018.F05