Purpose and Status
This report documents IQ Spark as it actually works in the browser. It is based on the live 36-item bank and scoring code, not on a general description of intelligence tests. Publishing these details lets visitors inspect what contributes to a result and decide whether the exercise is appropriate for their purpose.
IQ Spark is an independently developed pattern-reasoning exercise. It is not a standardized IQ test, a licensed form of Raven's Progressive Matrices, or a substitute for an assessment administered by a qualified professional. No representative norming sample, test-retest study, reliability coefficient, or clinical validation study has been completed for this item bank. The displayed number is therefore an educational IQ-style reference score, not a measured clinical IQ.
The intended use is narrow: practice visual-rule detection, review which item families caused difficulty, and compare attempts made under similar conditions. The score must not be used for diagnosis, school placement, gifted identification, employment, disability decisions, or any high-stakes conclusion about a person.
Current Item-Bank Snapshot
The inventory below was generated from the current source item bank. Each set contains 12 items. Answer positions are deliberately balanced so A, B, C, and D are each correct nine times. Items are shuffled only within their set, preserving the easier-to-harder progression while reducing a fixed presentation order.
| Set | Items | Item weight | Design emphasis |
|---|---|---|---|
| A | 12 | 3 | Single-rule completion, alternation, count, direction, and introductory matrices. |
| B | 12 | 4 | Analogies, overlays, rotations, spatial position, and combined transformations. |
| C | 12 | 5 | Multi-rule matrices and denser combinations of count, position, shading, and rotation. |
The category inventory is: matrix 9 items; rotation 5; spatial 5; pattern 4; transformation 3; sequence 3; logic 3; overlay 2; analogy 1; and mirror 1. These counts matter because the category chart is descriptive rather than a balanced subscale profile. For example, a result based on one mirror item cannot support a broad claim about a person's mirror-reasoning ability.
Exact Weighting Formula
Let A, B, and C be the number of correct answers in the three sets. The maximum weighted total is 144 points: 12 x 3 + 12 x 4 + 12 x 5. The browser calculates the result in two steps:
normalized_result = round((weighted_total / 144) x 36)
Difficulty weights are design labels assigned by the developer; they are not item-response theory parameters estimated from a norming sample. A correct Set C item contributes five points and a correct Set A item contributes three. The weighted total is then returned to a 0-36 range so it can use the site's reference map.
This design means two visitors with the same number of correct answers may receive different normalized results if one solved more later-set items. That is intentional, but it should not be confused with a validated estimate of item difficulty. Publishing the formula makes that trade-off visible instead of hiding it behind a proprietary-sounding label.
Reference Mapping
The normalized result is mapped to broad IQ-style reference anchors. Values between anchors are linearly interpolated. The map was created for readable educational feedback; it was not derived from observed population performance on IQ Spark.
| Normalized result | Reference anchor | Interpretation on this site |
|---|---|---|
| 0-14 | 70 toward 90 | Many item rules were not solved in this session. |
| 15-20 | 90 toward 100 | Core rules were partly solved; combined rules remained difficult. |
| 21-26 | 100 toward 110 | Middle reference range for this item bank. |
| 27-30 | 110 toward 120 | Most medium items and some hard items were solved. |
| 31-33 | 120 toward 130 | Strong performance across most item families. |
| 34-35 | 130 toward 140 | Nearly all weighted item rules were solved. |
| 36 | 140 | Maximum normalized result on the current form. |
The results screen also converts the reference number to a percentile-style value using a normal curve centered at 100 with a standard deviation of 15. That value is a mathematical illustration of the reference scale, not an observed percentile among IQ Spark visitors. IQ Spark does not transmit individual answer records to a server and therefore does not maintain a visitor norm table.
Checks Performed
The current release uses deterministic checks that can be repeated directly against the item bank and scoring code:
- Exactly 36 items are present, split evenly across Sets A, B, and C.
- Every item has four answer options and one answer-key position.
- Correct answer positions are balanced at nine A, nine B, nine C, and nine D.
- The maximum weighted total is 144 and a perfect response maps to normalized result 36.
- Age range and completion time do not change the score.
- Progress and the latest result are stored locally in the browser, not in an IQ Spark account.
These checks prevent implementation mistakes, but they are not a validation study. The project still lacks independent expert item review, accessibility testing with a representative user group, response-data calibration, internal-consistency estimates, test-retest evidence, and comparisons with a validated external measure.
Fairness, Accessibility, and Known Bias
Using symbols instead of vocabulary reduces one source of language demand, but it does not make a test culture-free. Familiarity with puzzle conventions, device size, visual acuity, motor control, education, practice, fatigue, and available time can all affect performance. The item bank has not been calibrated separately across languages, countries, ages, disability groups, or device classes.
The project follows the principle that score interpretations need evidence for their intended use. The open-access Standards for Educational and Psychological Testing describes validity as a property of proposed score interpretations and uses. The International Test Commission technology-based assessment guidance also emphasizes delivery conditions, fairness, accessibility, security, and privacy. IQ Spark cites these standards to explain its limits, not to claim compliance or professional certification.
Relationship to Raven's Matrices
IQ Spark uses the broad idea of finding a missing element in a visual relationship. It does not copy Raven item artwork, use Raven norms, or administer a Raven form. Pearson's official Raven's 2 product documentation describes a professional instrument with controlled forms, manuals, updated norms, standard scores, and percentile reporting. Those features are precisely why a licensed assessment supports interpretations that this independent browser exercise cannot.
Privacy and Reproducibility
The scoring calculation runs in the visitor's browser. IQ Spark uses localStorage for unfinished progress and a short-lived result cache. The application sends aggregate interaction events to Google Analytics, such as test started or completed, but it does not send the visitor's full answer record or displayed reference score as a user profile.
Because the scoring rules are documented above, a visitor can reproduce the weighted calculation with the set totals alone. The report intentionally separates code-level reproducibility from psychometric validity: a calculation can be implemented consistently and still lack evidence for clinical interpretation.
Version History and Ownership
| Version | Date | Change |
|---|---|---|
| 1.0 | July 27, 2026 | Published exact item counts, category distribution, weighting formula, reference map, privacy behavior, and validation gaps. |
Author and project owner: Emircan Agac, founder and developer of IQ Spark. No licensed psychologist has reviewed or endorsed the current item bank. Corrections can be sent to iqsparktest@gmail.com; include the page section, disputed statement, and a source that supports the correction.
Related first-party documentation: scoring guide, worked examples, methodology, and editorial policy.