top of page
.png)
Gamification Archetype Validations
Identify an archetype that is different from conventional gamification models and validate it using online survey tools in anticipation of its implementation on a distributed energy resource (DER) agreement platform - TESS
Lim, Reegine., Baltaduonis, Rimvydas., Malik, Mayank., and Chassin, David. P. (2024).
Evaluative Research
Quant Methods
Survey Model
🖼 BACKGROUND
Project Kickoff
TESS (Transactive Energy Service System) is an essential service designed to help everyone become more conscious of their utility consumption while simultaneously improving energy efficiency. A key requirement is to engage participants with the system so they remain consistently informed about the real-time energy market, enabling them to manage their household devices effectively and achieve more affordable utility costs.
Project Goals
-
Explore archetypes for energy consumer products
-
Understand how consumer wants to engage with energy consumer products
-
Identify features designed to engage energy consumers effectively
-
Map out gamification strategies that cater to the primary archetypes
-
Ideate ways in which an energy product like TESS can enhance the experience of energy consumers in achieving more affordable utility costs.
Evaluative Research Methodologies
-
Survey Planning
-
Literature Review
-
Competitor Analysis
-
Data Re-evaluation
-
Data Analysis
-
Regression Analysis
-
Product Ideation
-
Journey Map
-
Storyboard
👩💻🔎
📊🔎
🧑🎨🔎
Understand the shortcomings of Bartle Taxanomy of Player Types
Data collection via SurveyPro & Prolific
Identify features for product design based on identified archetypes
Plan and design the survey procedures
Analyze data with Marimo.io
Define the archetypes through visuals
Ran survey analysis n=1500 in total
Ran survey analysis, re-evaluation, and regression
Create survey app via Marimo.io with an intern
📚 IDEATION | Survey Inspiration
Bartle Taxonomy of Player Types, 1996
Bartle's taxonomy categorizes players into four types — Achievers, Explorers, Socializers, and Killers — based on their motivations and preferred activities in online games. Bartle calculates player types based on responses to the quiz questions, which are designed to measure preferences and motivations related to gameplay. Each response is scored and aggregated to determine a player's tendencies. The final scores reflect the extent to which a player aligns with each type, helping to identify their primary and secondary player types.
Achiever
Achievers regard points-gathering and rising in levels as their main goal, and all is ultimately subserviant to this.
Socializer
Socialisers are interested in people, and what they have to say.
Explorer
Explorers delight in having the game expose its internal machinations to them.
Killer
Killers get their kicks from imposing themselves on others.
📚 IDEATION | Survey Context
Create universal comprehensive survey questionnaire
Gaming Context
Achiever, Explorer, Socializer, Killer
We aim to identify archetypes applicable to non-game domains, in contrast to Bartle's study, which centers on gamer psychology. The survey we designed covers a broader range of topics and offers more diverse options, rather than focusing exclusively on gaming patterns.
Non-game Context
Achiever, Explorer, Socializer, Influencer
📚 IDEATION | Survey Diversity
Instead of binary question, we provide 4 diversity options
Bartle's Test of Gamer Psychology consists of 30 questions, each offering binary choices. In contrast, we opted to provide 4 diverse options per question, with each option representing one of the 4 proposed archetypes.
Bartle's Questions
Binary Questions
Our Questions
4 Diverse Options
🔎 Take a look at how researchers define the survey questions
Assessed by
8 Behavioral Questions
🕵🏼♂️
1️⃣
100 Survey Validators
Validators are assigned to label options provided based on the questions with four proposed archetypes
Researchers used the Delphi method to assess the agreement threshold, and found that 81.25% of our questions & options achieved 75% or higher agreement among the validators 🎉
Distributed
Assessed by validators
Demographic Questions
-
Susceptibility
-
Tech Savviness
-
Risk Aversion
-
Age Group
-
Location
-
Home Ownership
Distributed
🧑🏼💼
2️⃣
500 Survey Participants
Distributed
Preferred-Action Question
The same survey was given to the same participants two months later, without their knowledge
Participants must select an option that involves taking action according to the archetype they choose
💡 PLANNING | Recruitment
How do researchers recruit survey participants?
Behavioral and preferred-action questions, which include a 75% agreement threshold, are distributed alongside demographic questions to 500 survey participants through Prolific (survey recruitment platform).
