The regular meeting today seemed to go well. Main points to come out of the meeting:

• I need to think about whether my three proposed characteristics of technologies for smoking cessation – persistent, interruptive and responsive, should be collapsed down to two dimensions. Steve suggested persistent-transient and responsive-unresponsive. I’m not so sure about this … needs a bit more thought
• It probably is worthwhile interviewing the long-time quitters in the jaiku quit group
• I don’t need to be so concerned about the research question at this stage. My earlier efforts have focused the research somewhat, which has been useful. At the moment, it’s best to focus on the data and re-build the research question around that
• I need to get my literature review re-written, and my methodology written before next week. Procrastination be gone!

Going through this process also made me wonder about the role of the data I was intending to collect. How would the data support the assumptions made in the models? I found myself at one of those points where it seems like you’re doing research in reverse. So, back to square one, and time to re-ask some fundamental questions – What is the research question? Am I trying to modify a model, integrate some models or just use some models to get at some information? The first of these questions really should have been more present in my mind, but seems to have been ignored. The second, about models, can only really be answered once I have my data. There’s only so much you can guess about how a model can be changed when you don’t have any data.

I’ve resolved, at least for the moment, to press on with the following research question, which I have a feeling is at a good level of abstraction and will let me get the most advantage out of the contextual data I hope to collect (more on my new methodology later).

What are the influences on the ability of quitting smokers to access online social support at times of need?

In answering this, I will explore the activity of accessing support. I will use the TAU dimensions to explore this activity, and will use aspects of the DMR to explore contextual issues and temporal issues related to smoking.

I’m going to define ‘access’ as the sending or receiving of social support ‘content’. A ‘time of need’ may be defined either by the individual (eg. a craving or experience of withdrawal symptoms, intrapersonal negative affect), their context (eg. a high risk situation, conflict with others) or by research into appropriate times for intervention. In including the last of these, I hope to be able to cover interruption or persistence as a part of the investigation. Research has hinted that this is ‘needed’, but it is not a need that quitting smokers would necessarily express (note the inability to interrupt in online social support). Online social support will be defined as support in online communities from other quitters as well as micro-blog users who publish their quit experiences to their friends online. Some interesting comparisons could be brought up between conventional online communities and new ‘lighter’ social networking in micro-blogs, where communication is not only with other quitters but with a regular extended social network. “Online” may include mobile access to online technologies, eg twitter.

If I were to pre-empt some of the findings, I would be thinking about the interplay of rhythms and interruptions. The rhythms of the physiological effects of quitting, the rhythms of device use, rhythms of social interaction and other life rhythms may all have an effect. Interruptions would feature in a couple of ways. Experiencing a craving in a high risk situation could be seen as an interruption to the activity the individual wishes to be engaged in. We could also think of ‘remedial’ interruptions such as reminder emails.

Surely this needs another iteration, but I’m going to sit with it for at least 24 hours. I need some focus at least for the next little while. And I would really, really like to be able to keep most of my literature review! A possible problem: I think I (and other researchers) already implicitly know what many of the influences are. Also, the rhythms of internet use are perhaps not as interesting and “rich” (ie tied to interesting contexts) as the rhythms of mobile use. Then again, perhaps nobody has explicitly pointed out these inadequacies of online social support. Maybe I need to push for the distinction between tonic and phasic social support, and underline that an online community supports the tonic but is inadequate at supporting the phasic. This leaves a window to talk about mobile technology in the discussion, also highlighting the capacities of the different media to convey different kinds of social support. Another idea is to look at access to support in general at times of need. This could take in a variety of technologies … but I suppose it would better to just focus on one.

Re-thinking methodology

Well, failing some kind of miracle, it looks like I won’t have time to run the trial I was planning to. Hours of work building the website for the project and planning the logistics gone to waste. I knew there was a chance it would all be in vain, but I had to be prepared. Enough negative talk – how can I make this work? What data to I have at my disposal? How would that data help me to answer the question?

Possible sources of data

Interviews with online support community members: In order to get at ‘the influences on their ability to access online social support’, I would use the TAU and DMR concepts to probe their use of the community. This should lead to some discussion on the benefits of mobile technologies From members that have already contacted me, I should hopefully be able to tee up four or five online chat interviews. I could ask the administrators of the community if I could message some other members about the interview directly. Hopefully, I would have a few more bites this time. It would be good to get a list of members that had quit in the previous month or fortnight.

