|
[求助]
Major Revision:有关参数检验和非参数检验求助!
Manuscript中有关统计部分描述:
2.9. Statistical analysis
All results were expressed as the mean ± standard deviation of at least triplicate analyses. Data were subjected to analysis of variance using the linear model procedure of Statistix 9.0 (Analytical Software, Tallahassee, FL, USA). Significant differences (P < 0.05) between individual means were identified by the least significant difference procedure of all pairwise multiple comparisons.
Editor's Corrections
§2.9: Given the number of repetitions, you can not do a parametric test to compare differences between samples (for that you need at least 20 samples!). In fact, you may NOT carry out an ANOVA to compare data at different oil ratios for the same modality as these oil ratios are continuous variable and not independant occurences. You must instead model the evolution of the responses as a function of oil ratio. You may use a parametric test to compare OEH and HOE however, provided you do it over the WHOLE concentration range. Otherwise (comparison at one given oil ratio) you need to use non-parametric tests (such as Kruskal Wallis test or Dunn test) because you have too few samples (triplicates?).
我一共做了OEH和HOE两种处理,每种处理下都有四个油浓度梯度(2%、4%、6%、8%),每个浓度梯度做了三个平行样,另外进行两次重复。空白(对照)为不添加油。
按照编辑的意思是不能用方差分析(ANOVA)?于是我按照她的建议,用非参数检验(Kruskal Wallis test结合Dunn test)试着分析了数据,发现结果又和ANOVA有天壤之别,大部分的数据之间变得没有显著性差别了...
眼看文章修回的日期就要到了,实在不知如何处理,求助各位大神,谢谢~ |
|