提高回归模型拟合优度的策略(Ⅱ)——算术均值变换与其他变量变换

本文目的是介绍第二种提高回归模型拟合优度的策略,即算术均值变换与其他变量变换。具体方法包括以下几个方面:①对多值名义自变量采取"算术均值变换";②对定量自变量引入派生变量,包括"对数变换""平方根变换""指数变换""平方变换""立方变换"和"交叉乘积变换"的结果;③对定量因变量分别采取"对数变换""平方根变换""指数变换""倒数变换"和"Logistic变换";④构建回归模型时,在假定"包含截距项"与"不含截距项"的条件下,分别采取"前进法""后退法"和"逐步法"筛选自变量。得到了如下结论:①对定量因变量和自变量不做变量变换时,回归模型的拟合优度非常差;②根据资料所具备的条件,对定量因变量采取不同的变量变换方法,其回归模型的拟合优度是不同的;③对多值名义自变量进行"算术均值变换"是合理的,且有助于提高回归模型拟合优度;④对定量自变量引入派生变量是非常有价值的;⑤假定回归模型中不含截距项有助于提高回归模型的拟合优度。     

提高回归模型拟合优度的策略(Ⅲ)——校正均值变换与其他变量变换

本文目的是介绍第三种提高回归模型拟合优度的策略,即校正均值变换与其他变量变换。具体方法包括以下几个方面:①对多值名义自变量采取"校正均值变换";②对定量自变量引入派生变量,包括"对数变换""平方根变换""指数变换""平方变换""立方变换"和"交叉乘积变换"的结果;③对定量因变量分别采取"对数变换""平方根变换""指数变换""倒数变换"和"Logistic变换";④构建回归模型时,在假定"包含截距项"与"不含截距项"的条件下,分别采取"前进法""后退法"和"逐步法"筛选自变量。得到了如下结论:①对定量因变量和自变量不做变量变换时,回归模型的拟合优度非常低;②根据资料所具备的条件,对定量因变量采取不同的变量变换方法,其回归模型的拟合优度是不同的;③对多值名义自变量进行"校正均值变换"是合理的,且有助于提高回归模型拟合优度;④对定量自变量引入派生变量是非常有价值的;⑤假定回归模型中不含截距项有助于提高回归模型的拟合优度。     

提高回归模型拟合优度的策略(Ⅳ)——优化计分变换与其他变量变换

本文目的是介绍第四种提高回归模型拟合优度的策略,即优化计分变换与其他变量变换。具体方法包括以下几个方面:①第一,对多值名义自变量采取"优化计分变换";②对有序自变量分别采取"单调变换"与"优化计分变换";③对定量自变量分别采取"样条变换"和"单调样条变换";④对定量因变量分别采取"样条变换""单调样条变换"和"BOX-COX变换"。全部变量变换方法组合起来共12种,共创建了12个多重非线性回归模型。依据"拟合优度评价指标"的取值,从12个回归模型中挑选出一个,即本文中的"模型1",其"均方误差平方根=0.30935、R~2=0.9586、调整R~2=0.9527"。结合本期科研方法专题同类文章的结果和结论,得出提高回归模型拟合优度的策略主要在于以下四点:①应对"定量因变量""定量自变量"和"多值有序自变量"采取合适的变量变换方法;②在拟合回归模型的过程中,应尽可能多地引入派生变量;③应假定回归模型中不含截距项;④在构建回归模型的过程中,应尽可能多地使用筛选自变量的策略,如"前进法""后退法"和"逐步法"。     

非配对设计二值资料一水平多重Logistic回归分析

本文的目的是介绍非配对设计二值资料一水平多重Logistic回归模型的构建与求解方法。基于SAS软件分别对以列联表和数据库形式呈现的定性资料进行全面分析,并得出了4个对提高模型拟合优度很有价值的结论:第一,若资料以列联表形式呈现,应拟合"加权"Logistic回归模型;第二,若资料中包含定量自变量,不适合将其定性化;第三,若资料中包含定量自变量,应依据定量自变量和二值自变量产生出派生自变量;第四,若资料中有定性自变量时,必须将多值名义或有序自变量进行哑变量变换,不需要依据二值自变量产生出派生自变量。     

