Open Decision Ledger · Criminal law
Distortion of poll results never conducted, and the voter's overall impression
Supreme Court, Jan. 15, 2026, 2025Do16181 (judgment)
Case name: 공직선거법위반
Official text on the National Law Information Center
Question
Does Article 96(1) of the Public Official Election Act, which forbids distorting and publishing or reporting the results of an opinion poll on an election, cover conduct that fabricates poll results without having conducted any poll at all, and how is it judged whether a publication distorts those results?
Holding
It does. Reading together the dictionary meaning of distortion, the Act's use of 'false fact' and 'distortion' as alternatives and its legislative purpose of protecting the fairness of elections, Article 96(1) reaches falsifying all or part of a poll actually conducted, manipulating a poll under way, and fabricating poll results without conducting any poll. Whether an expression is a distorted publication of poll results is decided on the overall impression it gives an ordinary voter, judged from its objective content, the ordinary meaning of the words and the way the phrases are connected, read against the expression's overall gist; the speaker's inner intent and private interests cannot govern its objective meaning and are not considered unless they appear on the surface.
Summary of the holding, verbatim
여론조사를 실시하지도 않은 채 허위로 여론조사결과를 만들어 내는 행위도 포함된다고 보는 것이 타당하다.
Unofficial English translation
it is reasonable to regard such conduct as including an act of fabricating public opinion poll results without even conducting a public opinion poll.
Disposition
Partly reversed and remanded Stated in the decision
원심판결 중 피고인 1에 대한 여론조사결과 왜곡 공표로 인한 공직선거법 위반 부분을 파기하고, 이 부분 사건을 부산고등법원에 환송한다. 검사의 나머지 상고를 기각한다.
Unofficial English translation
The portion of the judgment of the court below concerning Defendant 1's violation of the Public Official Election Act arising from the publication of distorted public opinion poll results is reversed, and this part of the case is remanded to the Busan High Court. The prosecutor's remaining final appeal is dismissed.
Form of the test
Circumstances weighed Stated in the decision
표현의 객관적 내용, 사용된 어휘의 통상적인 의미, 문구의 연결방법 등을 종합적으로 고려하여 그 표현이 선거인에게 주는 전체적인 인상을 기준으로 판단해야 하고
Unofficial English translation
by comprehensively considering the objective content of the expression, the ordinary meaning of the words used, the method of connecting the phrases, and so forth, it must be judged on the basis of the overall impression that the expression gives to voters, and
Factors
- 여론조사를 실시하지도 않은 채 허위로 여론조사결과를 만들어 내는 행위도
- 표현의 객관적 내용, 사용된 어휘의 통상적인 의미, 문구의 연결방법 등을 종합적으로 고려하여
- 그 표현이 선거인에게 주는 전체적인 인상을 기준으로 판단해야 하고
- 표현한 사람의 내심의 의도나 개인적 이해득실 등 주관적 사정에 따라 그 표현의 객관적 의미가 좌우된다고 할 수 없으므로
Unofficial English translation
• an act of fabricating public opinion poll results without even conducting a public opinion poll
• by comprehensively considering the objective content of the expression, the ordinary meaning of the words used, the method of connecting the phrases, and so forth,
• it must be judged on the basis of the overall impression that the expression gives to voters, and
• since the objective meaning of the expression cannot be regarded as being determined by the subjective circumstances of the person who expressed it, such as that person's internal intention or personal gain or loss,
Relation to precedent
Follows earlier decisions Stated in the decision
대법원 2018. 12. 13. 선고 2018도8338 판결
Unofficial English translation
Supreme Court Decision 2018Do8338 decided December 13, 2018
Decisions referred to
[1] 대법원 2018. 12. 13. 선고 2018도8338 판결 / [2] 대법원 2003. 2. 20. 선고 2001도6138 전원합의체 판결(공2003상, 876), 대법원 2009. 4. 9. 선고 2005다65494 판결(공2009상, 608), 대법원 2010. 2. 11. 선고 2009도8947 판결
Burden of proof
Not addressed Not in the decision
Open question and revisit condition
If a candidate had posted the poll's full report and explained the two questions and the cross-tabulation in the same message, would the overall impression still be that he came first in the electability poll?
