The research of cognitive diagnostic models (CDMs) is becoming an important field of psychometrics. Instead of assigning one score, CDMs provide attribute profiles to indicate the mastering status of concepts or skills for the examinees. This would make the test result more informative. The implementation of many CDMs relies on the existing item-to-attribute relationship, which means that we need to know the concepts or skills each item requires. The relationships between the items and attributes could be summarized into the Q-matrix. Misspecification of the Q-matrix will lead to incorrect attribute profile. The Q-matrix can be designed by expert judgement, but it is possible that such practice can be subjective. There are previous researches about the Q-matrix estimation. This study proposes an estimation method for one of the most parsimonious CDMs, the DINA model. The method estimates the Q-matrix for DINA model by setting constraints on the generalized DINA model. In the simulation study, the results showed that the estimated Q-matrix fit better the empirical fraction subtraction data than the expert-design Q-matrix. We also show that the proposed method may still be applicable when the constraints were relaxed.

筆者在論文中引入一個名為加權距離檢驗的雙變項統計。此檢驗方法用於比較兩個隨機變數的集中趨勢。其優勢在於在有偏度的數據中，仍能穩定地控制第一型錯誤，並同時提供可觀的統計檢力。加權距離檢驗利用冪函數修正在偏度數據中的不對稱現象。與一般的冪函數轉換法不同，加權距離檢驗將冪值限制在0和1之間。文中亦提供了一個有效決定冪值的方法，以方便在實際運算中使用。 / 筆者總結了四個主流的雙變項統計方法，並利用蒙地卡羅模擬法在正態分佈、同程度偏度分佈以及不同程度偏度分佈三個情況中比較了它們與加權距離檢驗的表現。結果顯示，加權距離檢驗雖然沒有在任何一個情況中勝出，但卻於兩方面表現了其優勢。首先，它在任何情況下都能把第一型錯誤控制在合理水平之下；其次，它在任何情況下都不至於表現得太差。反觀其他四個檢驗方法總會在某些情況下表現失敗。由此可見，加權距離檢驗比起其他檢驗方法更能提供一個穩定而簡單的方法去比較集中趨勢。 / A new bivariant statistics, namely the weighted distance test, for comparing two groups were introduced. The test aims at providing reliable type I error control and reasonable statistical power across different types of skewed data. It corrects the skewness of the data by applying power transformation with power index ranged between 0 to 1. I also proposed in this thesis a possible way of deciding the power index by considering the skewness difference between the two groups under comparison. / I reviewed 4 commonly used inferential statistics for two-group comparison and compared their performances with the weighted distance test under 1) normal distribution, 2) skewed distribution with equal skewness across groups, and 3) skewed distribution with unequal skewness across groups. Monte Carlo simulations were ran to evaluate the 5 tests. Results showed that the weighted distance test was not the best test in any particular situation, but was the most stable test in the sense that 1) it provided accurate type I error control and 2) it did not produce catastrophically small power in any scenario. All other 4 tests failed in some of the simulated scenario for either inflated type I error, or unsatisfactory power. Therefore, I suggested that the weighted distance test could be one easy-to-use test that works fairly well across a wide range of situation. / Detailed summary in vernacular field only. / Detailed summary in vernacular field only. / Lee, Yung Ho. / Thesis (M.Phil.)--Chinese University of Hong Kong, 2012. / Includes bibliographical references (leaves 31-33). / Abstracts also in Chinese. / Chapter Chapter One --- : Introduction --- p.1 / Common methods in comparing central tendency --- p.2 / T-test --- p.2 / Median and rank --- p.3 / Trimming --- p.3 / Power transformation --- p.4 / Chapter Chapter Two --- : Weighted distance statistic --- p.5 / Definition --- p.5 / Statistical properties --- p.5 / Specification of Lambda λ --- p.7 / Estimation and inference --- p.7 / Chapter Chapter Three --- : Simulation --- p.10 / Study 1 --- p.12 / Study 2 --- p.15 / Study 3 --- p.18 / Chapter Chapter Four --- : Discussion --- p.21 / Summary --- p.21 / Limitation --- p.22 / Further development --- p.23 / Chapter Appendix I --- : Proofs of theorems of weighted distance statistic --- p.24 / Chapter Appendix II --- : Table of numerical results of simulations --- p.26 / Bibliography --- p.30

