Potential Divers — Card Sort Analysis

00 What this study set out to learn

Prospective divers — people who have not yet tried scuba — were asked to sort 39 PADI content cards into groups of their own making and name each group. The question behind it: how do future divers expect this content to be organised?

The study was run twice — once in English, once in Spanish, with different audiences — and the same seven groups emerged in both. Because every structural measure here reads only which cards were grouped together (never the translated labels), the two studies compare cleanly and act as independent confirmation of each other.

01 At a glance

How the two samples and their overall structure line up. The two studies were run independently, in different languages, with different audiences.

02 Who sorted the cards

Combined profile of the 50 valid participants across both studies — the people whose sorts feed every number in this report.

Profiles per study

The valid English (31) and Spanish (19) samples, side by side. Blue = English, orange = Spanish.

Source: recruitment screener & panel demographics embedded in each response file (column "Demographics").

03 Category mapping

An open card sort lets people name their own groups, so the raw labels vary wildly. Every label was standardised into a consistent set of themes — 13 in English, 15 in Spanish. This is the one step that depends on wording, so it is published in full: each theme lists the exact labels participants typed and how many used each. Switch between studies.

04 How the numbers are built

An open card sort: each participant sorted the same 39 cards into as many groups as they liked and named each group. Everything below is computed from those raw groupings.

See it both ways

The two views the pipeline produces — the co-occurrence matrix and the clustering tree — shown for each study. Colours mark the seven clusters used throughout this report.

05 The seven emergent clusters

The same seven mental models surfaced in both languages. Cohesion bars show how tightly each cluster's cards were grouped — English and Spanish computed independently on the shared card membership.

Cluster by cluster

For each cluster: the cards it contains with their dominant-placement % and strongest grouping partner, beside the EN/ES cohesion.

06 The pattern observed — run twice, same result

The study ran independently in two languages and produced the same seven groups with comparable cohesion — strong independent confirmation of the structure.

07 The strongest card pairs

The card pairs the most people grouped together — the load-bearing relationships to keep intact in any navigation. Shown for both studies.

The full top-10 strongest pairs

Percentages = share of valid participants who placed both cards in the same group. The two single strongest bonds — Our Difference + Our Mission & History (90% EN) and Learn to Scuba Dive + Scuba Basics Guide (87% EN) — anchor the About & Brand and Learn clusters respectively.

08 Ambiguous cards

The mirror image of the strongest pairs. These are the cards with the lowest average top-theme agreement across the English and Spanish studies — content participants scattered across many groups with no clear consensus on where it belongs. Each row reads how to interpret that ambiguity.

"# categories" = how many distinct groups the card landed in (the wider the scatter, the more contested). "Best placement" = its single most-common theme and the share of participants who filed it there. "Strongest pair" = the highest co-occurrence the card has with any one other card — often the only firm relationship an otherwise-scattered card has. Ranked by the average of the English and Spanish best-placement scores, lowest first.

09 Data quality & exclusions

Before analysis, sorts that carried no usable signal were removed — incomplete submissions and non-thematic / nonsense labelling — leaving 4 excluded in English and 6 in Spanish.

The higher Spanish exclusion rate (24% vs 11%) is driven almost entirely by no sort submitted — a recruitment/completion artefact, not a sorting-quality problem. After cleaning, the two valid samples are well-matched in effort: a mean of 5.2 (EN) and 5.3 (ES) groups per participant.

10 What the data tells us & next steps

The signal is strong, bilingual, and consistent. Four things to carry forward.

11 The raw sorts

Every valid participant's groups, exactly as they created and labelled them. Switch between the two studies — English groups are shown in English, Spanish groups in Spanish, as submitted.