线性结合可靠性-中英版
结果显示：信任、相关控制、关系满意度以及关系承诺运用在组织公共关系评价（OPRs）上是可行的。 信任度、关系满意度、承诺关系的指数α数值，以及相互控制关系的系数分别为71- , . 79 , . 73，和，58，。 这4个可靠性构建的数值分别为：74 , . 80 , . 72 , 和 62，。 在这4个数值中，相关控制的内在一致性和可靠性最低。其它3个构建数值的内在一致性和可靠性要高得多，达到了验收标准。
α数值可靠性和载荷指数的构建见表格1 第14项。可靠性的相关数值表明项目之间良好的内在一致性，然而，结合第14项目可靠性的数值，使用公式对内部结合的可靠性计算出的数值很高(89) (1978，Nunnally)。因此，可以考虑对第14项目使用新样品进行进一步的数据测试。
-规模纯化，第一阶段：CFA，第2步骤： -使用标准计算机程序EQS(Bentler，1992)重新引导-CFA。 这个步骤的目的是评价假设指数的适应性（Gentler和Newcomb，1986）。
CFA逻辑测试能成功分析探索性指数的两个原因是：首先,根据霍伊尔和史密斯(1994)的理论, CFA适合结构合理性得出来的假设，如指数的数量问题(即,潜在变量)，强调项目对测试的反映，以及因素之间的关系。其次，CFA使用的共差结构分析，能够提供试验统计数据，建议采用适合模型的（观测数据霍伊尔，1991）。
在CFA中，观测数据模型的范围(变化和共差项目)通过不同的拟合度指数显示。 根据Bentler和Bonett(1980)的理论，引进数个指数并且进行推广，拟合指数用于模型的评估，可以避免一些样品数量和分配不符合规格的问题。
他们已经在考虑实际应用意义(Hair et al.,1995). 因为在这项研究的过程中，项目可能受到交叉文化因素的影响，我在变量因数选择的过程中采用较高的标准。按照Galassi，Schanbcrg和韦尔Ware (1992)使用45变量共性作为标准。(注意：最小二乘法通常用作对Galassi et al研究的估算),我决定使用可比较价值数值(即,载荷因数；65)作为标度，确定一个规定的因素过程中，选择的一个变量。
74 huang
表1 第一阶段净化标度结果摘要
标度 标记 项目序号 可靠性指数 构造可靠性 项目 载荷指数
信任 YTR1 3 0.71 0.74 R1
R6
R13 0.79
0.80
0.79
关系承诺 YCM 4 0.73 0.72 R4
R9
516
R17 0.66
0.77
0.84
0.69
关系满意度 YST 4 0.79 0.80 R2
R7
R8
R12 0.74
0.71
0.84
0.85
相关控制 YMT 3 0.58 0.62 R3
R5
R10 0.82
0.65
0.75
线性结合可靠性 0.89
*R表示数据收集的第一阶段项目。**为项目的尺寸。
提议中合适的指数和评论,( 例如,Bentler，(990，1992； Gerbing &安德森，1993； Hu & Bentler，1995，1998，1999； McDonald& Marsh，1990； Tanaka，1993)，目前的评论和EQS提供材料属于CFI(Bentler，1990； Hu & Bentler，1995，1998； Sideridis，Kaissidis和Padeliadu，1998； Whang &汉考克，1997)。按照 Bentler(1990)和Hu *Bentler的建议(1995)， CFE是一个非常强大而且合适的指数，考虑了所有相关的问题(例如样品量，估计方法影响，正常状态破坏的影响，以及独立性)。通常，CFI数值的范围从90到1.00；一般认为反映出给良好适合度。
对起确定作用的分析来说，我采用摩根和亨特的(1994)建议，将建议的模型与竞争者模型比较。本质上来看，两个其它模型的倾斜模型适合的数据比较，决定测试项目反映出来的潜变量，
在这3 个模型之间的比较必须遵循下列问题。 OPRs 可以有目的就假设的尺寸数量进行描述吗？ OPRs数值是相互联系还是分开的(分量)？ 如果是他们之间的相互联系，他们的相互联系使用同一尺寸，而不是多重互相依存的尺寸？ 与模型比较，Akaike信息标准(AIC)用来选择最适合的模型(Naniwa和Ishiguro，ECato。1996; 波尔森，Juhl，克里斯滕森，贝克和Engelund，1996；Vinck，Vlietinck和Fagard，1999)。
A linear combination of reliability
Result shows: trust, relevant control, relationship satisfaction and relationship commitment (OPRs) as well as the use of public relations of the organization evaluation shall be feasible. Trust, relationship satisfaction, commitment relationship index of alpha values, as well as the mutual relationship between control coefficient, respectively, 71 -, 79, 73, 58, these four building reliability values are: 74, 80, 72, and 62. In the four values, related to control the internal consistency and reliability of the lowest. Other three build internal consistency and reliability of the numerical much higher, up to the acceptance criteria.