This research used middle managers as subjects to examine the relationships among sets of personal variables, leadership style predisposition in terms of task orientation and relationship orientation, and leaders' performance. A new model called the idiographic leadership model, that relates these variables, was introduced and tested in two parts In Study One the personal variables were conceptualized as personality traits. Ninety eight male police sergeants completed the Edwards Personal Preference Schedule, Blake and Mouton's Managerial Grid, and a self evaluation of leadership effectiveness. Each subject's performance was also evaluated by his supervisor. Task orientation and relationship orientation were found to be meaningful measures of leadership style predisposition. Using backwards elimination, regression equations of order and succorance on task orientation and deference and nurturance on relationship orientation were developed. Leaders with a task orientation performed better on tasks involving structure and objective goals than did leaders with a relationship orientation and the latter performed better on tasks involving personal contacts and orientation, but the differences were not statistically significant. The differences in performance between leaders with a congruent personality and leadership style and leaders for whom personality and leadership style were not congruent were not significant Exploratory Study Two employed 24 educational administrators as subjects. In this study the personal variables were measured as life stress and negative affects. Partially because of the small sample size, many of the relationships that were discovered were not statistically significant, but several themes for future research were developed. Life stress, as measured by Holmes and Rahe's Social Readjustment Rating Scale, was related to elevated negative affects (anxiety, depression, and hostility), as measured by the Multiple Affect Adjective Checklist, but the relationships were not significant. Both life stress and negative affects seemed to be related to task orientation, relationship orientation, and leader effectiveness, but only the correlations to effectiveness were significant / acase@tulane.edu

Thesis (M.Soc.Sc.)--University of Hong Kong, 2006. / Title from title page (viewed Apr. 23, 2007) Includes bibliographical references (p. 38-43).

Thesis (Ph. D.)--University of North Carolina at Greensboro, 2006. / Title from PDF title page screen. Advisor: Terry Ackerman, Robert Henson; submitted to the School of Education. Includes bibliographical references (p. 121-125).

Retrospective pretests provide a potentially useful elaboration on research methodology if they can be shown to be dependable under specific sets of conditions. Previous studies have examined response-shift bias and response-style effects, but less attention has been given to memory distortion associated with the retrospective recall of diverse types of variables. Identifying psychometric characteristics of these measures may help to clarify the picture emerging from retrospective accounts. The present study applied a methodology developed to measure the systematic error (i.e., memory distortion) that may be associated with variables involving a range of recall tasks. The study examined which types of variables account for the least measurement error in retrospective pretests administered at three time points. The types of variables examined in this study include students' self-ratings of academic abilities, self-reported attitudes and opinions about college, mood states, and perceptions of general health. The results of this study indicate that there was no main effect of time on any of the pairs of difference scores, and a moderate level of memory distortion was detected in the three variable types examined. The methodology applied provides an effective approach to understanding the effect of memory distortion on retrospective pretest variables. The author recommends that future applications of this methodology be applied to heterogeneous populations, investigate a range of complex variables, and include an examination of individual subject differences.

Within the past few decades, methodologists have made major advances in statistical methods for the analysis of change using longitudinal panel data, particularly in the area of modeling individual differences (Bryk & Raudenbush, 1987; Collins & Horn, 1991; Rogosa, 1991; Willett & Sayer, 1994; Willett, Singer, & Martin, 1998). These advances have made it possible for researchers to measure change and the correlates of change in ways that were not thought possible a few decades ago. These improvements should allow researchers to make stronger and more informed inferences regarding change over time. Despite the improvements individual growth modeling methods represent for the analysis of change, it remains unclear as to their adequacy for informing about individual differences with respect to change. The purpose of the present study was to directly compare three general classes of individual growth modeling strategies with each other and with two commonly used traditional fixed effects models of change in order to assess (a) the conclusions that can be drawn about change in general and about individual differences in change in particular; and (b) the robustness or stability of these various data analytic strategies.

Perceived risk of inherited susceptibility to cancer (ISC) historically has been determined by objective measures, which neither emphasize how people with ISC perceive their risk of cancer, nor address the meaning of being at risk. The few studies of perceived risk of ISC provide sparse data regarding the process of perceived risk. Knowledge of this process is important because perceived risk may affect how people act to reduce their cancer risk. This study tested the investigator-developed perceived risk of ISC (PRISC) model to learn about this process. The PRISC model, a latent variable (LV) model, is based on existing literature and a preliminary phenomenological study of women at high risk for breast cancer. LVs (Awareness, Perceived Risk, Fear, Support, and Action) in the PRISC model cannot be measured directly, rather, the score of the LVs is inferred through measurements of associated indicator variables. The purpose of this cross-sectional study was to: (a) estimate psychometric properties of instruments used to measure PRISC model indicator variables; (b) test the fit of the PRISC model with instrument-generated data; and (c) examine relationships among the LVs. The sample of 200 women had no cancer history and met established criteria for hereditary predisposition to breast cancer. Participants completed 12 self-report instruments. Analysis of instruments included descriptive statistics, estimates of internal consistency, and confirmatory factor analysis. Testing of the PRISC model used structural equation modeling (SEM) techniques. Psychometric analysis indicated eight instruments had acceptable standardized alpha coefficients of at least .70. Items in most scales loaded on one factor. SEM resulted in two models: one fit acceptably with the data but did not support relationships between LVs Awareness and Perceived Risk and Perceived Risk and Support. Respecification of the model (deleting Perceived Risk) resulted in poor fit with the data, but significant correlations between Awareness and Fear, Fear and Action, and Support and Action. Instability of the PRISC model indicated the need for theory-driven respecification and reconsideration of some instruments for future analysis. Although instability precludes generalization of findings, the model suggests that fear and support may positively predict action to reduce cancer risk.

No description available.

No description available.