Mapping Individual Subjective Values to Product Design

This is not an empirical study testing a single intervention. Instead, it is a **methodological framework development** with illustrative case studies. The author tested whether a structured, computer-aided process could:

Map user personality traits and value orientations (e.g., "adventurous," "conservative," "socially conscious") onto measurable product design features (e.g., shape curvature, colour saturation, material texture).

Allow engineers to predict which design variations would be preferred by specific user segments.

Enable optimisation of product appearance for individual subjective preferences across three domains: bionics (prosthetic limb covers), medical engineering (blood pressure monitor housing), and mobility (bicycle frame geometry).

The "comparator" was essentially the existing ad-hoc design process (designer intuition, focus groups, or generic ergonomic guidelines) versus the ACADE-guided process. No formal statistical comparison was made between these approaches.

The dissertation does not report a single controlled experiment with human participants. Instead, it presents:

**Case study 1 (Bionics):** 30 healthy adults (age and sex not specified) who rated prosthetic limb cover designs.

**Case study 2 (Medical engineering):** 25 adults (age and sex not specified) who evaluated blood pressure monitor housing designs.

**Case study 3 (Mobility):** 20 adults (age and sex not specified) who assessed bicycle frame geometry preferences.

All participants were German-speaking adults recruited through university networks. No exclusion criteria were reported. The sample sizes are small and the populations are convenience samples, limiting generalisability.

The author developed custom instruments for each case study:

**Personality/value assessment:** A modified version of the Schwartz Value Survey (SVS, a validated 57-item questionnaire measuring 10 universal value types: power, achievement, hedonism, stimulation, self-direction, universalism, benevolence, tradition, conformity, security). Participants rated each value "as a guiding principle in my life" on a 9-point scale (-1 = opposed to my values, 0 = not important, 7 = of supreme importance).

**Product impression assessment:** A semantic differential scale where participants rated product images on 7-point bipolar adjective pairs (e.g., "modern–traditional," "aggressive–gentle," "cheap–premium"). The specific adjective pairs varied by case study but typically included 12–20 pairs.

**Design parameter measurement:** Physical dimensions of product prototypes (e.g., curvature radius in mm, colour hue in degrees on the CIELAB colour space, surface roughness in μm) were measured using CAD software and physical callipers.

**Preference ranking:** Participants ranked 5–8 design variations of each product from "most preferred" to "least preferred."

No standardised psychometric instruments for product perception (e.g., AttrakDiff, User Experience Questionnaire) were used. The author created bespoke scales for each case study, which limits cross-study comparability.

### Study design

This is a **methodological framework development** with embedded **cross-sectional case studies**. The overall structure is:

1. **Phase 1 – Theory synthesis:** The author reviewed literature from psychology (personality theory, value theory), marketing (brand personality, product semantics), and engineering design (Kansei Engineering, Quality Function Deployment) to build the ACADE framework.

2. **Phase 2 – Instrument development:** The author created a step-by-step process chain linking user values → product personality dimensions → design parameters.

3. **Phase 3 – Case study validation:** Three small cross-sectional studies where participants completed value questionnaires, rated product images, and ranked design variations. The author then tested whether the ACADE model could predict preference rankings from value scores.

### Design details

**No randomisation:** Participants were assigned to product categories based on convenience (e.g., cyclists evaluated bicycle frames). There was no random allocation to conditions.

**No blinding:** Participants knew they were evaluating product designs. The researcher was not blinded to participant value scores when analysing data.

**No control group:** There was no comparison group using a different design method (e.g., traditional focus group approach).

**Duration:** Each case study involved a single session lasting approximately 45–60 minutes. No follow-up or repeated measures were conducted.

**Statistical approach:** The author used multiple regression analysis to predict preference rankings from value scores and design parameters. Specifically, partial least squares structural equation modelling (PLS-SEM) was used to model the relationships between values, product personality perceptions, and preference. No sample size justification or power analysis was reported.

### What this design can and cannot prove

**What it can prove:**

That a systematic mapping between user values and design parameters is *possible* in principle (proof of concept).

That there are *correlations* between certain value profiles and preferences for specific design features in small, homogeneous samples.

That the ACADE framework can generate design recommendations that are internally consistent with its own assumptions.

**What it cannot prove:**

That ACADE produces *better* products than traditional design methods (no comparative trial).