Interviews with quitext users: Quitext is a new service launched by an Australian company that sends SMS messages of advice for quitting smokers on a schedule defined by the individual. They’re quite a new service, though, and I’m not sure how many subscribers they have. I don’t think it’s that likely that this option will come through.

Channel on Jaiku of quitting smokers: These guys formed the group I was hoping to in my study. I could join the group and ask them to post more often, also ask if they’re posting from mobile or web. Could then do email pre-interview and web-chat post interview. Could be a bit risky. This is the last chance to get the micro-blogging trial idea off the ground. Otherwise, just interviewing some/all of them would be interesting. They seem to have stopped posting about their quit so much recently, but they have been posting to each other on other topics. I don’t know how many of these people knew each other beforehand. There is surely something interesting in here, I’m just not sure how relevant the TAU is to this any more. Need a rethink on the question if I go down this path.

Twitter feeds of people quitting: Across both micro-blogging platforms, there is one user on Twitter that posted the most about her quit attempt. About 20 times over the course of three weeks. She was already a regular twitterer, and she slotted her quit-related posts in between all her regular posts. She kind of ‘hijacked’ the communication channel in her quit attempt. The feeds themselves would be interesting to look at and the intervening posts would give some idea of the context. Interviewing her would be great, and her micro-blog would serve as a perfect aid to her memory, to get her back into the moments when she was posting and how she was thinking/feeling. If I could land an interview, then there could be a valuable methodological contribution regarding the use of micro-blog feeds to aid in interview recall.

I better stop crapping on now. I’ve read, re-read and re-written this post about 20 times. I’m feeling that focusing on the temporality of Quitnet use is a bit dull. I would ideally like to look at the micro-blogging quitters, but I need to define a research question and a contribution angle. I don’t know if TAU is going to cut it any more if I make so many changes to the methodology.

Hmmmmmm

Methodology

We discussed in detail my proposal for analysis of the data. Yesterday I went through the four dimensions of the TAU as proposed by Fabre, and thought about how I would get at them through the data. I was assuming that they would be the ‘lense’ and that the relationship between it and the theories of behaviour change/relapse (the Transtheoretical Model and the Dynamic Model of Relapse) would come out in the discussion. Steve suggested that I need to get at the constructs in those theories more clearly in my methodology, rather than just focusing on the TAU dimensions. This led to a discussion on what would actually be produced in the discussion section of the thesis – what will the contribution be? I’ve given up assuming that this question will have an answer before it’s handed in.

My intention to do some simple statistical analysis was questioned. We resolved that we would have to wait and see what data I get to see whether it can be made to play any part in the story.

Contribution

Our discussion today formalised what had been floating around in the back of my head. I’m going to look to synthesise the TAU framework with the temporal aspects of the models of addiction and relapse. The product would then be:

An model for the temporal aspects of usability in technology to support smoking cessation.

Obviously a shorter, snappier title needs to be formulated, but this is essentially what I’m trying to do. The solidification of this goal means that I need to make sure I can get at the temporal aspects of all of these in the data. Steve said he wanted to see a more definite indication as to how I will get at all of the constructs through the data. I think my mind map from yesterday demonstrates most of that, though I need to do some work on pulling out the constructs from the TTM and the DMR (maybe even just the DMR… we’ll see).

Contingency

Ethics clearance was finalised last night. I’ve contacted QuitNet and hopefully the message will go out to thousands of QuitNet members in the next couple of days. I need to come up with a contingency plan ASAP – what will I do if I don’t get any interest, or it takes participants too long to organise themselves into groups?

Timeline

I need to get participants going ASAP. I also need to make a call on whether there is enough time to do a one week pilot. If I get some participants by the end of this week, that may be possible, otherwise I really don’t know if I’ll have enough time.

By the end of September:

• All data should be in, with most of it having been analysed
• Draft of the methodology
• Draft findings

For October, this leaves:

• Draft dicsussion
• Finesse literature review
• Hand it in on October 26

Summary

I feel like I’m pretty well on track. I need to finally sort out the issues with the stage of change questionnaire, which have been bugging me for quite a while. I think I need to just put some together and get someone at TCCV to check over it. Also need to finish the mini-site with all the help info for participants in the study. And address all the issues brought up in the meeting. My work is cut out for me – it’s time to get cracking!