💡 PLANNING | Survey Data Allocation
How do researchers tabulate the collected data?
Each response provides a score of +1 or -1 in two dimensions, which are then averaged to determine the individual locus. Using these axes, quadrants are formed, categorizing participants into one of four types: Achiever (++), Influencer (-+), Socializer (--), and Explorer (+-).
💡 PLANNING | Design & Procedures
What do researchers aim to discover from the data?
Based on the three categories (behavior, demographic, and preferred-action questions), researchers came up with hypotheses:
-
An individual’s personality type can be determined from self-reported data about themselves
-
An individual’s personality type can predict their choices
-
An individual’s personality type corresponds to an archetype that is stable over time
Identified archetype from Behavioral Questions
Predict?
Identified archetype from Preferred-Action Questions
Demographic
Questions
Predict?
Identified archetype from Preferred-Action Questions
🔎 What methods researchers took to evaluate the hypotheses?
📊 Methodology
1. Data Analysis (n=500)
Both the simple average as well as the weighted average are computed based on the validator's weights assigned to each answer type to determine the locus point of a participant’s answers.
Without Validators' Weight
With Validators' Weight
📊 Methodology
3. Correlations
Table shows weak correlations within tech savviness, except for other savviness (how others view you) and self savviness (how you view yourself), which have a strong positive correlation of 0.92.
📊 Methodology
2. Data Re-evaluation (n=287)
With 57.4% of participants re-participating in the survey two months later, the shift in the sample focus was not significant, as the distribution between achievers and explorers remained nearly the same.
Without Validators' Weight
With Validators' Weight
📊 Methodology
4. Regression Analysis (Age & Sex)
The collected data is analyzed with a multinomial logistic regression, where the dependent variable is the individual’s choice and explanatory variables are their personality type.
It is confirmed that personality type predicts actions, except for the socializer
Age significantly explains some archetyped choices, especially among older age groups
📊 Methodology
5. Regression Analysis (Predictive Action Choice)
Regression results that capture the predictive margin of the action choices given the personality type. Significant action ordering patterns were found for achievers and influencers, with choices aligning with their archetypes, but not for socializers, possibly due to survey limitations.
Dominant is bold and statistically insignificant is in red with p-values provided under each estimate
✨ What conclusions can researchers draw from this research?
❓ Hypotheses
1. An individual’s personality type can be determined from self-reported data about themselves
Note1: Age is a significant explanatory variable for some archetyped choices with the older age groups standing out the most (see regression analysis).
❓ Hypotheses
2. An individual’s personality type can predict their choices (see regression analysis)
❓ Hypotheses
3. An individual’s personality type corresponds to an archetype that is stable over time
Note2: Archetype stability is not fully consistent at the individual level but it is remarkably consistent in the aggregate (see data-re-evaluation).
📍Researchers concluded that achievers and explorers are the identified primary archetypes, do we see any limitations here?
❗️Shortcomings
Sampling Bias
Inherit from Prolific, a global survey platform may lead to an overrepresentation of exploration and achievement traits. Participants in Prolific are often more exploratory and active due to the platform's recruitment process and their motivation for financial gain, which could skew results toward these characteristics.
Data Validity
Given that some participants may answer randomly on paid platforms like Prolific, researchers are considering using experimental economics tools to enhance participant attention and data quality.
🧠 What's next
Validating the survey through various channels
For instance, distributing the survey not only via new media but also through traditional channels like mail with a larger sample size that ensures diversity.
Refine questionnaires to focus on energy
The next phase will refine the methodology for incentivized surveys in the energy sector to understand participant archetypes and energy consumption behaviors.
Incorporate latent behavior
the researchers are also interested in incorporating latent (anti-social) behavior into this model to understand how participants engage in general electricity usage when incentives are involved.
Click the image to see how designers incorporate identified archetypes into TESS product design
More of this project? Supervised an intern this summer to develop a survey application
The intern was assigned to develop a custom survey application to enhance self-reporting precision through immediate personality feedback in behavioral surveying.
The intern automated the survey app to collect personality type information immediately after submission
With Marimo, the intern developed an interface of survey questionnaires to reflect immediate personality feedback once participants hit submit button
Introduce the most talented intern - Amanda Li 👑⭐️
bottom of page