适应性回归分析(Ⅳ)——与非适应性回归分析的比较

本文目的是分析一个已知真实情况的资料,比较适应性回归分析与非适应性回归分析建模的效果。结论如下:当资料中存在与因变量确有关系的自变量时,ADAPTIVEREG过程具有较好的甄别能力;REG过程具有较好的甄别能力,但需要满足一定条件,即采用"前进法"或"逐步法"筛选自变量,同时还需要"假定模型包含截距项"。当资料中不存在与因变量确有关系的自变量时,ADAPTIVEREG过程几乎完全失去了甄别能力;REG过程具有较好的甄别能力,但需要满足一定条件,即采用"前进法"筛选自变量,同时还需要"假定模型包含截距项"。若研究者基于"基本常识"和"专业知识"确定的自变量都与因变量有关系,对因变量进行Logistic变换,并且,假定回归模型中不含截距项时,会在回归模型中保留非常多的自变量。     

回归建模的基础与要领(Ⅲ)——变量状态与相互间关系

本文目的是介绍回归建模的基础与要领之三,即"变量状态与相互间关系"。首先,介绍"因变量状态"与"自变量状态";其次,介绍"自变量间相互关系",即"自变量间相互独立""自变量间有线性关系"和"自变量间有非线性关系";最后,介绍"自变量与因变量间关系",包括"自变量与因变量间无任何数量关系""自变量与因变量间有间接数量关系"和"自变量与因变量间有直接数量关系"。很明显,清楚"变量状态和变量间关系"是构建合理回归模型的重要基础与要领之一。     

M∶N配对设计二值资料一水平多重Logistic回归分析

本文的目的是介绍m∶n配对设计二值资料一水平多重Logistic回归分析方法。首先,介绍了需要了解的基本概念;其次,介绍了构建此类回归模型的基本原理;最后,通过一个实例介绍了使用SAS实现计算的全过程。在此过程中,获得了如下四点启示:其一,有必要确保所获得的科研资料是值得分析的;其二,有必要基于定量自变量产生派生变量;其三,有必要同时采用"逐步法""前进法"和"后退法"筛选自变量;其四,有必要采用多种方法评价不同回归模型对资料的拟合优度。     

适应性回归分析(Ⅱ)——排除噪声变量的干扰

本文目的是通过分析一个带有8个噪声变量的数据集,揭示适应性回归模型的实际应用价值。在数据集包含两个自变量与因变量有密切数量联系的前提条件下,适应性回归模型受噪声变量的影响接近于零;在数据集包含一个自变量与因变量有密切数量联系的前提条件下,适应性回归模型受噪声变量的影响较大,其分析结果出现了一定程度的"失真";在数据集包含零个自变量与因变量有密切数量联系的前提条件下,适应性回归模型受噪声变量的影响非常大,其分析结果是完全不可信的。得出的结论是:适应性回归分析模型不是万能的,其结果的可信度取决于数据集中是否真正包含"客观存在的规律性"。     

变量变换回归分析(Ⅰ)——拟合含间断点资料的方法

本文目的是介绍一种能很好地拟合具有间断点资料的方法。当资料中具有明确的间断点或整个资料包含多段不同变化趋势的曲线类型时,为了提高曲线回归模型对资料的拟合优度,需要充分发挥"节点"的作用。可基于两种不同视角来利用"节点":其一,人为设定不同数目的节点,利用样条变换方法拟合分段多项式曲线;其二,在客观存在的节点上,求曲线的一阶乃至四阶导数,并据此构建曲线回归模型。得到的结论是:后者的拟合效果优于前者。     

