# SMIL-566 item and Ukrainian translation verification

Date: 2026-07-11  
Scope: all 566 numbered positions in the adult male and female Sobchik SMIL forms, plus all 566 Ukrainian app entries in both effective sex variants.

## Result

The app has complete, sequential data: 566 Russian male positions, 566 Russian female positions, and 566 Ukrainian entries. The audit evaluated 1,132 effective Ukrainian form/item combinations.

| Ukrainian result | Form/item combinations |
|---|---:|
| Equivalent | 940 |
| Gender-specific equivalent | 138 |
| Service instruction | 54 |
| Needs wording review | 0 |

Independent QA adjudicated and resolved **18 form-specific findings across 17 IDs**: 13 material wording differences and 5 minor exactness differences. All now have status `equivalent` or `gender_specific_equivalent`; each retains its prior wording, recommendation, evidence, and `corrected_verified` resolution in the machine-readable artifact. The audit also removed two initial false positives: 245 female and 251 female remain unchanged because the chosen manual confirms their sex-specific wording.

Machine-readable evidence: [`smil-566-verification.json`](./smil-566-verification.json). It has one row per ID with exact SHA-256 hashes, male/female source status, effective Ukrainian status, correction audit trail, and repeat-pair membership. Rebuild it with [`generate-smil-566-verification.mjs`](./generate-smil-566-verification.mjs); its adjudication input is [`smil-566-ukrainian-qa.json`](./smil-566-ukrainian-qa.json).

## Source hierarchy and coverage

1. **Sobchik 2000 manual scan** — L. N. Sobchik, *Стандартизированный многофакторный метод исследования личности СМИЛ*, St. Petersburg: Rech, 2000. [Scan](https://kingmed.info/media/book/3/2892.pdf). This is the selected edition and highest-ranking item source.
   - OCR recovered 563/566 numbered male rows and 560/566 numbered female rows.
   - Three male and six female number markers were lost by OCR; their text and IDs remain covered by the archived Psylab lists.
   - Male comparison: 465 normalized exact, 14 contained despite page furniture, 40 OCR-equivalent, 27 service-equivalent, 17 manually adjudicated wording variants, 3 OCR marker gaps.
   - Female comparison: 456 normalized exact, 18 contained, 42 OCR-equivalent, 27 service-equivalent, 17 manually adjudicated wording variants, 6 OCR marker gaps.
2. **Psylab archived male and female forms**, both captured 2011-11-10: [male](https://web.archive.org/web/20111110215235/http://psylab.info/Стандартизированный_многофакторный_метод_исследования_личности/Текст_опросника_(мужской_вариант)), [female](https://web.archive.org/web/20111110212503/http://psylab.info/Стандартизированный_многофакторный_метод_исследования_личности/Текст_опросника_(женский_вариант)).
   - Female local corpus: 566/566 normalized exact.
   - Male local corpus: 564/566 normalized exact. ID 477 fixes an archived split word; ID 507 replaces a Latin `e` with Cyrillic `е`.
3. **Independent Clinli transcription** — [566-item male corpus](https://clinli.ru/blog/smil-test-mmpi-questions-answers-and-keys-to-questions).
   - 473/566 normalized exact.
   - 66 editorial/wording variants.
   - At all 27 Sobchik service positions, Clinli instead prints substantive legacy MMPI statements. Those statements cannot be combined with the selected Sobchik keys and norms.
4. **Sevastopol MSU teaching manual** — [2018 scan](https://sev.msu.ru/wp-content/uploads/2020/01/Uchebno-metodicheskoe-posobie.-Praktikum-po-psihodiagnostike.-Metodika-lichnostnoj-diagnostiki..pdf). It confirms that full SMIL has 566 positions and short SMIL has 398, but its appended questionnaire is the separate MMIL 377/384 form; it is not an item source for this audit.

Regular Tavily search/extraction was used. Tavily Research was not used.

## The 27 service positions and the `?` scale

The selected manual's male and female item lists each contain these 27 instructions:

`14, 33, 48, 63, 66, 69, 121, 123, 133, 151, 168, 182, 184, 197, 200, 205, 266, 275, 293, 334, 349, 350, 462, 464, 474, 542, 551`

The exact manual wording is “Номер данного пункта следует обвести кружочком.” Psylab paraphrases it as “answer ‘Не знаю’.” The app's forced `null` behavior is semantically consistent with circling the item number.

The manual is internally inconsistent: printed page 122 says that `26` removed positions form the initial `?` count, while the actual male and female lists contain 27 service instructions. Page 122 also says that `?` counts statements without a definite answer, `? > 70` invalidates the profile, 36–40 is permissible, and 41–60 indicates guarded responding.

Retain the app's actual 27-row baseline and raw `? > 70` rule. Do not substitute substantive items from another MMPI transcription: 14 service IDs occur in the F true key, and service IDs also occur in scales 1, 2, 4, 5, 6, 7, 8, 9, and 0. Answering them would change scores under keys and norms designed for the circled/service form.