That the value–preference relationships are *causal* (cross-sectional design, no experimental manipulation).

That the findings *generalise* beyond the specific products, populations, and cultural contexts tested (small convenience samples, German-speaking only).

That the preference rankings are *stable* over time (single session, no test-retest reliability).

That the design parameters identified are the *only* or *most important* drivers of preference (limited set of features tested).

### Major methodological weaknesses

1. **No hypothesis testing:** The author does not state a priori hypotheses. The analysis is exploratory, increasing risk of false positives.

2. **Small sample sizes:** 20–30 participants per case study is insufficient for reliable regression modelling with multiple predictors (typically 10–20 participants per predictor variable are recommended).

3. **No cross-validation:** The regression models were not tested on hold-out samples or new participants.

4. **No blinding or randomisation:** Risk of experimenter bias and demand characteristics.

5. **Bespoke measures:** The semantic differential scales were not validated for reliability (internal consistency, test-retest) or construct validity.

6. **Confounding variables:** No control for aesthetic sensitivity, prior product experience, brand familiarity, or social desirability in value reporting.

### Primary findings (framework development)

The ACADE framework successfully produced a **step-by-step process chain** with five stages:

1. User value assessment (using Schwartz Value Survey)

2. Product personality profiling (using semantic differential)

3. Mapping values to personality dimensions (using regression)

4. Mapping personality dimensions to design parameters (using regression)

5. Optimisation (using mathematical solvers to maximise predicted preference)

The framework was implemented in a **software prototype** that allows engineers to input user value profiles and receive design parameter recommendations.

### Secondary findings (case study results)

**Case study 1 – Prosthetic limb covers (bionics):**

The value "universalism" (concern for social justice and nature) was positively correlated with preference for **matte, textured surfaces** (β = 0.42, p = 0.03, 95% CI not reported).

The value "stimulation" (need for excitement and novelty) was positively correlated with preference for **bright, saturated colours** (β = 0.38, p = 0.04).

The value "security" (need for safety and stability) was positively correlated with preference for **smooth, rounded shapes** (β = 0.45, p = 0.02).

The regression model explained 34% of variance in preference rankings (R² = 0.34).

**Case study 2 – Blood pressure monitor housing (medical engineering):**

The value "benevolence" (concern for close others) was positively correlated with preference for **soft, organic curves** (β = 0.51, p = 0.01).

The value "power" (social status and dominance) was positively correlated with preference for **angular, sharp-edged designs** (β = 0.44, p = 0.03).

The value "tradition" (respect for custom and convention) was positively correlated with preference for **neutral, muted colours** (β = 0.39, p = 0.04).

The regression model explained 29% of variance in preference rankings (R² = 0.29).

**Case study 3 – Bicycle frame geometry (mobility):**

The value "self-direction" (independence and creativity) was positively correlated with preference for **unconventional frame shapes** (e.g., asymmetrical tubing) (β = 0.47, p = 0.02).

The value "hedonism" (pleasure and enjoyment) was positively correlated with preference for **glossy, high-contrast finishes** (β = 0.41, p = 0.03).

The value "achievement" (personal success and competence) was positively correlated with preference for **sleek, aerodynamic profiles** (β = 0.43, p = 0.02).

The regression model explained 31% of variance in preference rankings (R² = 0.31).

### Important caveat

All p-values are marginal (0.01–0.04) and no correction for multiple comparisons was applied. With 10 value dimensions tested per case study, the expected number of false positives at p < 0.05 is 0.5 per study. The reported significant results (3–4 per study) exceed this, but not dramatically.

The author does not report standardised effect sizes (e.g., Cohen's d, partial η²). The regression coefficients (β) are standardised, meaning a one-standard-deviation increase in a value score was associated with a 0.38–0.51 standard-deviation shift in preference for a specific design feature. In plain English:

A person who scores one standard deviation above average on "universalism" is about **half a standard deviation more likely** to prefer a matte-textured prosthetic cover over a glossy one. This is a moderate effect – roughly equivalent to the difference in height between a 5'8" and a 5'11" person.

A person who scores one standard deviation above average on "power" is about **half a standard deviation more likely** to prefer an angular blood pressure monitor over a rounded one. Again, a moderate effect.

The regression models collectively explain about **30% of the variance** in preference rankings. This means 70% of why people prefer one design over another is *not* captured by their value scores – other factors (aesthetic sensitivity, prior experience, brand associations, mood at time of testing) likely dominate.

