| 查看: 996 | 回复: 2 | ||||
[求助]
求助关于RASCH分析的一段翻译
|
| The numeric values of the categories (item scale steps) represent measurement on an ordinal scale – they only indicate relative order along the able-not able (or not participating-participating) continuum, but the differences between scale steps are not equivalent, and the values chosen for them are arbitrary. As a result, adding up the scores for the items to obtain the total score is not an operation that is allowed, based on the rules of mathematics. Therefore, total scores do not reflect position along a continuum that has a true zero point, and constant distances between scale points. (These constitute the definition of a ratio scale, which is what is required for calculating means for a group, or percentage improvement over time). However, research has shown that if the item category values are chosen “reasonably”, the sum of ordinal items corresponds quite well to values on a true ratio scale, at least for intermediate levels of the continuum of total scores (26). Rasch analysis, a mathematical procedure based on Item Response Theory (IRT) has been used to transform a set of scores on ordinal FA items into a score on an interval scale (10, 21, 26). The theoretical assumptions and mathematical manipulations underlying Rasch Analysis-based FA instruments are beyond the scope of this chapter; good introductions to the technique and its application to FA may be found in Bond and Fox (27). |
回复此楼» 猜你喜欢
博士延得我，科研能力直往上蹿
已经有7人回复
-
退学或坚持读
已经有27人回复
-
面上基金申报没有其他的参与者成吗
已经有5人回复
-
有70后还继续奋斗在职场上的吗？
已经有5人回复
-
遇见不省心的家人很难过
已经有22人回复
-
多组分精馏求助
已经有6人回复
zerohead
禁虫 (职业作家)
- 翻译EPI: 466
- 应助: 1 (幼儿园)
- 贵宾: 0.301
- 金币: 8727.8
- 散金: 850
- 红花: 15
- 沙发: 10
- 帖子: 3768
- 在线: 471.7小时
- 虫号: 1041741
- 注册: 2010-06-14
- 性别: GG
- 专业: 逻辑学
【答案】应助回帖
|
The numeric values of the categories (item scale steps) represent measurement on an ordinal scale – they only indicate relative order along the able-not able (or not participating) continuum, but the differences between scale steps are not equivalent, and the values chosen for them are arbitrary. 在每一种分类旁边的数字代表了一个次序，它们只代表在可以或者不可以之间（或者是不参与）的一个先后顺序。但是每个次序之间的差别是不按比例的，这些数字只是随意安排的。 |
2楼2010-08-08 21:01:07
lcg2009
荣誉版主 (著名写手)
-
专家经验: +1 - 翻译EPI: 1
- 应助: 2 (幼儿园)
- 贵宾: 1.53
- 金币: 14235.5
- 散金: 2100
- 红花: 20
- 沙发: 3
- 帖子: 2418
- 在线: 643.2小时
- 虫号: 899699
- 注册: 2009-11-10
- 性别: GG
- 专业: 知识管理
- 管辖: 教师之家
【答案】应助回帖
|
类别的数值（每一项的比例间隔）代表一种定序尺度上的尺寸——它们仅仅代表连续与不连续（或不参与）的相关次序，但比例间隔就不一样了，而且他们的取值是任意的。所以，从数学运算法则的角度来说，把每项相加得到总数的做法是不允许的。 因此，总数不能反映具有真值0点和刻度点之间恒定间距的连续的位置。(这些规定形成了刻度比的定义：要求计算一组数据的平均值或者随时间变化的百分比改善程度)。 然而，研究表明：如果项目种类值选择合理，项目顺序之和与真值上的比例刻度能很好的对应，这至少适用于总数连续的中间部分。 基于应理论的数学程序——拉希分析用于把FA项序数的数集转换成等距尺度。相关的基于FA的拉希分析理论假设与计算过程超出了本章的范围，这个方法的介绍及其在FA方面的应用可以参考Bond与Fox的文章。 |
3楼2010-08-10 13:12:58