SOCIAL MEDIA RESEARCH
With the explosion of social media, data being generated from these sources can be used to inform policy by understanding the general sentiment about the topic area in the general population. My colleagues and I recently presented on how the vaccine attitudes were travelling through social media. This can be ascertained by either following the path of certain data analytics or even social network analysis. At the end of the webpage, there is a brief comparison of traditional qualitative research techniques versus those that are AI-guided. As a general rule of thumb, AI-guided recognition or reading, if employed properly, can elicit far superior results due to the use of complex algorithms.
Start off by listening to this important podcast where Dr. Gjellstad (head of the ethics department). Dr. Stadtlander, and I describe the feasibility and ethics of social media research. Having also been involved in primary data collection process caught me a bit by surprise. We compare AI-driven approaches versus traditional qualitative research. You will hear about this in the latest podcast!
When you heard the podcast. what did you think? Did you feel like you were able to understand and grasp adequately what we were describing? If not, don't worry, below is another opportunity to view the same information--however this time much greater detail with original slides that I created. This research can help enhance the meaning from thematic analysis and phenomenological qualitative research.
Now that you have reviewed some of the AI research and have familiarizes yourself with the terminology, I will share with you the results from sentiment analysis. The first figure below is your first introduction into python code, which is showing how to connect directly to the API by using a Twitter developer account. For more information on how to apply for this Twitter development account, you can click on this link.
Now that you have seen how sample python code, let's take a look at some of the pretty graph that we can tell python to draw. The measures that are being assessed in the x-axis is polarity as seen below for Tweet sentiment on vaccinations since November 1, 2021:
Usually, NLP is good at positive or negative words. However, as you see above the combination of positive and negative words in a sentence, and how to assess this as a whole is where polarity comes it. Now that you see the polarity about attitudes vaccines in November. Please take a look at the graph below and the vaccination attitude since February of 2022.
Now that you see demonstrated some important sentiment analysis features, it is important to note that there are more complex sentiment analysis features you can discern through textblob. Now that you have seen how to organize numerical measures, I will show you how python completes a thematic analysis which is similar to NVIVO. Both programs use AI, but the coding in python is entered by the user.
From the words, you can observe some patterns. For instance, it was interesting that the word "need" was highlighted. This emphasizes that applying Maslow's Hierarchy of Needs, vaccination is considered a basic need that is required for people to feel safe. Additionally, the word "know" was really important, especially in the context of widespread misinformation that was present globally. Knowledge was so important, that based on this, daily policies were changing. If you are really interested in viewing rest of the webinar, please view the following video.
Now that I have presented the AI-driven research, it is important to revisit a traditional qualitative research techniques. As you listen to me explain this, please try to think if there are certain AI-driven research can in fact data collection process. In the following 3-minute clip, I differentiate between Hermeneutic versus Transcendental Phenomenological research. If you think reading this is challenging--then try pronouncing it!
Now that you understand phenomenology better, you can see how NLP-guided recognition of common terms may sometimes contain more information than conducting thematic manually. When conducting qualitative research, more novel techniques should be explored before determining an appropriate methodology to use.