交叉表之描述方法

A Method for Cross-Table Description

賴守仁(Shou-Ren Lai)

《中國統計學報》 47卷2期 (2009/06) Pp. 129-142

https://doi.org/10.29973/JCSA.200906.0004

摘要

交叉表是最常被使用的統計資料陳述方式，每個普查或調查報告通常附有許多交叉表，這些交叉表通常隱含寶貴的訊息，然其訊息並無法直接看出，必須經由分析及描述才能顯現。
有關交叉表分析的方法，一般會想到類別資料分析(Categorical Data Analysis)，其中最主要的是對數線性模型(Loglinear Model)，主要在分析各類別之主效應及各種交互作用效應，然而這些效應之訊息對決策助益不大。
對決策有用的訊息是目標類別（應變數）如何受其他類別（自變數）影響之明確敘述，因此要先判斷何類別爲該表之分析目標、何類別爲分析篡礎及分析的順序組合，然後對目標類別如何受其他類別影響作簡單清晰的描述。本研究提出以上分析描述方法，並提出二個分析範例。

關鍵字

交叉表 ； 影響 ； 描述

並列摘要

The cross-table is one of the most popular forms used for statistical data presentation. Almost every census or sampling survey report contains a lot of cross-tables. Such tables often carry precious information. However, the information within a cross-table can not be understood by simply being reviewed. Comprehensive analysis and description are often required to reveal their hidden messages.
Usually, categorical data analysis will be considered as one of the first methods of cross-table analysis. One of the most important models of categorical data analysis is the loglinear model which deals with the main effects and various interaction effects among the categories. However, these kinds of effects may not be that helpful for the decision-making.
The most useful information for decision-making is how the target category (response variable) is impacted by other categories (independent variables). So when analyzing a cross-table, some cause-effects must be clarified, such as: which category is the analyzing target, which one is the analyzing basis, and also the sequence of the categories to be analyzed. Then, a simple and clear description for the impact of other categories on the target category is concluded. The above described analyzing method and two analyzed cases are proposed in this study.

並列關鍵字

Cross-table ； impact ； description