The remand judgment of the Busan High Court on whether the leaflet distorted the poll results will show how the overall-impression test is applied to a text message leaflet. Any amendment to Article 96 or Article 252(2) of the Public Official Election Act, or a later decision changing the reading of distortion, would require this entry to be read again.
Cross-family AI independent reading & axis check
8 of 8 axes matched GPT-6 Astra (OpenAI, codex CLI) · 2026-10-24
A reader from another model family (GPT-6 Astra), given the official text but not the primary reading, agreed on all 8 axes.
Raw responses and comparison record: Protocol · Blind readings (JSONL) · Axis comparison (JSONL) · SHA-256 (JSON)
Lower-court contrast & scope of application
Lower-court reading vs. Supreme Court holding
Lower-court context: The Busan High Court, in its judgment of 17 September 2025 (2025No163), set aside the first-instance conviction and acquitted Defendant 1 on the charge of publishing distorted poll results. It reasoned that the graph's figures, 79.3%, 82.8% and 85.7%, matched the share of respondents who picked the same candidate in both the hypothetical match-up and the electability question, that their sum far exceeded the 100% base, so the graph plainly was not the electability result, and that the phrase claiming first place, read with that graph and the caption at the foot of the leaflet, was not by itself information likely to make voters believe Defendant 1 had ranked first.
Supreme Court standard: The Supreme Court held that this misapprehended the meaning of publication of distorted poll results under Article 96(1). An ordinary voter who did not know the questionnaire could not read the percentages as the cross-tabulated shares, and would instead take the largest type at the top of the card news as meaning Defendant 1 ranked first in the electability poll; the small caption at the foot, much of the leaflet being sent by text message, could not be checked. It reversed that part for misapprehending the legal principle and remanded it to the same court, while dismissing the rest of the prosecution's appeal, so the acquittal on the false-statement charge stood.
Distinguishing conditions
- A leaflet whose overall gist, read as an ordinary voter would receive it, does not convey that the candidate came first in the electability poll.
- A publication of poll results that are genuinely the electability figures on a 100% base, so that nothing has been falsified or manipulated.
- A case in which the distortion rests only on the speaker's inner intent, that intent not appearing on the face of the expression.
Record verification notes
Fix the medium first, since much of the material here went out as text messages, and then read the leaflet as an ordinary voter would, noting where the headline sat, its size and what the caption at the foot actually said. Set the percentages against the poll's two questions and its analysis, and keep the false-statement charge apart, because the Supreme Court dismissed the remaining appeal and left the acquittal on that charge undisturbed.
From the official text
Points at issue
[1] 공직선거법 제96조 제1항에서 금지하는 여론조사결과를 왜곡하는 행위에 여론조사를 실시하지도 않은 채 허위로 여론조사결과를 만들어 내는 행위도 포함되는지 여부(적극) [2] 어떤 표현이 공직선거법 제250조 제2항에서 정한 허위사실공표에 해당하는지 판단하는 방법 및 공직선거법 제96조 제1항에서 정한 선거에 관한 여론조사결과를 왜곡하여 공표 또는 보도한 경우에 해당하는지를 판단할 때에도 위 법리가 적용되는지 여부(적극)
