2016•Journal of Digital ConvergenceOpen access

The Study on Comparative Analysis of the Same Data through Regression Analysis Model and Structural Equation Model

Chang Ho Choi, Yen-Yoo You

Open full text 5 citations

Abstract

본 연구는 인과관계 분석에서 주로 활용되는 SPSS statistic(회귀분석)과 구조방정식모델을 구현하는 프로그램 중 하나인 AMOS 프로그램을 각각 활용하여 동일한 데이터에 대하여 실증분석을 실시하였다. 실증분석 결과, 회귀계수 및 유의확률에서 서로 다른 결과값이 나왔으며, 특히 매개효과 검정에서 귀무가설 기각역 근처의 유의확률값(즉, t값 및 C.R.값의 절대값이 1.96 근처)을 보이는 상황에서 SPSS statistic(회귀분석)에서는 매개효과가 있는 반면, AMOS 프로그램(구조방정식)에서는 매개효과가 없는 것으로 나타났다. 결국, 동일한 데이터임에도 불구하고 어떤 통계프로그램을 활용하느냐에 따라 다른 결과값(특히, 측정오차가 클수록 결과값이 크게 달라짐)이 나올 수 있음을 알 수 있다. This study analyzed empirically the same data through SPSS statistic(regression analysis) and AMOS program(structural equation model) used for cause and effect analysis. The result of empirical analysis was as follows. The different outcome of coefficients and p-values were deducted. Especially, in the mediated effect testing, meanwhile, SPSS statistic(regression analysis) pictured mediated effect, AMOS program(structural equation model) did not picture mediated effect on the reject zone of null hypothesis(absolute t-value and C.R.-value were nearby 1.96). Eventually, this study showed that what program used determined the outcomes of coefficients and p-values(In particular, the outcomes were differentiated further in the increasing measurement error) though using the same data.

Open-access reader

About this research paper

What this paper is about

본 연구는 인과관계 분석에서 주로 활용되는 SPSS statistic(회귀분석)과 구조방정식모델을 구현하는 프로그램 중 하나인 AMOS 프로그램을 각각 활용하여 동일한 데이터에 대하여 실증분석을 실시하였다. 실증분석 결과, 회귀계수 및 유의확률에서 서로 다른 결과값이 나왔으며, 특히 매개효과 검정에서 귀무가설 기각역 근처의 유의확률값(즉, t값 및 C.R.값의 절대값이 1.96 근처)을 보이는 상황에서 SPSS statistic(회귀분석)에서는 매개효과가 있는 반면, AMOS 프로그램(구조방정식)에서는 매개효과가 없는 것으로 나타났다. 결국, 동일한 데이터임에도 불구하고 어떤 통계프로그램을 활용하느냐에 따라 다른 결과값(특히, 측정오차가 클수록 결과값이 크게 달라짐)이 나올 수 있음을 알 수 있다. This study analyzed empirically the same data through SPSS statistic(regression analysis) and AMOS program(structural equation model) used for cause and effect analysis. The result of empirical analysis was as follows. The different outcome of coefficients and p-values were deducted. Especially, in the mediated effect testing, meanwhile, SPSS statistic(regression analysis) pictured mediated effect, AMOS program(structural equation model) did not picture mediated effect on the reject zone of null hypothesis(absolute t-value and C.R.-value were nearby 1.96). Eventually, this study showed that what program used determined the outcomes of coefficients and p-values(In particular, the outcomes were differentiated further in the increasing measurement error) though using the same data.

Why it matters

OpenAlex reports 5 citations for this work. Citation counts describe recorded attention and do not establish research quality.

Key contribution

A contribution statement is not available in the OpenAlex record.

Method / approach

Method details are not available in the OpenAlex metadata.

Main findings

Findings are not separately available in the OpenAlex metadata.

Limitations

Limitations are not available in the OpenAlex metadata.

Applications

Application details are not available in the OpenAlex metadata.

Available abstract

본 연구는 인과관계 분석에서 주로 활용되는 SPSS statistic(회귀분석)과 구조방정식모델을 구현하는 프로그램 중 하나인 AMOS 프로그램을 각각 활용하여 동일한 데이터에 대하여 실증분석을 실시하였다. 실증분석 결과, 회귀계수 및 유의확률에서 서로 다른 결과값이 나왔으며, 특히 매개효과 검정에서 귀무가설 기각역 근처의 유의확률값(즉, t값 및 C.R.값의 절대값이 1.96 근처)을 보이는 상황에서 SPSS statistic(회귀분석)에서는 매개효과가 있는 반면, AMOS 프로그램(구조방정식)에서는 매개효과가 없는 것으로 나타났다. 결국, 동일한 데이터임에도 불구하고 어떤 통계프로그램을 활용하느냐에 따라 다른 결과값(특히, 측정오차가 클수록 결과값이 크게 달라짐)이 나올 수 있음을 알 수 있다. This study analyzed empirically the same data through SPSS statistic(regression analysis) and AMOS program(structural equation model) used for cause and effect analysis. The result of empirical analysis was as follows. The different outcome of coefficients and p-values were deducted. Especially, in the mediated effect testing, meanwhile, SPSS statistic(regression analysis) pictured mediated effect, AMOS program(structural equation model) did not picture mediated effect on the reject zone of null hypothesis(absolute t-value and C.R.-value were nearby 1.96). Eventually, this study showed that what program used determined the outcomes of coefficients and p-values(In particular, the outcomes were differentiated further in the increasing measurement error) though using the same data.

Key concepts: Structural equation modeling, Statistic, Statistics, Regression analysis, Mathematics, Test statistic, Econometrics, Regression

Related papers

Back to paper searchBrowse research topicsOriginal source
The Study on Comparative Analysis of the Same Data through Regression Analysis Model and Structural Equation Model — Research Paper | ScholarLens