I’ve been doing some more work on social media scraping and data analysis. In particular, I’ve been thinking about what kind of measures we can put into place in order to examine the kind of learning that is taking place across the social media platforms. This also needs to be assessed based on the kind of data that is present in the social media platform that I’m scraping, and also what kind of data I can scrape and how I can treat that data for the purpose of analysis.
And, of course, all of this needs to be based on the ethical considerations that are important within this space, being especially mindful that anonymity is very difficult in an online space and that will influence the kind fo data that I gather, and how I present that data. I’ve found Annette Markham’s work about Fabrication to be helpful in this space – but I will discuss that further in another blog post.
So, let me continue. I think the first assumption that I’m making in this analysis is that learning is based on interaction and engagement; indeed, it’s a function of that engagement. But how can engagement be measured? Well, I think that a proxy for engagement might be the level of ‘likes’ a particular post garners. A proxy for interaction, on the other hand, might be the level of interaction a post creates through comments (and possibly shares/ retweets or whatever, although the platform I’m using for this example doesn’t natively allow sharing). These factors can be measured through calculated data like like: follower ratios, or comment to follower ratios. Another aspect worth considering here: targeted or @-posting.
Another aspect that might be worth exploring is the content of the posts that are being shared. I think that organisations and individuals that have very clear educative goals in mind probably have different kinds of posts: for example, there might be factual posts, calls to action, emotive posts, targeted posts. Examining the content of these – both at a caption or textual level, but also the affordances level, and also a visual level, too – might be indicative of the kind of posts -and I could then categorise those against the level of engagement, in order to make judgements about the efficacy of the learning.