## Ukrainian review method

Every ID received a form-specific status. The checks combined:

- prior bilingual item-by-item QA and the corrections already committed;
- exact ID/count/hash validation;
- source mismatch adjudication against the selected manual, Wayback forms, and Clinli;
- numeric-content preservation: zero numeric mismatches across 566 items;
- negation and polarity markers;
- frequency/intensity terms such as “always,” “never,” “often,” “rarely,” and “most of the time”;
- omitted clauses, added qualifiers, and stronger/weaker clinical wording;
- all 138 explicit female overrides plus source differences handled through neutral Ukrainian inheritance;
- 13 definite repeat/near-repeat pairs: 8/318, 13/290, 15/314, 16/315, 20/310, 21/308, 22/326, 23/288, 24/333, 32/328, 35/331, 37/302, and 38/311.

Four non-identical Ukrainian repeat pairs reproduce intentional source differences rather than translation drift: 23/288, 32/328 (male only), 35/331, and 37/302.

## Applied Ukrainian corrections

The independent audit retained the following 18 form corrections. All are applied and regression-tested. The audit-time text and source evidence remain in `smil-566-ukrainian-qa.json`.

| ID/form | Severity | Applied Ukrainian wording |
|---|---|---|
| 154 F | material | У мене ніколи не було нападів або судом (якщо «не було» — відповідь «Так»). |
| 190 F | material | У мене рідко болить голова. |
| 216 F | material | У моїй сімʼї стосунки менш теплі та дружні, ніж в інших сімʼях. |
| 231 F | material | Я люблю поговорити на сексуальні теми, тобто повʼязані з питаннями статі. |
| 245 M | material | Мої батьки та інші члени сімʼї часто чіпляються до мене. |
| 251 M | material | У мене бували такі стани, коли мої дії були перервані і я не розумів, що відбувається навколо мене. |
| 256 F | material | Єдина цікава сторінка в газеті — розділ гумору. |
| 290 M | minor | Робота дається мені ціною значного напруження (я працюю з великим напруженням). |
| 298 F | material | Коли кілька людей потрапляють у неприємну історію, їм краще домовитися, що говорити, а потім усім дотримуватися однієї лінії. |
| 299 M | material | Думаю, що я все відчуваю гостріше, ніж більшість людей. |
| 329 F | material | Мені рідко сняться сни. |
| 393 M | minor | Коней, які погано тягнуть, потрібно бити. |
| 393 F | minor | Коней, які погано тягнуть, потрібно бити. |
| 411 F | minor | Коли я чую про успіхи знайомих мені людей, я починаю відчувати себе невдахою. |
| 415 F | material | За сприятливого збігу обставин я могла б стати впливовою керівницею. |
| 526 F | material | Моє майбутнє здається мені безнадійним. |
| 556 F | minor | Я дуже уважно ставлюся до того, як я одягнена. |
| 560 F | material | Мене дуже часто дратує те, що я забуваю, куди кладу речі. |

The confirmed female source wording at 245 and 251 is intentionally preserved and covered by regression assertions.

## Manual wording variants already handled correctly

### Male manual variants (17)

| IDs | Outcome in Ukrainian |
|---|---|
| 97 | OCR merged item 98; item 97 itself is equivalent |
| 163 | Manual answer-polarity instruction preserved |
| 173 | Student-tense note retained from Psylab; compatible extension |
| 192, 210, 220, 482, 496, 563 | Equivalent syntax/word-order variants |
| 206 | Equivalent comparison; no polarity change |
| 245, 251, 299 | Corrected and verified above |
| 305 | Psylab duplicated “среди людей”; Ukrainian already matches the manual |
| 393 | Minor number alignment corrected above |
| 415 | “сприятливого збігу” follows manual “удачном стечении” |
| 459 | “принаймні одна” follows manual “по крайней мере одна” |

### Female manual variants (17)

| IDs | Outcome in Ukrainian |
|---|---|
| 70 | OCR merged item 71; item 70 itself is equivalent |
| 85 | Psylab typo `украсить`; Ukrainian correctly uses `вкрасти` per manual |
| 105, 210, 220, 241, 418, 482, 496, 548, 563 | Equivalent syntax/word-order variants |
| 173 | Student-tense note retained from Psylab; compatible extension |
| 297 | Past-tense parenthetical retained from Psylab; no polarity change |
| 245, 251 | Chosen manual confirms the preserved female-specific wording |
| 298, 329 | Corrected and verified above |
| 459 | Ukrainian already follows the manual |
| 507 | Manual OCR truncates at a page break; Wayback and male manual confirm current meaning |

## Limits

This audit verifies corpus identity, source alignment, and bilingual semantic fidelity. It does not establish Ukrainian norms, reliability, factor invariance, differential item functioning, criterion validity, or clinical cutoffs. Those require empirical data from Ukrainian samples. “Equivalent” in the JSON therefore means “no wording defect found against the selected source,” not “psychometrically validated.”