RAA3: Analyzing #insomnia on Twitter

Jamison-Powell, S., Linehan, C., Daley, L., Garbett, A., & Lawson, S. (2012). “I can’t get no sleep”: Discussing #insomnia on Twitter. Proceedings of the 2012 ACM annual conference on Human Factors in Computing Systems (pp. 1501–1510). Retrieved from http://dl.acm.org/citation.cfm?id=2208612
Purpose of the Research: 
(1) To explore the role of social media in the discussion of mental health issues, and with particular reference to insomnia and sleep disorders: whether individuals are disclosing their insomnia related information online, if so, what they are disclosing.
(2) To provide analytical results to those in the interaction design community interested in integrating online social media systems in health interventions.
A mixed-method approach is used to analyze 18,901 tweets using hashtag #insomnia. First, an automatic content analysis is used to confirm “whether individuals are disclosing their insomnia related information online”. Second, a qualitative thematic analysis is performed to answer the question “what individuals are disclosing about their insomnia”.
(1) Content analysis using a software named Linguistic Inquiry and Word Count (LIWC) . Simply put, this software contains a pre-defined English dictionary which includes categories of negative words and positive words, as well as word roots related to health, time, anger, anxiety, sadness, etc. The software performs frequency counting and statistical analysis of the input text based on this dictionary. In order to compare these #insomnia tweets with the general posts on Twitter, 17,532 random general non-specific tweets on Twitter were used as a control group. LIWC is used to compare the #insomnia tweets and the non-specific tweets, and produces table 1 below.
(2) Inductive thematic analysis. The authors acknowledge that the LIWC software analysis results maybe inaccurate because of the subjectivity and ambiguity of  informal human language, as well as the large amount of abbreviations used in tweets due to the character limit. So they perform inductive thematic analysis. A sample of 749 is taken from every 25th tweet of the original 18,901 tweets. One researcher first read the 749 tweets for several times, and comes up with 49 categories, and gives the results to two other researchers. The three of them discuss and decide on 45 categories. Then they take another sample of 514 tweets from every 35th tweet of the original 18,901 tweets, and code the 514 tweets into the 45 categories independently.  Inter-rater reliability is also examined.  All the researchers finally refine the categories into a hierarchy in figure 1.
Main Findings: 
(1) Results in table 1 below confirms that individuals do disclose insomnia related information online.
(2) Inductive thematic analysis categorize the content of #insomnia tweets into two major categories: describing the insomnia experience and coping with insomnia.
The authors claim that this paper, in a generic sense, contributes to a growing body of literature regarding online self-disclosure. Based on the analysis results, they come up with a brief design requirement for a social network structure allowing insomnia therapy to be delivered in an online group. “The structure would be a useful format for users to exchange support and practical information. To allow users to exercise catharsis,  a space could be provided to disclose symptoms and frustrations together with an area where the users can “communicate with” insomnia. To comply with the principles of sleep hygiene, a digital barrier should be in place in any interactive platform, this would reduce the amount of access a user has to an online support network at times when they should be refraining from stimulating environments. However, rather than simply restrict all access a “life-line” should be kept open, providing information on good sleep hygiene and progressive muscle relaxation exercises to aid sleep. “
This paper relates to what we learned in CGT512 how to analysis qualitative data and thus produce design requirements.
This paper also relates to my research interests in social media analytics for instrumenting the social and emotional aspects of college students’ life. It provides information on useful software and methods.

