24小时热门版块排行榜

返回列表

brjun

新虫 (正式写手)

应助: 0 (幼儿园)
金币: 1605
散金: 591
红花: 6
帖子: 590
在线: 88.1小时
虫号: 2334928
注册: 2013-03-10
性别: GG
专业: 污染生态学

[求助] 大修，，，文章中统计的问题！！！跪谢！！！

今天收到大修的消息，其中有个审稿人提出下面一个意见。我真不懂啊！大家帮帮忙啊！！！
我的实验看生活垃圾黑炭施用对稻田重金属和温室气体的影响的。就两个处理对照和施用黑炭，每个处理3个重复！
下面是一个审稿人的意见！我这应该怎么回复啊！因为实验简单，结果也就那样，所以投的1.3的文章，大家帮帮忙！
I am concerned that the experimental design does not have the statistical power required to reliably detect some of the relatively small expected effects being investigated. For example, the authors observe a 5.1% increase in wheat grain yield, very similar to the 5.6% global average crop yield increase reported in the biochar-crop yield meta-analysis cited, yet the study report 'no change' (Conclusion 1) because the result did not achieve statistical significance. However, at n=3 replicates the power of the study to detect such an effect is only around 50% by my estimate, and thus there is a good chance that the yield improvement is real and the authors are making a Type II error. Likewise, Pb concentration in rice grain was observed to double, yet the change was judged as not statistically significant (again, 'no change' reported as Conclusion 3). However, given the high variance and low number of replicates, this experiment could only reliably detect (i.e. achieve power > 80%) an increase of a factor of -2.5x, and thus again the reported result could easily be a Type II error. While I am not a statistics expert, I would think that it is incorrect to report 'no effect' when there is in fact evidence of an effect but it cannot be confirmed due to low experimental power. Thus I would like to see either some additional statistical analysis indicating that the experimental design has sufficient power to support these conclusions, or an adjustment of the language in the Abstract and Conclusions sections to reflect the limited ability to detect such effects.

产量的数据：水稻：对照 8.7±1.2a，处理 9.0±0.2a，小麦：对照 5.9±1.0a，6.2±0.3a
稻米铅的数据：对照：0.04±0.02a，处理：0.08±0.04a
如果说3个重复不能说事的话，那我们实验室所有大田都是3个重复，那以前的文章不是全没用吗？

回复此楼