Alpha numeric reliability and load index construction item 14 see table 1. The reliability of the related numerical show that good internal consistency between projects, however, according to the reliability of the numerical 14 project, using the formula of internal combined with the reliability of the calculated value is very high (89) (1978, Nunnally). As a result, it can consider to 14 projects using the new samples for further test data.
- the size of the purification, the first stage: the CFA, step 2: - use the standard computer program EQS reboot - the CFA (Bentler, 1992). The adaptability of the purpose of this step is to evaluate hypothesis index (Gentler and Newcomb, 1986).
CFA logical test can successfully analysis exploratory index is two reasons: first, according to hoyle and Smith (1994) theory, the CFA out suitable structure rationality hypothesis, such as index of the problem of the number of potential variables (i.e.,), emphasizes the project, responding to the test, and the relationship between the factors. Second, the CFA using total differential structure analysis, can provide the test statistics, recommend suitable model (observation data hoyle, 1991).
In CFA, the range of observed data model (change and poor project) through different fitting degree index. According to Bentler and Bonett (1980) theory, the introduction of several index and promotion, fitting the evaluation index is used to model, can avoid some sample quantity and distribution does not conform to the specifications of the problem.
They have in consideration of practical application significance (Hair et al., 1995). Because in the process of the study, project may be influenced by cross-cultural factors, I used in the process of variable factor to choose the higher standard. According to Galassi, Schanbcrg and weir Ware (1992) 45 variables in common use as standard; (note: least squares method is usually used for estimating of the Galassi et al research), I decided to use a comparable numerical value (that is, load factor; 65) as the scale, determine a regulation factors in the process of selection of a variable.
Huang, 74
The results in this paper is as follows, the first stage purification scale in table 1
Scale mark project serial number reliability index of structural reliability project load index
Related control YMT 3 0.58 0.62 R3
A linear combination of reliability (0.89)
The first phase of the project * R data collection. The size of the * * for the project.
Proposed proper index and commentary, (for example, Bentler, (2; 990199 Gerbing & Anderson, 1993; Hu & Bentler, 1995199, 8199, 9; McDonald & Marsh, 1990; Tanaka, 1993), current comments and EQS provide material belongs to CFI (Bentler, 1990; Hu & Bentler, 1995199 8; Sideridis, Kaissidis and Padeliadu, 1998; Whang & Hancock, 1997). According to the Bentler (1990) and Hu * Bentler suggestion (1995), CFE is a very powerful and appropriate index, considering all the relevant issues (such as sample weight, estimation method, the influence of the normal damage, and independence). Usually, the range of CFI from 90 to 1.00; is thought to reflect to the good fitness.
To determine the role of analysis, I use of Morgan and hunt (1994) suggested that the suggested model model to compare with the competitors. In essence, two other models of slope model is suitable for data comparison, decided to test project of latent variables,
In the comparison between the three models must follow the following questions. OPRs can have purpose is assumed to describe the size of the amount? OPRs values are interrelated or separate (weight)? If is a connection between them, they connect with each other using the same size, rather than the size of the multiple interdependent? Compared with the model, the Akaike information criterion (AIC) used to select the most appropriate model (Naniwa and Ishiguro, ECato. 1996; paulson, Juhl, christensen, baker and Engelund, 1996; Vinck, Vlietinck and Fagard, 1999).
2014.6.6