Marc Smith NodeXL Talk Sep 22 @ Purdue

I have an interest in Social Media Analytics, so I went to a talk by co-founder of Social Media Research Foundation Marc Smith titled: Charting Collections of Connections in Social Media: Creating Maps and Measures with NodeXL. Some takeaways and thoughts I get:
1. Marc mentioned that Usenet is still growing at a pace of 11% per year. I was wondering why, because it seems nobody around me here use it anymore. Maybe it’s growing more outside of US, I know in mainland China and Taiwan, BBS is still very popular, especially in higher education institutions. In a later Social Media class, Dr. V also mentioned that Usenet is still widely used in certain groups of people who started with it. So I think maybe the body of a new generation who is born with Google+ will grow larger in the future, but then the old generation will still maintain on Facebook (at least before this generation die away), so Facebook don’t need to worry and panic now, maybe it should take some strategies to prepare for the new grown-ups 10 or 20 years later. Some of the thought here is inspired by this blog post by Dr.V.
2. Marc talks about one of the reasons for developing social media monitoring tools. When we see a large group of crowd on the street physically, we can actually see some patterns of the crowd. For example, they have formed various sub-groups, talking in different languages, and wearing different clothes. However, the shape of the crowd on the web is totally hidden. “The interface of social media is remarkably anti-social”. The nicely and neatly organized lists make everybody’s tweets and profiles the same format, thus break the shape of the crowd. So we want to take closer look at the content to see the patterns and shape of the crowd. To deal with large amount of data created by the crowd, we need social media monitoring tools with great computational ability. Here comes NodeXL–and MS Excel add-in that could help social scientists investigate the shape of the crowd with 0 line of programming.
3. Another interesting topic he talked is about strong and weak ties. Internet has increased the number of relationships we maintain in a great deal, however, most of them are weak ties. Strong ties need dedication of time and effort to maintain, and we only have limited time and attention. So are these weak ties useless? No. They matter a great deal! We usually share almost the same information with our strong ties,because we are very close to each other. Our weak ties are people who can bring new information, new possibilities, and job opportunities. Think about your own situation, if everybody you know knows everybody else you know, your job is not secure, because you are a redundancy, you can not bring new information to these people.
4. Marc showed a graph made with NodeXL about the connections among users who recented twitted “Purdue” when queried on Sep 16 2011.
20110916-NodeXL-Twitter-Purdue Composite
A larger version of the image is here: http://www.flickr.com/photos/marc_smith/6155750905/sizes/o/in/photostream/
Based on my understanding of Marc’s opinion, some of the in-between users in the middle green and light green areas are more important than the ones who are in the largest networks (the two big “hairballs”), because these in-between users connect the scattered people with the two big networks. Without them, these scattered people will lose the connections. Also sometimes, the big balls probably are created by famous patrons who don’t really belong to the group. I cannot remember the term Marc use to call them.
5. Another interesting graph Marc showed is:
2010 - September - 5 - NodeXL - Twitter - GOP
This is about people’s political opinions about GOP. Red represents GOP supporters, and blue represents critics. The in-between dot is political.com. This pattern has only been found in the US, not in other countries. So America is by far the most polarized country in terms of political opinions. This is an interesting finding that proves America is exceptional somehow.
Reflection about Tweeting the talk live:
I got a mission via Twitter from Dr. V at the beginning of the talk to tweet the talk live. I saw other people do so before but never done it myself. It’s quite interesting and intense and overall beatifically to me and maybe to whom read my tweets and this blog.

I felt super busy because I have to listen, tweet, and write my own notes at the same time. Later I abandoned my own notes because I couldn’t write notes and tweet at the same time. Today when I write this blog, I have to go back to my tweets rather than my notes to help memorize what was going on there yesterday.

One good thing about tweeting compared with taking notes myself is that I am responsible to make comprehensible sentences for my twitter audience, while taking notes I can use misspelling words, symbols and incomplete sentences. Some of the misspelling, symbols, and incomplete sentences never got corrected after I took them.

One drawback of making responsible sentences is that I have to think while couldn’t pay attention to listen. There are some words I don’t know how to spell exactly and I have to go to look it up. For example, Marc talked about the number of relationships chimpanzee could maintain vs. human could do. I couldn’t remember how to spell “Chimpanzee“, and couldn’t find it in a short time, I only find gorilla in my dictionary. I know gorilla and chimpanzee are different but I couldn’t remember what is the difference in the Chinese words, so I didn’t input the right Chinese word for chimpanzee, so I cannot find how to spell chimpanzee. I gave up to tweet this finally, and wrote a misspelling word in my own notes, but it took me a while.

Although this kind of things took some attention, but overall tweeting makes me pay more attention to the talk, because I am responsible to tweet as accurate as I could to my twitter audience. If I were not tweeting I probably would think about what I should have for dinner and daydreaming some other stuff that could take even more attention than looking for how to spell Chimpanzee. Overall, I think tweeting while listening and write a blog post about it is the most beneficial method to get the most out of a talk that you are interested in.

How to Find People to Follow and my Anger about Twitter

I said in class that I tend to blog when I am angry or unhappy, since I am so angry about Twitter now, I decided to write why I am angry first and then write about how to find interesting people to follow there.

What makes me angry about Twitter (disclaimer: these are all just my personal opinions based on the way I use Tiwtter, the purpose I use Tiwtter, my situation, my technological background, etc. I understand that different people use it for different purposes, and different tools are designed for different purposes. ):

Twitter Web UI

1.First of all, I tends to use more than 140 characters whenever I really want to express my idea. So blog is way more suitable for me than twitter. I know I have to learn to be concise, but I don’t want to be concise to a point that I have to write so many abbreviations and acronyms that don’t make sense to me, nor other people. I am intending to substitute twitter with blog, they are different platforms anyway, but something like Facebook status or Google+ stream would be better.

2. I set the preference for notices whenever I get mentioned or replied, but I get notice sometimes, sometimes NOT.

3. Why I heard so much complain about privacy issue on FB, not that much about Twitter. People just forgive it because it’s simple, and because it’s meant for public?

4. There is not a concise way of organizing the conversations, they are all just mashed together. I cannot just follow a the conversation about one hashtag, I have to hear all the irrelevant tweets by people in the list.

5. I cannot use Google Voice number to send text to or get text from Twitter! This maybe a problem of Google Voice, rather than twitter.


1. I took me forever to find the “favorite” button, and actually it’s not a button.

2. I don’t know what is the difference between group and list here, when I created a group, it automatically turned to a list, and the group is empty. If you don’t want the concept of a group, why leave the function there?

3. I feel it’s convenient to use the notice window to reply people’s tweets. However, if I write reply here, the autoname function disappears and the auto-including hashtag function disappears too. I have to manually type them in.

