Behavior Designing for Confidence in the Cryptocurrency and Blockchain Space

Motivation For Study:

I found that while I was confident navigating making important decisions in some aspects of my life, that I was not as confident in making crypto-asset investment decisions. I found myself completely captivated by cryptocurrencies and blockchain but somehow still I was lacking a huge amount of confidence, and I saw how this was affecting my behavior in terms of engaging with others in the space and also in pursuing my own interests in this space. In an attempt to address this problem I went on a pursuit of as much knowledge in the space as I could get. As an undergraduate junior I weaseled my way into one of the most popular GSB classes "Cryptocurrencies” taught by Susan Athey and Kathryn Haun. After taking the class, I continued to do more reading in the space, joined Stanford’s blockchain mailing list, and started to connect with others in the space. However, I realized that while gaining more knowledge in the space definitely increased my confidence to an extent, I still was not very confident about my place in blockchain, and realized that, as compared to my other endeavours in Computer Science, that I was missing critical feedback and affirmations in this space.

For this reason, I sought out to fix my own problem by starting to understand confidence some more. This spring quarter in BJ Fogg’s Behavior Design Lab, I started to research confidence and I applied behavior design principles to pilot a study on confidence. With the field of confidence being not very well fleshed out, I had a lot of thinking to do and a lot of mistakes to make before getting to the final study design. After much work and dedication we were able to come up with a final experiment design that can not only be applied to studying confidence, but also to many other kinds of behavior change research.

Carving out the Confidence Space Confidence is defined as a sense of self-efficacy tied to one’s self-belief and self-trust to achieve some goal. We know that boosting people's confidence influences people to be more likely to engage with what is asked of them. It has also been found in research that the sex difference in confidence depends upon situational variables such as the specific ability area, the availability of performance feedback, and the emphasis placed upon social comparison or evaluation. We also consider the main different categories of confidence boosts, namely encouragement and affirmations. We define affirmation as ‘fact based persuasion’, as compared to encouragement which is defined as ‘a form of boosting that does not include or reference any relevant facts’.

Journey to the Right Study Design: In attempting to find a way to investigate confidence from a behavioral standpoint, we considered several different approaches. When discussing what makes someone confidence, the difference between confidence and self esteem, the different types of confidence boosts, the different ways to express these different types of confidence boosts, and all the other nuances around confidence that we could think of in lab, it was incredibly easy to fall into a rabbit hole. In exploring all the different options when it came to confidence boosting we were able to canvas the space and flesh out several different nuances that we may encounter when we think of confidence. Before we knew it, we had curated an entire repertoire of experiment designs ranging from manipulating encouragement vs affirmations vs no confidence boosts, to manipulating the importance level with which the study participant considers the topic that the study focuses on, to manipulating the different types of affirmations that can be given to a person based on BJ Fogg’s 32 unit grid on framing success. We limboed between in-person interviews, online surveys, and SMS as mediums through which we could carry out our study with participants. We considered confidence profiling - profiling what exactly makes one individual feel confident, versus what kind of confidence boost may allow another type of person to feel confident. We considered several different success metrics that the different communication mediums would allow for, and in true Stanford undergraduate style, created a beast of an overly ambitious study that would take somewhere in the ballpark of 4 years (at least) to complete. From the proposing models of what influences confidence (encouragement, knowledge/expertise level, affirmations, self esteem), to thinking about how these confidence influencers could vary cross-culturally, I finally crashed. I went from riding the confidence wave to being soberingly washed ashore to the land of one-experimental-maniupulation-at-a-time, and then took it home from there.

Study Protocol: The goal of our study is to investigate how an effective boost in confidence may affect a study participant’s willingness to carry out an action in alignment with their specific area of interest within cryptocurrencies and blockchain. We decided on communicating with study participants over SMS text message because people are more responsive to text messages than emails and other mediums of communication, and also because SMS would allow for a more conversational approach than an online survey (so as to get people’s opinions on different topics in the blockchain space), while also allowing more quantitative metrics (such as response time and drop-off) than an in-person conversation would allow. Furthermore, we designed the conversation such that it would contain an easy question followed by a hard question. Both questions were open-ended questions, so as to allow the study participants to share their thoughts on the space, rather than giving black and white responses. Open-ended questions were preferred over closed-ended questions, as they would allow for more flexibility in affirming the study participants, and also allows us to have challenging conversations with study participants at a varying different levels of expertise in the area (since there is no single ‘right’ answer to these questions). Easy questions are open-ended questions which can be answered with a simple ‘yes’ or ‘no’ (even though the study participant may have a longer response -perhaps up to even a paragraph long, depending on the level of engagement), while hard questions can not be answered with a simple ‘yes’ or ‘no’ response.

​​​​​1) Easy Question Example:

Do you think that there is potential for bitcoin to be used in the banking & financial services sector?

2) Hard Question Example:

What would you say to someone who says bitcoin may be a pump and dump scheme?

3) Throughout the experiment we track the trajectory of the person’s confidence level, and we manipulate whether or not the person receives a confidence boost.

4) At the end we give the study participant a specific task to complete and use the completion of this task to determine the conversion rate, thus achieving our goal of investigating how a boost in confidence may affect a person’s willingness to do a behavior in alignment with their interests.

This conversational design is used throughout the entire experiment and the study is designed to be run over a period of 3 consecutive days.

Ethical Considerations: Confidence boosting and profiling ought to be used for good. Everything that we study in the lab we do with the intent for the insights to be used for good. When integrating confidence boosting into a service, user experience or any other application, no false confidence boosts should be given. Confidence boosts, such as affirmations, should be truthful so as to not give anyone a false sense of self-efficacy. On the other hand, insights into confidence should not be used to tear down anyone’s sense of efficacy either. Confidence boosts are meant to be applied in situations in which we are helping someone to achieve a positive goal and/or aspiration for themselves (For example, helping women who have the ability but perhaps not the belief that they can be financially secure).