正確使用統計圖表呈現處理間比較

Correct Usage of Graph/Table for Treatment Comparisons

呂秀英(Hsiu-Ying Lu)

《台灣農業研究》 60卷1期 (2011/03) Pp. 61-71

https://doi.org/10.6156/JTAR/2011.06001.06

摘要

精確表達資訊是統計圖表製作的首要原則。農業試驗研究最常見的資料分析，為利用變方分析(analysis of variance, ANOVA)及多重比較(multiple comparisons)來探討多組處理間的差異性。多重比較包含很多種方法，其中Fisher的最小顯著差異性測驗(Fisher's protected least significant difference test, LSD test)是農業學術期刊中最普遍接受的多重比較方法，主要用來進行所有處理平均值間的兩兩比較是否存有差異。這些統計分析的結果，若沒有在圖或表中予以適當呈現，會令人不易閱讀甚至產生誤解，進而影響研究報告發表的品質。因此，本文以實例說明如何用適當的圖表來有效呈現出ANOVA及LSD之分析結果，並給予精確的文字解讀，期能有助於農業科技研究成果發表之參考。

關鍵字

變方分析 ； 多重比較 ； 成對處理平均值間比較 ； 最小顯著差異性測驗

並列摘要

Conveying accurate information is key to the effective presentation of scientific graphs or tables. The analysis of variance (ANOVA) and multiple comparisons are the methods commonly used for comparison of several treatment means in data collection from agricultural research. There are many methods available for multiple comparisons of treatment means; among them, the Fisher's protected least significant difference (LSD) test is the favorable method for all pairwise comparisons of treatment means. A graph or table is of no scientific merit if the data are not analyzed statistically and presented accurately. The objective of this report is to describe proper procedure for preparation and presentation of statistically analyzed data in the form of scientific graphs or tables for improving quality of scientific publications.

並列關鍵字

Analysis of variance ； Multiple comparisons ； Pairwise comparisons of treatment means ； Least significant difference test