Fascinating — and a little frightening: measuring acceptance of modular smartphones

Overview

Modular smartphones promise sustainability and personalisation: swap the camera, the battery or the storage instead of replacing the whole device. Google tried it with Project Ara and cancelled it in 2016. So the question we set out to answer wasn’t can it be built — it was would people actually accept it?

Working in a team of three for the Usability & UX course of our MSc, we designed and ran a technology acceptance study grounded in the TAM3 model: a physical prototype, an explanatory video, and a questionnaire that collected 136 valid responses from the Portuguese population.

The headline finding is a paradox. Perceived Enjoyment scored highest of all constructs (mean 5.50 on a 7-point scale) — and Computer Anxiety came second (5.43). People found the concept fascinating and intimidating at almost exactly the same intensity. That tension, not raw enthusiasm, is what would decide adoption.

Why modular phones?

A modular phone lets the user replace individual components — camera, battery, processor — instead of discarding the entire device. The appeal maps onto circular economy principles: extending one year of a phone’s useful life measurably reduces the CO₂ emissions tied to its production and disposal.

The obstacles are just as real: universal connectors are technically immature, modularity fights against waterproofing and slim aesthetics, and the business model of the industry is built on replacement cycles, not longevity.

Project Ara is the emblematic case. Google’s ATAP division planned developer kits for 2016 and a consumer launch in 2017, then cancelled the project over connector reliability and production costs. We used Ara as our reference concept — the study asks whether the idea it represented would land with Portuguese consumers today.

Research design

Why TAM3

The Technology Acceptance Model 3 (Venkatesh & Bala, 2008) explains adoption through two core beliefs — Perceived Usefulness and Perceived Ease of Use — and models the factors that feed them: self-efficacy, anxiety, enjoyment, external support, job relevance. For a product that doesn’t exist on the market yet, this was the right lens: it measures dispositions toward a technology rather than experience with one.

The questionnaire operationalised nine TAM3 constructs, two items each — 18 Likert items on a 7-point scale (1 = strongly disagree, 7 = strongly agree): Perceived Usefulness, Perceived Ease of Use, Computer Self-Efficacy, Perception of External Control, Computer Anxiety, Perceived Enjoyment, Objective Usability, Result Demonstrability and Job Relevance. Additional questions captured phone experience, current device, future purchase intention and perceived usability versus current phones; items were deliberately interleaved as a comprehension check.

We formulated five hypotheses before fielding, centred on whether education level and tech familiarity shape acceptance — for example, whether participants without higher education would show more enthusiasm than those with degrees (spoiler: they did).

Making the invisible tangible

You can’t ask people to judge a product they can’t picture. Before the questions, respondents watched a video we produced ourselves, built around a physical prototype of our own modular phone concept.

The video runs through six scenes designed to show range, not features: an elderly woman with a stripped-down basic configuration, a gamer who snapped controller buttons onto the back, a photography influencer, an athlete with a health-monitoring module — closing with an honest recap of pros and cons, and a prompt asking viewers what they would build. Each persona answers the same implicit question: what does modularity do for someone who isn’t you?

Fielding

The questionnaire ran on Google Forms across seven pages — framing, concept introduction, the video, scale explanation, then the items. After collection we cleaned the data (removing straight-lined responses, normalising open answers), leaving 136 valid responses. The sample skewed female (70.6%), young (19–25 was by far the largest age bracket) and educated (51.5% with a bachelor’s degree) — a limitation we return to below.

Findings

The enthusiasm–anxiety paradox

Ranking the construct means tells the story in one glance:

• Perceived Enjoyment — 5.50 (median 6.00): the strongest signal in the study. People find the concept genuinely fascinating.
• Computer Anxiety — 5.43 (median 5.50): almost as strong. The same people fear breaking something when swapping modules.
• Behavioral Intention to Use — 5.25: a real willingness to adopt.
• Perception of External Control — 4.43: the weakest construct. Respondents don’t trust that manuals, tutorials and support will actually be there for them.

The design implication writes itself: the barrier isn’t desirability, it’s confidence. A modular phone succeeds or fails on whether swapping a module feels as safe as plugging in a USB cable — and on visible, credible support infrastructure.

Intention is high; commitment is not

Asked directly whether they’d consider buying a modular phone, 53.8% answered “maybe”, 31.1% “yes” and only 15.2% “no”. And when asked to compare usability against current phones, the sample split almost perfectly: 59 said “better” — 58 said “I don’t know enough to have an opinion.”

That second number matters more than the first. Half the market can’t evaluate the product category at all yet. Any launch strategy is an education strategy first.

Education changes the shape of acceptance

Splitting the sample by higher education produced the study’s most actionable contrast:

• Participants without higher education were more enthusiastic — Perceived Enjoyment 5.82 vs 5.39, Behavioral Intention 5.54 vs 5.14 — and simultaneously more anxious (Computer Anxiety 5.65 vs 5.35).
• Participants with higher education were more reserved across the board, suggesting a more analytical, show-me-the-value stance.
• Both groups distrust external support equally: Perception of External Control bottomed out at 4.41–4.47 regardless of education.

Two audiences, two strategies: the degree-holding group needs proof of practical value; the non-degree group needs reassurance and intuitive design to convert their enthusiasm before anxiety kills it.

Learnings

Descriptive data can frame hypotheses — it can’t test them. We formulated five hypotheses but the analysis stayed at means and medians; no inferential tests were run, so group differences like 5.82 vs 5.39 are patterns, not confirmed effects. Running Mann-Whitney U tests on the education split would be my first move in a follow-up — the same lesson about statistical honesty I carried from reporting a non-significant Wilcoxon result in the CP Para Todos study.

A sample is a lens, not a mirror. 70.6% female, dominated by 19–25-year-olds, university-heavy: our “Portuguese population” was really “people our questionnaire could reach”. The report acknowledges this openly, and it disciplines every claim above.

Showing beats describing. The video-plus-prototype approach is the piece of this study I’d defend hardest. Acceptance research on hypothetical products lives or dies on whether respondents are all imagining the same product — the six-scene structure gave 136 people a shared, concrete referent in just over a minute of their attention.

Measure the fear, not just the desire. If we had only asked TAM’s classic usefulness and ease-of-use questions, this study would have concluded “people want modular phones”. Including Computer Anxiety and Perception of External Control is what surfaced the real adoption barrier — a reminder that the constructs you don’t measure are the findings you don’t get.