• 응용통계연구
• /
• 제5권1호
• /
• pp.19-28
• /
• 1992

본 논문에서는 부분균형배열을 이용하여 $2^t$ 요인 실험계획에서 최소의 실험횟수를 가지 고 2인자 교호작용까지 분석할 수 있고, 추정량의 분산-공분산행렬이 균형을 이루고 있는 실험계획법을 기술하였다. 그리고 제안된 실험계획법의 최적성을 검토하여 트레이스 최적실 험계획을 찾아보았다. 이러한 최소균형 Resolution V $2^t$요인 부분실험법은 직교배열을 이용하여 주효과만 분석하는 기존의 다구찌방법과는 달리 2인자 교호작용까지 분석하고자 할 때 효과적으로 사용될 수 있으며 그 응용방법을 검토하였다.

• Journal of the Korean Statistical Society
• /
• 제32권3호
• /
• pp.235-244
• /
• 2003

In unreplicated factorial or fractional factorial experiments, the presence of one or more outliers can seriously affect the analysis of variance. Using the normal plot of t residuals to identify outliers in factorial or fractional factorial is an easy method to find these dubious points. In some cases, the t residuals form the identical pairs. One can not tell from the plot which is doubtful. This phenomenon occurs for all minimal designs of resolution IV, which fits the model containing all main effects and some two-factor interactions, whether it is orthogonal or not. In these kinds of models, when we drop one point or two points (not foldover pair) from the fraction, the phenomenon of identical pairs of t residuals may still occur. In this paper, the theoretical background of the phenomenon and its sequences will be investigated in detail.

• Journal of the Korean Statistical Society
• /
• 제21권1호
• /
• pp.80-86
• /
• 1992

The concept of Balanced Factorial Experiment (BFE) was introduced by Shah (1958). The conditions for BFE were set up by Kurkjian and Zelen (1963) and Kshirsagar (1966). Generalized Cyclic Factorial Experiment (GCFE), which is more wide class of designs than BFE, do not satisfy the condition of BFE. So all contrasts belonging to the same interaction are not estimated with equal variance. The main purpose of this paper is to show that GCFE have orthogonal factorial structure and the scheme of the size of variances for all normalized contrasts in GCFE is similar to the original intra-block association scheme.

• 품질경영학회지
• /
• 제40권1호
• /
• pp.15-24
• /
• 2012

For the construction of fractional factorial designs, the various arrays can be widely used. In this paper we review the statistical properties of fractional designs constructed by two arrays such as orthogonal array and partially balanced array, and develop a quick and easy method for analyzing unreplicated saturated designs. The proposed method can be characterized that we control the error rate by experiment-wise way and exploit the multivariate Student $t$-distribution. Especially the proposed method can be used efficiently together with some exploratory analysis methods, such as half normal probability plot method.

• Journal of the Korean Statistical Society
• /
• 제8권2호
• /
• pp.125-130
• /
• 1979

It is well known that $L_2(m)$ type PBIB designs have the Property A, so they are BNAS PBIB designs. However, $L_3(m)$ type PBIB designs are not of type of Property A but do have the factorial structure (Cotter, John, and Smith(1973)). In this paper, the properties of the $L_3(m)$ type PBIB designs are investigated. Extended Property A and fractional BNAS are defined and a solution formula for the treatment effects in the $L_32(m)$ type designs is obtained.

• 품질경영학회지
• /
• 제28권3호
• /
• pp.59-67
• /
• 2000

In analysis of resolution IV fractional factorial experiments, the main effects only are analyzed, even though we can get some useful information on the confounded 2-factor interactions. In this paper, we introduce an exploiting method of the confounded structure of interactions, especially for the near minimal resolution IV 3$^{t}$ fractional factorial designs developed by Anderson and Thomas (1979). Moreover, in this paper the application way of the proposed method is also discussed by analyzing some simulated data.

• Communications for Statistical Applications and Methods
• /
• 제6권3호
• /
• pp.695-706
• /
• 1999

When fractional factorial experiments contain some infeasible treatment combinations called debarred combinations we should construct experimental designs so that those debarred combinations are to be excluded by selecting defining contrasts appropriately. By applying Franklin(1995)'s procedure for selecting defining contrasts to Cheng and Li(1993)'s method this paper presents a method of selecting defining contrasts to construct orthogonal 3-level fractional factorial experiments which exclude some debarred combinations.

• Communications for Statistical Applications and Methods
• /
• 제15권2호
• /
• pp.235-244
• /
• 2008

실험계획의 요인배치법에서 부분실험을 설계할 때, 직교배열을 이용한 실험설계방법이 널리 사용된다. 그러나 부분실험의 해상도(resolution)가 큰 경우, 직교배열을 일반화한 균형배열이 효과적으로 사용될 수 있다. 특히 2-수준 요인실험법에서는 강도(strength)가 4인 균형배 열은 resolution-V인 부분실험법과 동일하다는 것이 알려져 있다. 본 연구에서는 강도(strength)가 4인 균형배열의 직교성을 평가하는 측도를 제시하고자 한다. 그리고 본 연구에서 제시된 평가측도를 응용하여 실험횟수가 가장 적고 직교성에 가까운 최소 균형 resolution-V 부분실험법의 설계방법을 제시하고자 한다.

• 한국연초학회지
• /
• 제5권1호
• /
• pp.49-56
• /
• 1983

• 품질경영학회지
• /
• 제46권1호
• /
• pp.27-38
• /
• 2018

Purpose: A large number of attributes with mixed levels are often considered in the conjoint analysis. In the cases where attributes have two or three levels, we research on the efficient design of survey questionnaire to estimate all the main effect and two factor interaction effects with a reasonable size of it. Methods: To reduce the number of questions in a questionnaire, the balanced incomplete block mixed level factorial design with minimum aberration was proposed by Lim and Chung (2016). Based on the number of questions and that of the respondents in that design, D-optimality criterion is adopted to find efficient designs where the main effect and two factor interaction effects are estimated. Results: The list of the number of questions and that of the respondents in efficient designs for survey questionnaire are recommended based on the D-efficiency of each design and the proposed selection criteria for the number of both questions and the respondents. By analyzing all the respondents survey data generated by the simulation study, we find the proper model. Conclusion: The proposed methods of designing survey questionnaires seem to perform well in the sense that how often the proper model is found in a simulation study where all the respondents survey data are generated by the simulation model.