China
Africa
Explore chapters and articles related to this topic
Cluster Analysis
Published in M. Venkataswamy Reddy, Statistical Methods in Psychiatry Research and SPSS, 2019
While few but well-defined methods are available in determining the number of components in factor analysis (Eigen values of one and above, Scree test, and Horn’s Parallel Analysis), numerous methods of determining the number of clusters in the data have been proposed. Some of these methods are based on intuition like examining the dendrogram for large number of clusters between fusions, and the minimum number of entities in the ultimate clusters.
Correlational-based methods
Published in Claudio Violato, Assessing Competence in Medicine and Other Health Professions, 2018
Factor analysis is a collection of methods used for exploring the correlations between a number of variables seeking the underlying clusters or subsets called factors or latent variables. According to the principles of factor analysis, variables correlate because they are determined in part by common underlying influences. Patterns of correlations among individual personality variables, for example, are thought to reflect underlying processes that effect students’ behaviors and performance. “Conscientiousness” is thought to be a personality trait characterized by organization, purposeful action, self-discipline, and a drive to achieve. These behaviors, therefore, should be highly correlated.
Morphometries of Craniofacial Form
Published in D. Dixon Andrew, A.N. Hoyte David, Ronning Olli, Fundamentals of Craniofacial Growth, 2017
We now turn to an allied method called principal components analysis which is frequently used in morphometries. The difference between these two computationally similar procedures, factor analysis and principal components analysis, has engendered considerable confusion. The major emphasis of factor analysis is to obtain a small number of easily interpretable variables, or factors, that reveal the essential information (underlying structure) present in the data set. These factors are intended to explain the interrelationships between the original variables. In contrast, principal components analysis is primarily focused on extracting a set of components which explain as much of the total variance as possible.
Assessment of reliability and factor structure of the hypnotic induction profile (HIP) scale
Published in American Journal of Clinical Hypnosis, 2022
Shahyad Somehsaraei Sabet, Mohammad Ali Rahmani, Susan Emami Pour, Hasti Atashi Shirazi
Evaluation of the reliability of the scale: In order to evaluate the reliability of Hypnotic Induction Profile (Induction score and Profile score), two methods, content validity, and factor analysis methods were used. In order to determine the validity of the content, the opinions of experts were used. In this regard, the test was given to the five clinical psychologists and psychiatrists who were sufficiently familiar with the subject of hypnosis, and then appropriateness of the test materials was evaluated. Modifications in some of the materials were made through editing of the translation, thus ensuring that hypnotizability was measured with items mentioned on the HIP scale, and therefore prepared for performance on the samples. Before performing factor analysis on the test results, five assumptions were considered: Adequacy of sampling.Assurance that the data correlation matrix in the dataset is not equal to zero.The factor load of each item in the factorial matrix and the rotated matrix should be at least 0.3 and preferably higher.Each factor must belong to at least three items.Factors must have sufficient validity.
Design and psychometric properties of a questionnaire for assessing sexual and reproductive health needs of married adolescent women: an exploratory sequential mixed methods study
Published in Journal of Obstetrics and Gynaecology, 2022
Ashraf Ghiasi, Afsaneh Keramat, Farid Zayeri, Maryam Farjamfar, Katayon Vakilian, Leila Bagheri
Some limitations of this study should be mentioned. First, subjects were recruited from two counties in the northeast Iran, which may limit generalisability of our findings to other regions. Another limitation is the sample size, which was the minimum required for factor analysis and not relatively large. The other limitation is the long length of the questionnaire, which might have led to participants’ boredom and could have influenced the accuracy of the participants when completing the questionnaire. A further limitation of the current work is lack of comparable valid and reliable instruments in the literature. In addition, the possibility of response bias is an inherent problem with any self-report measure. Our research also has some strengths. The questionnaire is developed based on the experiences of target group and a comprehensive literature review. Selection of MAW from urban–rural areas is other strength. Another strength of this study is utilising a mixed methods sequential explanatory research design. Moreover, the psychometric properties of the questionnaire were assessed through analyses of its face, content, construct validity, internal consistency and stability.
Bayesian factor models for multivariate categorical data obtained from questionnaires
Published in Journal of Applied Statistics, 2021
Vitor Capdeville, Kelly C. M. Gonçalves, João B. M. Pereira
The normal assumption for the latent factors is a common choice in the standard factor analysis, mainly for facilitating analytical manipulations derived from multivariate normal distribution properties. In general, the latent factor distribution is changed in according to the error assumption. For example, to deal with a multivariate t-distribution, [23] replaced the normality assumption not only in the error distribution but also in the distribution of the latent factors, [22] deal with Poisson distribution for the errors and showed that assuming normal and gamma distributions for the latent factors yield, under special cases, to known marginal distribution for the response variable. In this work, although we only considered the normal distribution for the latent factors, taking advantage of the conjugation with the model assumed in (2) for the errors, other distributions could be assumed.