The main core of Structural Equation Modeling (SEM) is the parameter estimation process. This process implies a variance-covariance matrix Σ that is close as possible to the sample variance-covariance matrix of data input (S). The six Sigma survey uses ordinal (rank) values from 1 to 5. There are several weighted correlation coefficients that overcome the problems of assigning equal weights to each rank and provide a locally most powerful rank test. This paper extends the SEM estimation method by adding the ordinal weighted techniques to enhance the goodness of fit indicators. A two data sets of the Six Sigma model with different statistics properties are used to investigate this idea. The weight 1.3 enhances the goodness of fit indicators with data set that has a negative skewness, and the weight 0.7 enhances the goodness of fit indicators with data set that has a positive skewness through treating the top-rankings.
The traditional way for statistics in any statistical book starting with descriptive statistics, followed by Probability and ends with inferential Statistics. The probability is considered as the link between descriptive and inferential statistics. Inferential Statistics has wide definition and is defined as “the branch of statistics concerned with using sample data to make inferences about a population. In inferential statistics, predictions are made and conclusions are drawn for the target population based on the sample”. The main topics of inferential statistics are Estimation, Testing Hypotheses about means, variances, goodness of fit and proportions, Correlation, Regression and Time series. In this article we are trying to organize the statistics by splitting the inferential statistics into two parts, namely Diagnostic Statistics and Predictive Statistics and explaining the importance of each part. Also we will discuss a Perspective Statistics. Based on that we will have four levels in statistics that can be used to analyze data (Descriptive, Diagnostic, Predictive and Perspective Statistics). Descriptive statistics are primarily concerned with graphs, frequency tables, measures of central tendency, measures of variation, and measures of shape. Diagnostic statistics are primarily concerned with the effects of the Independent variables (inputs) on the Dependent (Target) variable, as measured by Tests of Correlation or Association, Tests for Mean Differences, and Tests for Classification. Predictive statistics are primarily concerned with estimation, regression techniques, and time series analysis for the dependent (target) variable. Perspective statistics mainly related to the previous three levels and acts as a prescription to how to solve or prevent the problem, which means to take a decision in advance. In this article, we will clarify idea through giving an example on a real data related to Gynecologic Cancer, and show how the perspective analytics can prevent it.
'/%3e%3cpath fill='%23fff' d='m8.751-17.959 10.11 11.373L3.997-9.844l13.94-6.1-7.692 13.129z'/%3e%3c/svg%3e)
