9.8
가설 검정의 결과는 귀무가설을 기각하거나 기각하지 않는 것으로 나타납니다. 이 결정은 자료 분석, 적절한 검사 통계, 적절한 신뢰 수준, 임계 값 및 P-값을 기반으로 내려집니다. 하지만 귀무가설을 기각할 수 없다는 증거가 나오면 귀무가설을 '채택'한다고 말하는 것이…
실험에서, 감염된 식물이 있는 농장에 널리 적용 가능한 살충제를 투여합니다.
이 살충제는 적용 후 건강한 식물의 수를 증가시킬 것으로 예상됩니다. 그러나 실험이 끝날 무렵 건강한 식물과 감염된 식물의 비율은 동일하게 유지되었습니다.
여기서 살충제는 효과가 없다는 귀무 가설이 성립하는 것처럼 보이지만, 이 가설을 받아들여야 합니까 아니면 거부하지 말아야 합니까?
이 가설을 받아들이는 것은 살충제가 효과가 없고 식물의 건강을 향상시킬 수 없다는 것을 의미합니다.
이 결정은 실제로 관찰된 결과에 대한 다른 그럴듯한 설명을 간과합니다.
이 경우 처방되지 않은 양이나 농도의 살충제를 사용해도 효과가 없을 수 있습니다.
살충제가 표적으로 삼을 수 없는 물질에 식물이 감염될 가능성이 있습니다.
귀무 가설을 기각하지 않으면 기대 효과 또는 관찰된 효과에 대한 충분한 증거가 없음을 의미합니다.
만약 과학자들이 귀무가설을 받아들였다면, 식물 바이러스의 발견이나 많은 멸종된 종의 재발견은 불가능했을 것이다.
View the full transcript and gain access to JoVE Core videos
Q1: Why is 'fail to reject' better than 'accept' when describing null hypothesis test results?
Accepting a null hypothesis implies it is proven true, but hypothesis testing only shows insufficient evidence against it. Failing to reject means the data lacks support for the alternative hypothesis, not that the null is definitively true. This distinction matters because accepting prematurely can halt further investigation and overlook alternative explanations for observed results.
Q2: What are the consequences of accepting rather than failing to reject a null hypothesis?
Accepting a null hypothesis may lead to severe consequences in critical fields like criminal trials, drug testing, and species research. It implies the hypothesis is proven and needs no further study, potentially preventing discovery of viruses, extinct species, or other important findings. Failing to reject leaves room for future evidence to challenge existing conclusions.
Q3: How can alternative explanations affect the interpretation of null hypothesis results?
When a null hypothesis cannot be rejected, multiple plausible explanations may exist beyond the hypothesis being true. In the insecticide example, ineffective results could stem from incorrect dosage, insecticide limitations against specific pathogens, or other factors. Accepting the null overlooks these alternatives, while failing to reject acknowledges the need for further investigation.
Q4: What statistical factors determine whether to reject or fail to reject a null hypothesis?
The decision depends on data analysis, test statistic values, confidence level, critical values and significance level, and P-values. These elements work together to determine if sufficient evidence exists to reject the null hypothesis. When evidence is insufficient, the appropriate conclusion is to fail to reject rather than accept.
Q5: Why can't a null hypothesis be proven true through hypothesis testing?
Hypothesis testing begins by assuming the null hypothesis is true, then evaluates whether data contradicts it. The test can only show insufficient evidence against the null, not prove it true. Absence of evidence against a hypothesis differs fundamentally from evidence proving it true, which is why failing to reject is the correct statistical conclusion.
Q6: How does the insecticide experiment illustrate the difference between accepting and failing to reject?
When plant health remained unchanged after insecticide application, accepting the null would conclude the insecticide is ineffective. However, failing to reject acknowledges other possibilities: incorrect dosage, pathogen resistance, or unsuitable insecticide type. This distinction preserved opportunities for further research that might reveal these alternative explanations.
Q7: What role does newer scientific evidence play in hypothesis testing conclusions?
Newer scientific evidence often challenges existing studies and conclusions. Accepting a hypothesis suggests it is proven and requires no further study, potentially blocking important discoveries. Failing to reject keeps investigations open, allowing future evidence to refine understanding and potentially overturn previous conclusions in fields like virology, paleontology, and medicine.