Should I set up their twitter accounts for them? 

I should offer that I can do this for them. If I set up an account, I will need to use a new email address to do it. I will need to make sure that the participant enters their own email address at the start of the study – in case they want to be sent email reminders or forget their password. They may also want to change the password.

Should I maintain full access to the accounts?

I was thinking that maybe I should be able to have access to their accounts so that I can see the private messages they have sent to each other. My supervisors suggested that I should keep them private and ask them to disclose as much about their private messages as they are comfortable with. This might be as little as stating how often they sent private messages, or as much as cutting and pasting messages they are comfortable with me seeing.

How can I ‘refine my data collection approach’?

Steve suggested that I try to collect some dummy data and start to figure out how I may approach the analysis. This should hopefully streamline my approach ahead of getting the real data and help me to write up my methodology more clearly.

Deadlines

I now have a fortnight to produce a rough draft of my literature review. For some reason I’ve had it in the back of my head that this will be pretty easy, but I think I’m getting a bit ahead of myself. The literature review needs to be a well crafted story.

Background & Context: Introduce the problem space

Problem: A problem that has been identified as important by the relevant academic communities – could be the temporality aspect for smoking cessation. Provides the scope for the ‘work of others’ literature review.

Work of Others: What others in the research community have done to address the problem

Gap: Shortcomings of the work of others in addressing the problem. Note that there is some interplay between the gap and the problem. Some things may be interchangeable between the two depending on the story to be told

Research Question: A re-articulation of the gap as a question

Basically, I’m going to be exploring the ways that micro-blogging can help improve the timeliness of support for quitting smokers in online communities. I will be guided in my investigation by Fabre et al.’s (2000) theory of the dimensions of the Temporal Aspects of usability: duration, location, frequency and contingency.

I will conduct a trial that will involve the participants using twitter for a fortnight. Participants will be groups of 3 or 4 that have had some interaction with each other within the QuitNet online community. Through observing their feeds, I will be able to see the frequency and context of posts between the participants as the use twitter to assist them in quitting smoking.

References

Fabre, J. (2000). Designing time at the user interface. Behaviour & Information Technology, 19(6), 451-463.

Drawing on studies of context awareness and CSCW, Neustaedter et al. develop a model for interpersonal awareness through a qualitative analysis of the relationships of a number of participants. Relationships were put into three categories. These are placed on a spectrum in order of decreasing need for awareness and decreasing need for detail: home inhabitants, intimate socials and extended socials. The analysis also explored the types of information needed and the reasons for needing awareness. Fundamental reasons for needing awareness were distilled to co-ordination, promoting connectedness and desire to share personal knowledge. The three fundamental types of awareness information were of location, activity and status.

This study is relevant to the current problem situation as it deals with context awareness in a social context. Many studies in IS research have focussed on awareness needs in a work context. My first thoughts on possible ways to expand this model are through investigating:

• how online relationships with no face to face component fit into the model
• how support relationships fit into this model
• people’s need to express their status or convey information. Neustaedter et al. focus on the awareness of others that people require. Apart from cases involving co-ordination, they do not mention people’s own need for others to be aware of their location or status.
• how aware people are of others in an online support community, and whether increasing awareness is of any benefit

I was also thinking of looking at the relationship between interpersonal awareness and perceived social support through obtaining a subjective account of their perceived supportedness before and after the trial.

Anyway, these are thoughts that require a bit more rumination over the weekend, I think.

References

Neustaedter, C., Elliot, K., & Greenberg, S. (2006). Interpersonal awareness in the domestic realm . Proceedings of OZCHI 2006, 15-22 .

The sticking point, for many weeks now, has been deciding what the contribution of this research is, exactly. One of my supervisors initially suggested studying it from the perspective of appropriation, but that turned out to be unsuitable for a number of reasons. This was then refined to the idea of technology acceptance, with the suggestion that I look at the Technology Acceptance Model (TAM). I found a paper which included a number of models of user acceptance. In the next section I offer a brief summary of three of them, their core constructs and how I think I could use them in my study.