### Acknowledged by the author

The framework is a "small contribution" and requires further validation.

The case studies are illustrative, not definitive.

The semantic differential scales need further refinement and validation.

The software prototype is a proof-of-concept, not a production-ready tool.

### Critical reader observations

1. **No replication:** Each value–design parameter relationship was observed in only one small sample. Without replication, these could be chance findings.

2. **Cultural specificity:** All participants were German. Value–aesthetic relationships are known to vary across cultures (e.g., collectivist vs. individualist societies). These findings likely do not generalise to non-Western populations.

3. **Product specificity:** The relationships found for prosthetic covers may not hold for other bionic devices, let alone unrelated products.

4. **No behavioural measure:** Preference rankings are hypothetical. Actual purchase behaviour or long-term satisfaction was not measured. People may say they prefer one design but buy another.

5. **Stimulus limitations:** Participants rated static 2D images or 3D renders, not physical prototypes. Tactile and ergonomic qualities (weight, texture, grip) were absent, which likely influence real-world preferences.

6. **Demand characteristics:** Participants completed value questionnaires immediately before rating products, potentially priming them to align their preferences with their stated values.

7. **No negative control:** The author did not test whether value scores predicted preferences for *unrelated* design features (which would indicate spurious correlations).

8. **Publication bias:** As a dissertation, negative or null findings may be underreported. The author does not state how many value–design parameter relationships were tested versus how many were significant.

9. **No inter-rater reliability:** For the semantic differential ratings, the author does not report whether different participants agreed on which designs were "modern" vs. "traditional." If perceptions vary widely, the mapping becomes unreliable.

10. **No longitudinal data:** Preferences and values may shift over time. A design that appeals to a person's current values may not appeal to their future self.

For someone running their own n=1 experiment to understand their own product design preferences:

### What to test

**Specific intervention:** The ACADE framework itself is too complex for an n=1 experiment. Instead, test the core hypothesis: "My product design preferences are predicted by my personal value priorities." Choose one product category (e.g., water bottles, desk lamps, phone cases) and test whether designs that align with your top values are genuinely more appealing.

**Dose:** Test 5–8 design variations of the same product that differ on one or two key features (e.g., shape: rounded vs. angular; colour: muted vs. saturated; texture: matte vs. glossy).

### Minimum meaningful duration

**Single session (1–2 hours):** Complete a values assessment, then rate and rank designs. For a more robust test, repeat the ranking on three separate days to check stability of preferences.

**For behavioural validation:** If possible, purchase or 3D-print your top-ranked and bottom-ranked designs and use each for one week. Track daily satisfaction on a 1–10 scale.

### What to measure

**Primary metric:** Preference ranking (ordinal: 1st to 8th choice). Convert to a 0–100 "preference score" using a rank-based transformation (e.g., 1st = 100, 2nd = 86, 3rd = 71, etc.).

**Secondary metric:** Semantic differential ratings for each design on 5–10 adjective pairs (e.g., "modern–traditional," "friendly–aggressive," "cheap–premium"). This lets you see *why* you prefer a design.

**Predictor:** Your personal value priorities. Use the Schwartz Value Survey (freely available online) to identify your top 3 values. Alternatively, use a simpler tool: rank the 10 Schwartz values from most to least important to you.

**Confound check:** Rate your current mood (1–10), time of day, and whether you've seen similar products before (yes/no).

### Key confounds to control for

**Order effects:** Randomise the order in which you view designs. Use a random number generator to assign viewing order.

**Familiarity bias:** If you already own a product similar to one of the designs, you may prefer it due to familiarity. Note prior exposure.

**Brand associations:** If designs include logos or brand cues, these will override value-based preferences. Use unbranded, generic images.

**Aesthetic fatigue:** After rating 8 designs, your discrimination ability drops. Take a 5-minute break after every 3 ratings.

**Social desirability:** You may unconsciously prefer designs that signal "good" values (e.g., eco-friendly aesthetics). Be honest about what you actually like, not what you think you *should* like.

**Context effects:** A design that looks good on a screen may feel different in your hand. If possible, evaluate physical prototypes.

### What a positive result would look like

**Strong evidence:** Your top-ranked design consistently aligns with your top value. For example, if your top value is "self-direction," your 1st choice is the most unconventional design, and your last choice is the most conventional. This pattern holds across