A COMPARISON OF GOODNESS-OF-FIT TESTS FOR ANALYSIS OF NEST ORIENTATION IN WESTERN KINGBIRDS (TYRANNUS-VERTICALIS)
A COMPARISON OF GOODNESS-OF-FIT TESTS FOR ANALYSIS OF NEST ORIENTATION IN WESTERN KINGBIRDS (TYRANNUS-VERTICALIS)
复制标题
DOI：
10.2307/1368619
复制
发表时间：
1991-02-01
期刊：
影响因子：
3.1
通讯作者：
BERGIN, TM
中科院分区：
文献类型：
--
作者：
BERGIN, TM
采用计算机模拟和西部王鸟（Tyrannus verticalis）巢向数据比较了四次拟合优度检验的性质和假设。评估了各测试的结果和检测不同巢向模式的相对能力。计算机模拟表明，每种方法在检测单峰模式方面都很有效。然而，Rayleigh的Z和Watson的U2缺乏Rao的u所证明的检测双峰模式的鲁棒性。计算机模拟表明，在检测方向回避模式方面缺乏鲁棒性和低功耗，但也表明，每种方法在分析无模式数据时都具有低I型误差和高功耗。西部王鸟的巢向在年际间表现出高度的变异，但总体上表现为北、南、东三种模式的多模态。巢朝向的显著模式的检测取决于所使用的分析方法。卡方分析显示，四年中的每一年和所有年份的总和都明显偏离均匀分布。瑞利Z仅在1986年才显示出重要的巢向。Watson的U2和Rao的间距检验在所有年份加起来都显示出显著的巢向，1985年（仅Rao的间距检验）和1986年也是如此。Rao的间距检验在分析巢方向和其他圆形数据时比其他常用的拟合优度检验具有明显的优势，特别是在数据是多模态的情况下。Rao的间距测试解释了研究人员在分析巢取向数据时必须考虑的许多统计含义。它更灵活；能够以更少的限制假设处理更多类型的循环数据；能检测多态；而且在小样本量下更有效。此外，西王鸟和其他物种的巢向可能受到小气候、捕食和栖息地结构等多种因素的影响，因此在确定西王鸟巢向格局的性质时，利用Rao的间距测试是必不可少的。
Computer simulations and Western Kingbird (Tyrannus verticalis) nest orientation data were used to compare the properties and assumptions of four goodness-of-fit test. The results and relative abilities of each test to detect different patterns of nest orientation were evaluated. The computer simulations showed that each method was powerful in detecting unimodal patterns. However, Rayleigh's Z and Watson's U2 lacked the robustness for detecting bimodal patterns demonstrated by Rao's U. Computer simulations demonstrated a lack of robustness and low power in detecting directional avoidance patterns, but also demonstrated that each method performed with low Type I error and high power when analyzing data with no pattern.Nest orientation of Western Kingbirds exhibited a high degree of variation between years, but appeared to be generally polymodal with modes in the north, south and east. The detection of a significant pattern of nest orientation was dependent on the analytical method used. Chi-square analysis showed significant departures from a uniform distribution in each of four years and all years combined. Rayleigh's Z indicated significant nest orientation only in 1986. Watson's U2 and Rao's spacing test both indicated significant nest orientation for all years combined, and years 1985 (Rao's spacing test only) and 1986.Rao's spacing test has distinct advantages in the analysis of nest orientation and other circular data over other commonly used goodness-of-fit tests, especially if the data are polymodal. Rao's spacing test accounts for many of the statistical implications a researcher must consider when analyzing nest orientation data. It is more flexible; able to handle more types of circular data with fewer limiting assumptions; can detect polymodality; and is more powerful with small sample sizes. In addition, the nest orientation of Western Kingbirds and other species is probably affected by multiple factors such as microclimate, predation and habitat structure which make the use of Rao's spacing test indespensible in determining the nature of the pattern of nest orientation of Western Kingbirds.