Blog Archives

Researching images on social media – nuts and bolts

Images and videos are pervasive online, these days, web articles include at least one image or video. On Twitter, Facebook and Snapchat these visual contents are even more common, and social media platforms such as YouTube, Vimeo, Vine, Instagram, and Pinterest are entirely dedicated to their sharing.

screw-mother-metal-iron-60060

Images can emphasise textual messages, or even convey a message without text at all (Hankey et al., 2013), and they can increase the visibility of a tweet and how often it is shared (Yoon and Chung, 2013). There are so many images on social media that these platforms have become picture databases, and these have become subject to research. For example, Vis et al. (2013) explored images production and sharing practices on Twitter during the UK riots in 2011; Tiggemann and Zaccardo (2016) analysed Instagram images related to the #fitspiration movement, addressing their potential inspiration for viewers and negative effects on viewers’ body image; and Guidry et al. (2015) investigated the content and the engagement of pro- and anti-vaccine images shared on Pinterest.

My Ph.D. research uses one of these databases – it focuses on vaccine images used for advocacy that are shared on Twitter. Sourcing the images that are my data may sound simple, after all, I only need to download my data from Twitter, right? However, it is rather more complex than that. To start with, there are many different communities on Twitter, and they share images on a range of different topic. They may also share images on the same topic from different angles; for example, if we search #health on Twitter, we will see pictures related to healthy food, obesity, fitness, losing weight, public health policy, etc. So, the biggest challenges are how to find the communities of interest and then to develop a data analysis strategy that uncovers how they use their pictures.

To help me narrow the potential field of image research for my PhD, I asked the following questions:

  1. What topic am I interested in? Which communities do I want to study?
  2. Which social media outlets would I find most interesting/useful for my research?
  3. Each social media platform is used by different audiences, so it is important to think about the overall question we are asking. For example, young adults use Facebook, whereas teenagers prefer Snapchat, and Chinese people may be on Weibo.
  4. Where are these communities from? Which language(s) do they use?
  5. If we focus our research on Europe, we have to take into account that Europeans speak different languages. If we focus on English language, we have to consider that our images will come from all over the world, but especially from the US, UK and Australia.

Afterward this initial sifting, I had more questions to answer:

  1. What keywords should I use to search on my chosen social platform (in my case, Twitter)?
  2. Each topic and each community has its own “slang” or “dialect” and therefore keywords. On Twitter, for example, users in favour of vaccinations tweet their content including the hashtag #vaccineswork, whereas people against vaccines use mainly the hashtag #vaxxed and/or #CDCwhistleblower.
  3. How can I find the relevant keywords?
  4. Previous research on social media can suggest some terms; in my case, keywords such as vaccine(s), vaccination(s), vaccinate(d) and immunes(z)ation (Love et al., 2013; Salathé et al., 2013). Searching for these generic words, I found both tweets with and without hashtags that talked about vaccines. However, some communities use specific keywords which may not include these terms (e.g. #vaxxed) and they may use these keywords to label their tweets as relevant to the topic. For example, a tweet claiming “They’re poisoning our children #CDCwhislteblower” and showing an image with a child whilst being vaccinated, would be relevant to vaccinations even if it did not mention “vaccine” or “vaccination”. This tweet would not appear in my research if I set my data collection using only generic words, thus I needed to search for relevant hashtags as well.
  5. How do I find relevant hashtags?
  6. A first step would be considering which hashtags previous studies used, then searching Twitter for generic hashtags and see which other hashtags people use. There are also some online tools that can be helpful, such as Hashtagify.me, Get Tags and RiteTag.com. These online software packages suggest correlated hashtags and their popularity.

Answering these questions helps us define the criteria for data collection, but they also show how complicated research on images shared on social media is. As with any data collection method, planning, defining and developing are key for research drawing on online images. We need to be able to justify the approach we took and show that the data collection process is robust. This means, as with many other types of data collection, that we need to pilot and test our data collection methods ensuring that they deliver the material we anticipate and which will validly help us to address our research question. There are so many pictures online, uploaded, downloaded, edited and shared, that the choice of image collection methods becomes key to ensuring the quality of the study overall.

 

Elena Milani

 

References

Hankey, S., Longley, T., Tuszynski, M. and Indira Ganesh, M. (2013). Visualizing Information for Advocacy. Nederlands: Tactical Technology Collective.

Love, B., Himelboim, I., Holton, A. and Stewart, K. (2013) Twitter as a source of vaccination information: content drivers and what they are saying. American Journal of Infection Control [online]. 41(6), pp. 568-570.

Guidry, J.P., Carlyle, K., Messner, M. and Jin, Y. (2015) On pins and needles: How vaccines are portrayed on Pinterest. Vaccine [online]. 33(39), pp. 5051-5056.

Salathé, M., Vu, D.Q., Khandelwal, S. and Hunter, D.R. (2013) The dynamics of health behavior sentiments on a large online social network. EPJ Data Science [online]. 2(1), pp. 1-12.

Tiggemann, M. and Zaccardo, M. (2016) ‘Strong is the new skinny’: A content analysis of #fitspiration images on Instagram. Journal of Health Psychology [online].

Vis, F., Faulkner, S., Parry, K., Manyukhina, Y. and Evans, L. (2013) Twitpic-ing the riots: analysing images shared on Twitter during the 2011 UK riots. In: Weller, K., Bruns, A., Burgess, J., Mahrt, M. and Puschmann, C. (2013) Twitter and Society. New York: Peter Lang Publishing Inc., pp. 385-398.

Yoon, J. and Chung, E. (2013) How images are conversed on twitter? Proceedings of the American Society for Information Science and Technology [online]. 50(1), pp. 1-5.

Advertisements