Unofficial English translation
[1] Whether an act of fabricating public opinion poll results without even conducting a public opinion poll is also included in the act of distorting public opinion poll results prohibited by Article 96, paragraph (1) of the Public Official Election Act (affirmative) [2] The method for determining whether a given expression constitutes the public announcement of false facts under Article 250, paragraph (2) of the Public Official Election Act, and whether that legal principle also applies when determining whether one falls within the case of publishing or reporting public opinion poll results relating to an election in a distorted manner as provided in Article 96, paragraph (1) of the Public Official Election Act (affirmative)
Summary of the holding
[1] 공직선거법 제96조 제1항은 "누구든지 선거에 관한 여론조사결과를 왜곡하여 공표 또는 보도할 수 없다."라고 규정하고, 제252조 제2항은 "제96조 제1항을 위반한 자는 5년 이하의 징역 또는 300만 원 이상 2천만 원 이하의 벌금에 처한다."라고 규정하고 있다. 이는 여론조사의 객관성·공정성에 대한 신뢰를 이용하여 선거인의 판단에 잘못된 영향을 미치는 행위를 처벌함으로써 선거의 공정성을 보장하려는 규정이다. ‘왜곡’의 사전적 의미는 ‘사실과 다르게 해석하거나 그릇되게 함’이고, ‘그릇되다’의 사전적 의미는 ‘어떤 일이 사리에 맞지 아니하다.’이다. 사실에 대한 왜곡은 일부 사실을 숨기거나 허위의 사실을 덧붙이거나 과장, 윤색하거나 조작하여 전체적으로 진실이라 할 수 없는 사실을 표현하는 방법으로 이루어진다. 공직선거법은 ‘허위의 사실’과 ‘사실의 왜곡’을 선택적인 것으로 규정하기도 하고(제96조 제2항), 허위를 배제하지 않는 의미로 ‘왜곡’을 사용하기도 한다(제8조의6 제4항). 이와 같은 왜곡의 의미와 용법에 앞에서 본 공직선거법 제96조 제1항, 제252조 제2항의 입법 목적을 종합하면, 여론조사결과를 왜곡하는 행위에는 실제로 여론조사를 실시한 후 그 결과 중 전부 또는 일부를 허위로 만들어 내거나 실시 중인 여론조사에 인위적인 조작을 가하여 그릇된 여론조사결과를 도출하는 경우뿐만 아니라 여론조사를 실시하지도 않은 채 허위로 여론조사결과를 만들어 내는 행위도 포함된다고 보는 것이 타당하다. [2] 어떤 표현이 공직선거법 제250조 제2항에서 정한 허위사실공표에 해당하는지 판단하기 위해서는 먼저 그 표현의 의미를 확정해야 한다. 이는 일반 선거인이 그 표현을 접하는 통상의 방법을 전제로 그 표현의 전체적인 취지와의 연관하에서 표현의 객관적 내용, 사용된 어휘의 통상적인 의미, 문구의 연결방법 등을 종합적으로 고려하여 그 표현이 선거인에게 주는 전체적인 인상을 기준으로 판단해야 하고, 표현한 사람의 내심의 의도나 개인적 이해득실 등 주관적 사정에 따라 그 표현의 객관적 의미가 좌우된다고 할 수 없으므로, 이러한 주관적 사정은 겉으로 드러나지 않은 이상 표현의 해석에 고려할 것은 아니다. 한편 이러한 법리는 공직선거법 제96조 제1항에서 정한 선거에 관한 여론조사결과를 왜곡하여 공표 또는 보도한 경우에 해당하는지를 판단할 때에도 적용된다.
Order
원심판결 중 피고인 1에 대한 여론조사결과 왜곡 공표로 인한 공직선거법 위반 부분을 파기하고, 이 부분 사건을 부산고등법원에 환송한다. 검사의 나머지 상고를 기각한다.
Unofficial English translation
The portion of the judgment of the court below concerning Defendant 1's violation of the Public Official Election Act arising from the publication of distorted public opinion poll results is reversed, and this part of the case is remanded to the Busan High Court. The prosecutor's remaining final appeal is dismissed.
Provisions referred to
[1] 공직선거법 제8조의6 제4항, 제96조 제1항, 제2항, 제252조 제2항 / [2] 공직선거법 제96조 제1항, 제250조 제2항
Read 2026-10-24 · Cross-read 2026-10-24. All quotes matched against the official text. The Chinese-character placeholders the court used for anonymised parties are given in their Korean readings, and parenthetical Chinese-character glosses are omitted.
Every entry is published only after all quotes have been matched character by character against the official text, and audited by an independent second reading from a different model family across all eight axes. Both agreed axes and text-grounded interpretive divergences are recorded openly. The ledger is a research record, not an assessment of the court or advice on any matter.
Back to the ledger