Search peer-reviewed journals and articles

In many fields of research including Education, Psychology, Social Science, and Marketing Research, data is generally collected through a questionnaire, which is often referred to as the survey instrument. For the scale development of these questionnaires two different approaches, Classical Test Theory and Item Response Theory, are mostly prevalent. In our study a comparison is made between these two theories by the means of item parameters as well as person parameters. Guttman scaling is selected for the classical test theory, whereas one and two parameter models are used for the item response theory. Our data consists of an instrument containing 64 items assessing the adolescent life of hardcore criminals, petty criminals, and community persons. The study reveals that the difficulty parameters of classical test theory are highly negatively correlated with one parameter model of item response theory whereas the discrimination parameters of Classical test theory and two parameter model of Item response theory are uncorrelated. The present study also reveals that the person parameters of classical test theory, and parameters under two models of Item response theory are highly positively correlated.

Keywords: Classical test theory, Item response theory, Item parameters, Person parameters.

Agarwal, G.G., & Asthana, A. (2022). Comparison of Classical Test Theory and Item Response theory in the Discrimination of Hard Core and Petty Criminals. Journal of Applied Statistics and Machine Learning. 1(1): pp. 113

Mycobacterium tuberculosis is an infectious agent of human tuberculosis whose genome revealed 12 membrane proteins of MMPL family. Accurate description of amino acids of MMPL12 is very important in prediction of structure and calculation of energy which is done with the help of Bioinformatics tools and software approach. Ramachandran plot of the ö, ø values for the amino acids in a MMPL12 protein is further done. Using this computational approach, the total energy of MMPL12 in Mycobacterium tuberculosis is found to be 38098.148KJ/mol. Repeating energy trends at each of the molecular, functional group, and atomic levels is also observed. Retinols and retinoic acid play essential characters in variation of gene expression and overall growth of embryo in Mycobacterium tuberculosis.

Keywords: Retinol; MMPL12; Adipose tissue; P value; Computational energy.

Tiwari, A., Katiyar, A., Srivastava, S., Sazena, V.L. & Agarwal, N. (2022). InSilico Structure Prediction, Energy Calculation and Phylogenetic Analysis of Membrane Protein Large 12 (MMPL12) of Infectious agent of Human Tuberculosis (Mycobacterium Tuberculosis). Journal of Applied Statistics and Machine Learning. 1(1): pp. 14-26

Regardless of the size or complexity of a research study or reporting, there is usually more than one approach for managing the data that is collected to answer the questions posed by the investigators. In this paper, we suggest Smart Information Assimilation & Aggregation System (SIAAS) that follows a layerbased architecture to handle most of the data issues. The architecture and applications of SIAAS in the COVID-19 situation is discussed in detail.

Keywords: Data discrepancies, Data diagnosis, Reliability, Layer based Architecture, Analytical Layer

Roy, S., Agarwal, G.G. & Phanord, D. (2022). Layer based Architecture for Data Management. Journal of Applied Statistics and Machine Learning. 1(1): pp. 27-38

Classes of estimators of population mean which are cost efficient under measurement and nonresponse errors using auxiliary information are presented. These classes of cost efficient estimators have been proposed when both the errors occurs simultaneously as an alternative to the class of estimators proposed for only nonresponse by Singh & Kumar (2010) and Singh & Bhushan (2012). The results of the proposed classes are derived. The proposed estimators are put to test against Singh and Kumar (2010) and Singh and Bhushan (2012) estimators under the cost efficiency criteria. The estimators are compared theoretically and empirically.

Bhushan, S., Kumar, A. & Shukla, S. (2022). Cost Efficient Estimation in Presence of Measurement errors and Non-Response using Auxiliary Variable. Journal of Applied Statistics and Machine Learning. 1(1): pp. 39-59

In this study, the response and impact of COVID-19 is analyzed on a state by state level. Using data from 1/1/2020 through 7/16/2021, every state’s restriction and response during this time period on a daily basis are compiled. Conclusions are drawn regarding the differences in actions taken by Republican and Democratic governors, finding conclusive evidence and numerical specifics on their differences, including that Republican governors, on average, spent less than half as much time as Democrats in stage 3, 4, and 5, lockdowns, the most restrictive ones. Further nuance and predictors are found by comparing states with legislatures controlled by a different party than their governors, showing divided state control to be a moderating factor on governors’ actions for both parties. Statistical significance is found between increased unemployment claims and the most stringent lockdowns for every single state that enacted such a restriction, as well as the vast majority of states in stage 3 and 4.

Franchetti, A., Singh, A.K. & Gallagher, T.M. (2022). Analyzing Political and Economic Variation in United States’ Covid-19 Response. Journal of Applied Statistics and Machine Learning. 1(1): pp. 60-72

This paper uses the case of the Bengal famine of 1943 – 1944 in order to compare poor agricultural data collection methods with the ones that are reliable. This study reviewed the methods used to estimate land area under cultivation and yield per unit area prior to the Bengal famine, and the methods used to arrive at these estimates after the tragic event. A comparison of these methods highlighted the deficiencies of the agricultural data estimation methods used prior to the famine. The ambiguity of guidelines, guesswork, and lack of properly trained field personnel played a central role in the faulty estimation of crop production statistics prior to the famine. The Bengal famine became an important motivator for change. A threestage randomized technique to locate plots where cropcutting experiments were to take place was developed. Additionally, permanent field personnel were recruited and properly trained to do the job. Any reliable system of collecting crop production statistics must rely on well trained field personnel that are educated in the random sampling process. We conclude that this alone cannot produce reliable agricultural production data unless there is close oversight of the field personnel. A perspective discusses the development of other agricultural trends that appear to set the course for future famines in India, given that these trends are emulating the manmade factors contributing to the previous famines.

Keywords: agricultural data, Bengal famine, stratified sampling,sampling bias

Sadri, P. (2022). Lessons from Gathering of Crop Production Statistics in India: A Comparison of Methods used before and after the Bengal Famine of 1943-1944. Journal of Applied Statistics and Machine Learning. 1(1): pp.73-88