Before I do this, I should comment overall on what I believe to be the rationale behind investigating user acceptance (in the form of intention or actual usage) for this problem. It is just one of many possible dimensions of the phenomena that could be analysed. Firstly, an interesting issue here is to see how an existing model of user acceptance, developed for organisational contexts, can be applied to the case of technology to assist in smoking cessation. Interviews and the qualitative data from their two weeks using the technology would provide rich insights into their difficulties and attitudes towards the technology. Secondly, it will provide me with a theoretical framework to guide my enquiry which may be lacking in other approaches to the situation. Overall, insights into what effects the usage and intention to use technologies can inform more successful design and introduction of those technologies.

Following is a brief overview of three models of user acceptance, along with their core constructs as defined in Venkatesh, Morris, Davis & Davis (2003). I have attempted to note how some of these constructs may manifest in the proposed study.

Technology Acceptance Model – Extended (TAM2)

Core contstructs

• Perceived usefulness
  • How much they think it will help them in their quit attempt
• Perceived ease of use
  • How easy it is to use the technology to post (eg. usability issues with sms, web forms)
  • How the posting and maintenance of awareness fits in with their life in the world at large. ie does it become irritating to receive updates via sms?
• Subjective norm (perception that important others believe they should use the technology)
  • How do others in the community perceive its use

The following two models come from sociology and psychology and have been used to model smoking cessation as well. I’m not sure at this stage if these parallels are useful in any way – are the two behaviours entirely separate or do they become in some way intertwined?

Motivation Model

Core Constructs

• Extrinsic motivation – motivation through belief that the behaviour will achieve valued outcomes
  • The extent to which users believe that using the system will help them in their quit attempt
• Intrinsic motivation – motivation through a desire to perform the activity
  • The extent to which users ‘feel better’ when using the system, eg when posting, reading others’ posts.

Social Cognitive Theory

Core Constructs

• Outcome expectations
  • Do they expect the technology to help them in their quit attempt? (closely tied to their expectations re: their quit attempt in general)
• Self-efficacy
  • How confident they are at using the technology
• Affect (individual’s liking for a particular behaviour)
  • How do they feel when using the technology
    • Do they feel better after posting?
    • Do they feel comfort reading the posts of others?
• Anxiety (anxious responses to do with performing a behaviour)
  • In tech context this can represent, for example, fear of using a technology.
  • Subjects will likely be anxious when posting, but not about the posting itself.

Possible issues with studying user acceptance

Operationalising the constructs

Validated measures of the constructs for the TAM2 were developed for organisational contexts and may not be appropriate for the smoking cessation context. The existing validated measures are in the form of questionnaires. If I couldn’t use the questionnaires at all, would I just code my qualitative data keeping these constructs in mind and validate my measures through sound reasoning? There seems to be a tension here between positivist/quantitative research, which is what has typically been done to explore models of user acceptance, and the qualitative research I am proposing.

‘Actual usage’ will not be a useful construct since I will have requested that they use the system. I could only really measure their intention to keep using the technology after the trial has been completed, and this may not be an entirely useful measure. Measuring this beforehand will be useless since they will have signed up with the intent to use the system.

Time-frame too short, contrived conditions, selection bias

The time-frame I am dealing with makes it impractical to see if the technology is actually adopted. Also, the particpants have signed up, we would assume, with an expectation that the technology will be effective in some way, so they would not be a representative sample.

Not enough of a social emphasis

Usefulness/relevance

I have my doubts that the research design as it is envisioned is long enough or can easily use adequate measures to explore one of these models. Also, we are looking at the acceptance of the technology before we have ascertained that it is actually useful in any way for this problem.

References

Dijkstra, A., De Vries, H., Kok, G., & Roijackers, J. (1999). Self-evaluation and motivation to change: social cognitive constructs in smoking cessation. Psychology and Health, 14, 747–759.

Venkatesh, V., Morris, M.G., Davis, G.B., & Davis, F.D. (2003). USER ACCEPTANCE OF INFORMATION TECHNOLOGY: TOWARD A UNIFIED VIEW. MIS Quarterly, 27(3), 425-478. Retrieved June 18, 2007, from http://search.ebscohost.com/login.aspx?direct=true&db=buh&AN=10758835&site=ehost-live.

I have a meeting with my supervisors tomorrow, and I really haven’t made much progress since the last meeting. If anything, things have gotten slightly worth with the news that I won’t be able to get ethics approval any earlier than late july. The 2 week trial of twitter with international participants that I have been planning may just be too risky to do in the time frame I’ve got. I need a way to get the data I need out of some simple interviews, but who would I interview, and how would that help me with my research problem?

Questions, questions…. early rise tomorrow to have a think about it.

What kind of study is this?

Last week I had to submit an assignment on my intended research methods for my project. It was a bit of a struggle trying to fit HCI research instruments like technology probes into the strict Information Systems research categories like ‘Field study’, ‘Case study’, ‘Ethnography’ and ‘Experiment’. I suggested to my supervisor that this is possibly an interpretive field study. Instead, he suggested that however I label my study, it should mention that it is qualitative, online and involves polling/probing in the form of an interview. It may be that there is an existing term to describe this kind of research, or I may have to adapt an existing term or come up with my own. He also said that what I call it will depend on the audience.

Justifying the number of participants

In order to justify the number of participants before the analysis of the data, I need to make a pragmatic argument and also refer to examples of similar research. After the research, I need to make an argument that some kind of theoretical saturation had been achieved from the data, that an analysis of more participants would not reveal any more insights.

From appropriation to adoption

After investigating appropriation it was looking like it was too ill-defined a term. There are few theories about technology appropriation. One that looked like it may be useful is only concerned with whether or not a particular technology is appropriated and the reasons behind its appropriation, rejection and ‘disappropriation’ (Just what do the youth of today want? Carroll et al. 2002). Such theories do not probe to the level of providing a rich account. Also, given the short timeframe available to me to collect data (2 weeks), it is unlikely that the full process of appropriation will run its course.

My supervisor then suggested looking at technology adoption theories instead, which are a lot more well-defined and focus on the stage where technology begins to be used. I asked whether the theory would simply be a tool or whether it was intended for my contribution to be testing how adoption models relate in the context of technology support for behavioural change. The answer I got was, effectively ‘both, but possibly only the first one’. I would use the theory as a way of analysing and understanding their behaviour (in addition to my qualitative analysis). Essentially, it is a tool for investigation (the qualitative analysis to follow would complement the perspective given by this tool). I’m still unsure as to whether this is the appropriate tool if I don’t intend for my contribution to be to adoption theory. I need to think some more about this and clear this up.

Data Analysis

We discussed hermeneutics and ‘open and axial coding’. Steve seemed to think that open and axial coding was more appropriate. His take on the hermeneutic approach was that it would take too long as it involves continually collecting data as you progress through the ‘hermeneutic circle’. I don’t recall that constantly pulling in new data was part of hermeneutics, I thought the principle was to become immersed in the data.

Open and axial coding is generally used as part of a ‘grounded theory’ approach, where the theory emerges from the data. Peter suggested I read a book by Strauss & Corben on grounded theory, which is apparently a practical guide to using grounded theory.

Contribution

I think I’ve been somehow assuming that having theory emerge from the qualitative analysis of the data means that you don’t have to define what your contribution is beforehand. I’m starting to see that that’s a little misguided. What would emerge would be a theory describing something in the area of contribution you have defined. Some possible areas for contribution that were talked about today:

• methodology – how to analyse micro-blogging data
• models of technology acceptance (adjusting the TAM)

My assumption about the contribution was that it would be in the form of design guidelines (which has an implicit assumption that they actually use it) for technology for behavioural change and for micro-blogging in particular in this context. I never seem to be able to articulate this in a way that seems sufficient

Unfinalisability

I spoke to a researcher in the department this week, Tuck Leong, about the concept of unfinalisability. It may be relevant if it seems like the micro-blog is being used for varios purposes. Each of these uses may require the user to express themselves in a different way (eg. reflect, request social support). It seems that it is the kind of thing that I could consider as a minor point in the discussion section of my thesis.

What next?

Well, Steve suggested I find some models of technology acceptance to employ. It seems to be a constant struggle to see how his suggestions fit into my expectations for this thesis. Whenever I have an approach that I’m entirely comfortable with, the question that invariably comes back to haunt me is ‘what is the contribution’. If I’m going to come up with a research approach I need to be able to answer this question clearly.