零膨胀Poisson分布模型回归分析

本文目的是介绍零膨胀Poisson分布模型回归分析。首先,介绍零膨胀计数资料及其零膨胀Poisson分布回归模型构建原理,包括"零膨胀Poisson分布回归模型的形式"和"零膨胀Poisson分布回归模型的求解";其次,介绍"零膨胀Poisson分布回归模型的SAS实现",包括"创建SAS数据集""呈现因变量Y的频数分布""求出因变量Y的均值和方差"和"基于全部自变量对因变量Y构建多重零膨胀Poisson分布回归模型"。本文结果提示,当计数资料为非严重过离散的零膨胀计数资料时,拟合"多重零膨胀Poisson分布回归模型",可获得满意的拟合效果。     

协变量在因果中介效应分析中的作用

本文目的是介绍因果中介效应分析的理论基础以及结合一个实例采用SAS实现因果中介效应分析的具体方法。因果中介效应分析的理论基础包括基本概念以及定义因果中介效应的反事实框架。实例是关于父母提供的鼓励性环境是否会影响儿童的认知发展,分别采用传统的多重线性回归分析、不考虑协变量和考虑协变量的因果中介效应分析,通过比较3种分析方法所得到的结果,得出如下结论：①当资料中存在中介变量时,不适合采用传统的多重线性回归分析取代因果中介效应分析;②当资料中存在协变量时,不适合在忽视协变量的条件下进行因果中介效应分析。     

Treating control of voicing in apraxia of speech with variable practice

Background: Apraxia of speech (AOS) is generally considered a phonetic‐motoric disorder. As such, it is reasonable to draw on the motor learning literature to develop interventions for improving articulation. The often cited problem of impaired voicing control is used to test the application of a variable practice approach to training skilled movements in AOS. It is predicted that variable practice—practising a behaviour over a range of possible values or contexts—increases accuracy and stability of a trained behaviour.

Aims: The aims of the study were to test the influence of variable practice conditions on acquisition and long‐term maintenance of voiced and voiceless phonemes in words at the phrase/sentence level as well as on generalisation of treatment effects to phonemes of same and different manner. The protocol was tested on two individuals with moderate AOS, one of whom exhibited a concomitant moderate aphasia.

Methods & Procedures: A single subject multiple baseline across‐subjects design was used to examine efficacy of treatment for improving control of voicing for three fricative/affricate phonemes (Participant 1) or three plosive phonemes (Participant 2). The training sets included two voiced phonemes and one voiceless phoneme, each presented in the initial position of 10 different words. This provided a range of voice onset times and a range of phonetic contexts for each target phoneme, thus providing the variable practice. The inclusion of a voiceless phoneme that was produced at a high level of accuracy during baseline was hypothesised to reduce the chance of overgeneralisation from voiced to voiceless phoneme production.

Outcomes & Results: Consistent with predictions, participants demonstrated improved production of trained voiced phonemes and generalisation of treatment effects to untrained phonemes of same manner only. These effects were maintained up to 3 months post‐treatment. Results support training multiple fricatives or plosives in parallel in a variety of phonetic contexts (i.e., variable practice) as well as including a combination of voiced and voiceless phonemes in treatment to ensure maintenance of accurate voiceless phoneme production.

Conclusions: Results from two participants, varying in overall communication impairment severity, provide promising evidence that a variable practice approach to retraining speech behaviours in AOS is effective. Specifically, this approach resulted in long‐term maintenance of treatment effects and generalisation of treatment effects to untrained phonemes within manner class. Just as importantly, it did not result in overgeneralisation of voicing to voiceless phonemes. Further testing is required before recommending general clinical application.     