4. The biggest problem here is that I cannot see the back-n-forth conversation among people very clearly here. If there is one original tweet, and several replies. When read the original ones, I cannot see the replies. When I see one reply, I can click “in reply to” to see the original one and this one reply, and also many other irrelevant conversation between these two people will also show up, and I cannot see other relevant replies. The Web UI has a right arrow at each tweet that is nicer. The right arrow and the “in reply to…” are different functions, and tweetdeck only has “in reply to…”

5. When I reply somebody’s tweet, I cannot see my own reply in a thread, which will make think I didn’t reply successfully. I have to go to check my tweets again.

6. A good thing here about Tweetdeck is that I can create a search column, so I can only follow the relevant tweets here.

Twitter iPhone:

I am very disoriented when I first started to use the tweetdeck iPhone, I don’t know which column is which, and how to go to the next column. The screen is so small and cannot display the whole name of the column sometimes.

My iPhone, my email and Tweetdeck send me notices at different paces, and sometimes one of them will omit some tweets, or delayed for a long time. Or sometimes I am not in front of my laptop and have to rely on my phone. So I cannot turn either of them off, in case I didn’t see some important tweet. However, sometimes they are really disturbing, I just want to turn all of them off and do something else!

As we are going to talk about Internet Culture next week, I feel every phenomenon is existing and has evolved around a culture. Twitter is like this too. It is designed for 10 people, and has evolved to this point today because of the collective effort of people who like it or use it. These people grew up with this culture and are very much used to this culture. Other people, like me, very different to enter into this culture. Of course, the difficulty also depends on whether I really want to enter into this culture or not, whether I think it’s worthwhile. If I decide that I have to enter into this culture no matter what, because it will bring value to my future, I will do it no matter what. Similarly, I am always angry that there are so many abbreviations and acronyms in English, but this is how this language has evolved, thus if I decide to use English for the rest of my life, I have to embrace it.

Now I think the key point probably is that to put important time urgent information on Twitter is a bad idea. Time urgent information should be left for emails and phone-calls. Twitter is just a loosely coupled information system, where we may find interesting people and interesting articles to read if we want. To put timely urgent information there is simply too much pressure, because then I cannot turn if off as I want anymore. I understand the practices on Tech621 is just for us to immerge in the social media world, but no more in my normal life.

Okay, let’s go to talk about the real topic: how to find interesting people to follow on Twitter. Although twitter has made angry for many parts, but there is still one part that is crucial and useful to me, that is, I can find important people to follow and find interesting article to read.I am sure there are many ways to identify people, and Dr.V has mentioned some in class. Here I just provide one unique one that works for me:

First, follow an organization. Second, follow the people that showed up in the tweets of the organization. Third, unfollow the organization.For example, I followed an organization @socialmedia2day. It will tweets people’s names and their interesting blogs, then I will go to look at the blog and decide if I am going to follow the people. After a certain period of time, then I starts to know many people to follow, and starts to utilizing other more advanced ways to follow people, then I can unfollow the organization.

Twitter Reflection

Rethinking Recruiting….

I read this blog about how useful twitter is for recruiting.  It should also useful for us in similar way (for job hunting, and academic career), and I always have this belief which has been enhanced by TECH621 class. However, I am not quite there yet. I still feel it’s a lot messy on my twitter, and feel burdensome just to clean it up, and there are interesting people there I am following, but I simply just read too slow to follow all those interesting posts. Hopefully I will get there this semester.

I definitely understand why Dr. V push us to collaborate on Twitter, because of all those potential benefit mentioned above if we get really professional on Twitter. We did a fairly good job. However, Twitter seems not a very good tool for this kind of collaboration, especially for some of us who are at the starting stage of using twitter. We are not automatically included in one conversation if our name didn’t get mentioned in one tweet, we have to manually click the #Tech621 hash-tag to see what was going on. We should have created another hash-tag just for this presentation. Since we didn’t do that, there are probably hurdles in between the conversation: somebody posted the topic they want to do, but another person didn’t see that tweet, and posted the same topic later. I heard about complain about that A wanted to a certain topic, and posted on Tweet, but B “stole” A’s topic later on, probably because B didn’t see or didn’t pay attention to A’s earlier post.  I made a summary tweet of who’s going to do what topic at the early stage to make things clearer. Sometimes there are too much traffic on twitter and too much distraction, it’s very easy to miss some relatively important messages, especially for starters.We coordinate well on twitter, but me and my partner still met offline to go through the slides and clean things up.

I would think a Skype discussion group (by including other people in the conversation) would be better, or any other tools of the similar nature (I’ve used QQ messenger and Fetion messenger before, and they works just fine. Gtalk would do the work too) where conversations happen exclusively among the group members without outside disturbing, also the messages will be saved if a certain group member is offline, thus he or she will see the messages next time upon login.

I should dig more into Twitter to make it truly useful for my career. Hopefully, I will get on the right track at the end of this semester.

P.S. Something for your laugh and thought. Here is an interesting video about twitter. A young man who doesn’t tweet said those who tweet all day for the reason of hanging out with friends don’t really have friends. This caused the Fail Wale, which caused a huge panic: