…and we’re back

After a bit of hard work over the weekend, and half a dozen interviews, things seem to be back on track.

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!

Change of plan

A number of factors, time being the most prominent one, are forcing me to abandon my plan to run a trial as part of my thesis. The sending of an internal message to members of the online community I was hoping to recruit from has taken longer than anticipated. There have also been some unexpected complications, such as the fact that text messaging is not included by default in many North American mobile phone plans. Short of a miracle occurring in the next couple of days, I think I have to lay the plan for a trial to rest, and come up with something else quickly and smartly.

Technology Acceptance

Well, I’m pretty sure that my research design is locked in. I will recruit 3 or 4 groups of 3 from an online quit smoking support community and get them to sign up to twitter. I’ll ask them to send updates to their mini-blog as to how they feel, what they’re thinking etc (I should be careful not to specify that their posts should only be smoking related). All the members of the group will be able to see each other’s updates.

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.

A bit of a break

After getting soaked up in university assignments, job interviews, Aikido gradings and mixing songs in the studio, I just had to have a couple of days off.

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.

Oh, clarity, I barely knew ye …

Currently trying to put the finishing touches on my research proposal. Keep losing the ability to tell what is important. I’ve read so much and feel that I need a few days for everything to digest in my subconscious. The research proposal as it stands feels a bit thrown together. There are some gaping holes and things just aren’t hanging together as nicely as I’d like them to. I’m not sure whether to look into appropriation literature, or just to give an account of the appropriation in the form of a summary and analysis of user behaviour. I suppose I should at least find out the best way to do this. The big problems I’m coming up against here is that I’m dealing with a lot of really new things that have hardly been covered in the literature. Technologies for behavioural change, pervasive technologies, API based web services (technologies that are in a sense designed solely for appropriation, there is little assumption about how they should be used). All of this is pretty new, so I’m finding it a bit hard to ground myself in the literature. Obviously, there is nothing at all on micro-blogging, so I need to look at what I can – existing technologies for behavioural change, technologies similar to micro-blogging (maybe….).

Hopefully I can get my proposal looking respectable. I think it’s almost there. I realise that at regular intervals through my research I’ve been waiting for times when things become clear. They seem to be few and far between. I put a lot of work into my presentation last week, but most of that work, while good for my background knowledge, is not helping me at all right now. I suppose I just need to get used to this state of incompletion, or lay down some ‘milestones for understanding’ so that I can feel like I’m getting somewhere. First milestone, what is appropriation in this context and how should I deal with it?

hmm…

This is the research question

Just had a meeting with my supervisor and have settled on one of the options from the last post. Drumroll, please:

How is micro-blogging appropriated for smoking cessation?

The basic research design is still similar. I will get three groups of three people from an online community and place them into groups on twitter. At the end of a two-week trial I will interview all of the participants and then code and perform a simple statistical analysis of their posts.

Rather than using psychological models of behavioural change as a lense for my enquiry, I will be using the characteristics of temporality, placedness, reflexivity and social supportiveness established recently in the HCI literature. My justification for this is that these characteristics have emerged from  research that has taken the psychological models into account. I will use these as a starting point and observe how each of these characteristics is manifest in micro-blogging and hopefully refine their definitions and add any other characteristics that may become apparent along the way. In this way I am grounded more in the field of HCI.

So what is the contribution here?

• Description of micro-blogging being appropriated for smoking cessation
• Contribution to theories of technology for behavioural change
• Design guidelines for using micro-blogging to aid in smoking cessation

Basically, it is a technology trial with some theoretical outcomes rather than requirement outcomes.

Questioning research questions…

Things seem to stay stable round here for a maximum of about five minutes.

After settling on the following research question:

What effect does interpersonal awareness have on self-efficacy for behavioural change?

I became increasingly concerned that my view of self-efficacy, the confidence in one’s own ability to refrain from engaging in a particular behaviour, was too simplistic. I believed it was something that could be easily measured and something that needed to be high in order to effect behavioural change. After reading a few papers and meeting with Ron Borland at the Cancer Council Today, I have realised that Self-Efficacy, and indeed many of the internal and external variables in behavioural change, is a strange beast. Depending on the stage of the quit attempt, and depending on the strength of other characteristics, high self-efficacy can have either a positive or negative effect. For example, high self-efficacy combined with insufficient coping skills results in the individual being overconfident and failing to refrain from the behaviour. Also, with a small sample size a measure of this kind would be next to meaningless.

Ron then suggested a different kind of study, where I could attempt to identify what characteristics of micro-blogging address the psychological constructs known to be important in behaviour change. I could get my participants to sign up to twitter and use it for a couple of weeks. I could then observe their posting behaviour and survey/interview them with a view to finding out how the technology was used in relation to those key constructs. For example, did they use it to reflect on their position or to request help. Did they feel it was useful? Do they feel it had an effect on the outcome? While I was taken with this idea for a couple of hours, when talking it over with one of my supervisors I was alerted to the fact that this is merely a technology review. In order to make a contribution to the literature I need to go a step further. Being able to elicit some design guidelines from the data may be enough, but it may not be. I will find out tomorrow when I talk to my main supervisor.

I have since fallen into the idea of focusing on the concepts of temporality, placedness, reflexivity and social supportiveness, which have been identified as possible important characteristics when designing technology to assist in behavioural change. Rather than using the psychological concepts as a lense, I could use these as a lense to probe the users’ experiences and see if they are sufficient, what they mean in the context of micro-blogging (how they manifest) and how micro-blogging services could be better to designed to enhance these characteristics and any others which become evident through the investigation. Again, I’m not sure how significant the contribution is here. The only real difference between this and the previous option is the choice of lense. In this one I am taking for granted some design guidelines and seeing how they hold up and can possibly be elaborated upon or added to. In the previous one I am making no assumptions on design guidelines, instead I am starting from principles of behaviour change and hopefully waiting for design guidelines to emerge. It may be that the existing research into technology for behavioural change has already taken this into account and there is no need for me to dive to that level again. Perhaps it is useful to refine, elaborate and extend the existing guidelines in the literature.

Ok, so, any other big ideas before I sign off? I think some kind of contribution related to design guidelines is a useful one. I’m pretty much following the model of Consolvo et al’s (2006) trial of their mobile phone application to encourage physical activity, only in my case I am not designing the artefact to be tested, I’m just taking advantage of an existing one – just being a connectionist. Whether this contribution is significant is the subject of tomorrow morning’s meeting.

Gotta get this straight…

Whoa. I have read way too many articles today.

I’m about to try to sort out this whole reciprocity issue. The way it is at the moment, it would be pretty straightforward for me to design a solution taking into account everything I’ve read about reciprocity, which would mean completing option 1 from my previous post. Before I decide on doing that, though, I’d like to see if there’s an opportunity for me to make any contribution to reciprocation literature (option 2 from my previous post). I’ve got a few leads, but I keep procrastinating when it comes time to try to focus. Reminds me of meditation.

At least now I know I’ve got option 1 as a safety net. I’m a bit concerned that I don’t seem to have one paper in my collection that spells it out, though.

Anyway, this is the last bit of procrastination before I dive in. Here goes nothing….

Thesis topic crunch time: theoretical indigestion

The context

A couple of weeks ago my research question looked something like this:

How can existing online social networks be used to assist in behavioural change?

The context for my investigation is smoking cessation. So far I’ve been focusing on how online social networks are being used for social support. I’ve found that some substantial research has been done into social support online (Maloney-Krichmar & Preece 2005), but little has focused on extending this support to pervasive technologies such as mobile phones. Some studies in the field of Human-Computer Interaction (HCI) – which is where my thesis is supposed to be placed – have investigated text message support services for behavioural change, while others have looked at social support in particular using mobile applications (Consolvo et al. 2006). A study by my supervisors (Graham et al 2006) found that temporality, placedness and reflexivity were important factors to consider when designing technologies to aid behavioural change.

To my knowledge, none of the existing online social support networks are addressing the concepts of temporality and placedness (context awareness). All of them allow interaction only through the website – unless you are fortunate enough to have the internet on your mobile phone, you can’t access support when and where you need it. So I started thinking about how peers in a social network could start to address these temporality and placedness issues – delivering emotional or even informational support when the concerned individual needs it, with an awareness of the person’s context. This is not to say that the technology does not exist, only that it is not yet being effectively employed towards the end of social support.

Web 2.0 services like twitter and facebook allow users to post their status from their mobile phone while they are on the go. A quitting smoker could post their status (having a good/bad day, in need of help etc) to the online group, which could observe the stream of messages coming from community members. Members that are online could reply, sending messages of support. There are a whole host of other ways to exploit this technology in this context (eg buddies, subscribing to users’ feeds) but for the moment I’m trying to keep my focus small, lest my imagination run away with me. The concept underlying all this has something to do with reciprocity, either between individuals or between the individual and the group.

Theoretical indigestion

The idea of reciprocity between an individual and the network came to me while reading Barry Wellman’s Network capital in a multi-level world (2001). The paper is a multi-level analysis of social support. It is multi-level in the sense that it looks at the individual, relational and network characteristics of a social support network, and tries to observe any emergent network properties, instances where the ‘whole becomes greater than the sum of its parts’. Although the paper was more a demonstration of the technique of multi-level analysis, some interesting conclusions were drawn on the nature of social support:

• Reciprocity is a multi-level phenomenon. Small acts are likely to be reciprocated quickly by the receiver of support, whereas more substantial support may be reciprocated at some point by some other part of the network. A family is a good example of this latter form of reciprocal support.
• Particular kinds of relational ties are more likely to be supportive when embedded in a network of similar ties. These are cases where network capital is generated:
  • Less intimate but active relational ties between individuals were more likely to be supportive in a network with more active ties
  • Ties with high accessibility (eg ease, frequency of contact) were more likely to provide support when embedded in a network of accessible ties.

I have the feeling that both of these observations will be useful to me at some point, but for the moment I’ll try to focus on reciprocity. When interpreting Wellman’s conclusions, it is important to keep in mind that the data it draws from is based on very broad definitions of emergency support and everyday support, and that the sample was from a middle-class British-Canadian population in a Toronto suburb in 1968. Is the emotional support needed by a quitting smoker emergency or everyday support? Could patterns of reciprocity have changed since 1968?

The next step

So perhaps now it is time for me to find a theory of reciprocity (or some other related concept) and see if it is evident in online social support networks. If a theory already exists for technology-assisted reciprocity, then this will be used, otherwise a theory of face to face reciprocity will be employed.

Once the theory has been selected, I have two choices (as suggested by my supervisors):

Option 1: Contribution to user interface literature

In this option I would look at the theory and see how web 2.0 techniques could be used to employ the theory. My empirical component would then be a laboratory study comparing a ’standard’ web 2.0 technology, such as twitter, with a solution which I would develop. While part of me would like to develop something like this, I have my reservations

Pessimisms

• A laboratory setting may not reflect the usability issues that would become apparent in the real world
• I may not be able to develop a solution in time
• Unreliability of the technology. Facebook does not allow text messaging from Australia and Twitter has currently disabled text messaging from Australia. (This may not be a problem – in a lab environment we could just use IM to simulate messages)
• The contribution to UI literature may not be all that significant. People are churning out web2.0 interface designs.

Optimisms

• Output may actually be significant – ‘designing for reciprocity in online social support networks’. Output would be what works and doesn’t work in the UI design.
• I would get to do some development, have a product at the end
• Designing – may be easier to tap into some kind of ‘flow’

Questions

• Would there have to be a couple of different prototype designs to compare?

Option 2: Contribution to reciprocity literature

In this option I would look at the theory and do some empirical study of existing online social support networks. For example, I might look at a number of posts over a period of a week and observe reciprocation between community members. I could then see if the theory was verified or whether it needs to be extended to account for the observed behaviour.

Pessimisms

• This may only be useful if no study has already been done into reciprocation in online support communities (suppose I better find out, huh?)
• I won’t get any technical experience with the new technology. This may effect my employment options after graduation

Optimisms

• Providing it hasn’t already been done, this would be a meaningful contribution

Questions

• Would I still design a prototype GUI?

Where is smoking cessation in all this?

It seems like the context of smoking cessation may be becoming irrelevant in all this. My intuitive assumption that smokers will need more immediate support than members of other social support groups (eg depression, injury recovery) may actually be incorrect. If it is incorrect, then we could use any context. If I do go ahead and use the smoking cessation context, I will have to justify my reason for doing so, citing reasons other than the fact that there is research being done in this context in my department. Alternatively, I may be able to argue that temporality and placedness are in fact more important for smoking cessation in particular, since the consequences of missing support (ie. a relapse) are more severe than in other contexts.

Do you care?

I honestly don’t think anything I’ve written today will be useful (or comprehensible) to anyone other than myself, but I certainly do feel a lot less confused for having written it. Hopefully once my theory and methods are locked down I can start writing posts that are more useful to more people