Familial adult amyotrophic lateral sclerosis: report of cases

We examined 8 cases of familial ALS (amyotrophic lateral sclerosis) in three different families from our province, admitted to our hospital between 1970 and 1989. Clinical criteria for diagnosis were satisfied in all cases; EMG was performed in 6 out of 8 patients. 4 cases showed classical onset and 4 cases bulbar onset. The average age at onset was 65.7+10.6 years. The average survival was 19.1+9.2 months. In two families two generations were affected, in the other only one. The mode of transmission was found to be autosomal dominant with variable penetrance. Neither environmental nor toxic factors seemed to be involved in the development of the illness. Genetic investigations may help to elucidate the pathogenesis of familial ALS.

---

Sommario Abbiamo esaminato otto casi di SLA familiare (Sclerosi Laterale Amiotrofica) appartenenti a tre diverse famiglie isolate nella nostra provincia e ricoverati nel nostro Istituto tra il 1970 ed il 1989. I criteri clinici di diagnosi sono stati soddisfatti in tutti i casi e l'EMG è stato eseguito in 6 casi su 8. Quattro casi mostravano un esordio di tipo classico e quattro un esordio di tipo bulbare. L'età media di insorgenza della malattia nelle tre famiglie era di 65.7+10.6 anni. La sopravvivenza media era di 19.1+9.2 mesi. La modalità di trasmissione era autosomica dominante con penetranza variabile. In due famiglie erano affetta due diverse generazioni, nella terza famiglia una sola. Fattori ambientali non sembrano essere coinvolti nello sviluppo della malattia. Ulteriori studi genetici potranno far luce sulla patogenesi della SLA familiare.

     

基于因果图模型应用调整集估计数据的因果效应

本文目的是介绍因果图模型的基础知识、因果图过程的内容以及基于SAS/STAT中的CAUSALGRAPH过程构造和搜索调整集的方法。因果图模型是图论与概率论相结合的产物，它可以基于用户设定的变量之间的作用关系找到包含最小调整集在内的所有可能的调整集。因果图过程的内容主要包括三种识别标准、两种操作模式和一种验证检查方法。本文基于SAS中因果图过程对两个实例进行因果效应分析，并对输出结果做出解释。     

Phenotypic differences of human neutrophils of carriers of the PSGL-1 A and B-allele in binding to immobilised P-selectin under flow conditions

P-selectin glycoprotein ligand-1 (PSGL-1) interacts with P-selectin expressed on endothelial cells and platelets. PSGL-1 extracellular mucin-like domain displays a variable number of tandem repeats (VNTRs) polymorphism. The wildtype consists of 16 decameric repeats (designated A isoforms) and variants with 15 (B allele) and 14 (C allele) repeats that are assumed to be associated with reduced risk of vascular disease. We investigated the adhesion of these natural variants to P-selectin in native human neutrophils. Healthy volunteers were genotyped and the adhesion of neutrophils expressing the PSGL-1 isoforms A/A, A/B and B/B were studied under static and physiologic flow conditions. Homozygous B/B neutrophils attached significantly weaker to P-selectin at elevated shear rates from 24 up to 64 dyn/cm² than A/A and A/B neutrophils. No difference in adhesion rate was found under static conditions and shear stress below 24 dyn/cm², but B/B neutrophils rolled significantly faster than A/A neutrophils at shear stress ≥ 12 dyn/cm². There was no difference in the adhesive capacity between A/A an A/B neutrophils. These data support the view that the role of the decamers is to extend the ligand binding domain far above the cell surface to support stable leukocyte adhesion and rolling.     

Moving from the quality of life concept to a theory

Background The Special Interest Quality of Life Group has updated its set of statements defining the quality of life (QOL) construct to reflect emerging areas of agreement and the framework for understanding better the QOL construct. Method This article examines the major areas currently under discussion involving the objective–subjective dichotomy, needs, and core domains. Results It is concluded that while the new statements constitute a significant advance, further progress requires testable theory. In order to facilitate such future research, a conceptual model is proposed that distinguishes causal and indicator variables within the framework of a homeostatic management system. Conclusion Several lines of empirical investigation are suggested to test this and similar theoretical models with a view to taking our conceptualization of QOL to the next level.