Research
Email: pmsdg@unilorin.edu.ng
OBAFEMI, Kayode Ezecheal
07/25OY009
RELATIVE EFFECTIVENESS OF THINK-PAIR-SHARE AND JIGSAW-IV STRATEGIES ON PRIMARY SCHOOL PUPILS’ ACHIEVEMENT IN MATHEMATICS IN ILORIN, KWARA STATE
March 2023
The use of teaching strategies that promote active participation of pupils in teaching and learning activities has been suggested by researchers and examples of such strategies are Think-pair-share and Jigsaw-IV strategies. The poor performance of primary school pupils in Mathematics has been attributed to the use of the teacher-centered method of teaching. The study, therefore, investigated relative effectiveness of Think-pair-share and Jigsaw-IV strategies on primary school pupils’ achievement in Mathematics in Ilorin, Kwara State. The objectives of the study were to: (i) establish pattern of pupils’ academic achievement in Mathematics; (ii) examine effect of treatment on final year primary school pupils’ achievement in Mathematics; (iii) investigate interaction effect of treatment and gender on final year primary school pupils’ achievement in Mathematics; (iv) determine interaction effect of treatment and school type on final year primary school pupils’ achievement in Mathematics; (v) explore interaction effect of treatment, gender and school type on final year primary school pupils’ achievement in Mathematics.
This study adopted a 3X2X2 pretest, posttest, non-randomized control group. The population of the study consisted of 41,486 primary school pupils in Ilorin West Local Government Area of Kwara State, while the target population was 4,257 final year pupils in public and private primary schools. Purposive sampling technique was used to select 164 pupils in six intact classes. The treatment groups were exposed to treatments packages on four topics in Mathematics. The control and treatment groups were subjected to pretest and posttest using validated Mathematics Achievement Test (MAT) with a reliability coefficient of 0.74. The data generated were analyzed using percentage, mean and Analysis of Covariance (ANCOVA) at 0.05 level of significance.
The findings of the study were that:
The study concluded that though Think-pair-share and Jigsaw-IV strategies improved final year primary school pupils’ achievement in Mathematics, Jigsaw-IV was a better strategy. This implies that primary school pupils performed better with the use of Jigsaw-IV and Think-pair-share strategies than conventional method. The study, therefore, recommended that Primary school teachers should use Jigsaw-IV and Think-pair-share strategies for the teaching of Mathematics to the final year pupils.
DEVELOPMENT AND EVALUATION OF A GLOCALISED TECHNOLOGY-ENABLED LEARNING APPLICATION FOR TEACHING PHYSICS COMPONENT OF NIGERIAN BASIC SCIENCE CURRICULUM
KOLEDAFE, Olawale Sunday
10/25PC031
Glocalisation is the process of presenting global content in the local context. Sciences
concepts can be glocalised to enhance students’ understanding of instructional context.
Science is universal, and glocalised teaching in the local language (L1) is a right to learning.
This study deals with the development and evaluation of a glocalised learning app (G-TELA) for teaching physics components of the basic science curriculum. The objectives were to (i) develop a glocalised learning app (G-TELA) on selected physics components of the Basic Science and Technology (BST) curriculum; (ii) assess the usability of the developed G-TELA by students; (iii) examine G-TELA effectiveness on students’ academic performance; (iv) investigate students’ attitude to L1 and use of G-TELA; (v) determine BST teachers’ attitudes of towards the use of G-TELA; and (vi) analyse the effect of gender on the performance, attitude to L1 Medium of Instruction (MOI) and use of G-TELA in learning. The study adopted design-based research that combines qualitative and quantitative research methods. Standardised test items with a reliability value of 0.82, open-ended, and close-ended questionnaires with a reliability of 0.71 were used for quantitative data collection. The sample was 70 participants in the experimental or control groups or as experts’ validators. The system usability scale, mean score, standard deviation, independent-sample t-test, and paired-sample t-test were used to analyse the quantitative data at a 0.05 level of significance. Thematic analysis was used to analyse the qualitative aspect of the research.
The findings of the study were that:
i. glocalised learning app (G-TELA) was successfully developed for the teaching of
physics components of BST;
ii. glocalised instructional content is a multi-disciplinary endeavour requiring several
experts. The Culture-Based Model (CBM) and Analysis Design, Development,
Implementation and Evaluation (ADDIE) synthesis guided development;
iii. participants rated the usability of the G-TELA high reflected in an overall rating score
of sus = 88.70 from 100;
iv. there was a significant difference between participants exposed to the application and those taught using the conventional method, t(51) = 8.097; p = 0.00;
v. participants had a positive attitude towards the L1 medium of instruction and the G-
TELA for learning with a grand mean x̄ = 3.0 out of 4.0;
vi. the BST teachers had a positive feeling towards using G-TELA as a resource for
teaching BST in schools; and
vii. no interaction gender effect was established on academic performance and attitude
toward G-TELA. However, gender effects existed in their attitude to the use of the L1
medium of instruction, t(26) = 1.265; p = 0.22; and t(31) = 3.45; p = 0.02.
The study concluded that G-TELA was successfully developed to enhance students’
performance in physics concepts. These results established the usability and instructional impact of G-TELA teaching science and technology in Nigeria. The implication is that students will learn and perform better when instructional content is presented using a technology-enabled L1 medium of instruction. The study recommends that educational institutions and curriculum developers should encourage and promote the use of glocalised packages for teaching sciences curricula in Nigeria.
Name: ADEBOLA, Christiana Oluwatoyin
Title: PERCEPTION, PREVALENCE AND CONSEQUENCES OF DRUG ABUSE
AMONG STUDENTS OF TERTIARY INSTITUTIONS IN KWARA STATE, NIGERIA
Year: February 2022
Drug abuse is the harmful use of mind-altering drugs, illegal drugs and legally prescribed drugs among students of tertiary institutions. The harmful use can be a result of poor or negative perception of students to drug as an instrument that aid boldness, learning and assimilation. This negative perception had led to high prevalence of drug abuse yearly and in turn led students to early grave, deteriorated health, rustication and poor academic performance. This study therefore examined the perception, prevalence and consequences of drug abuse among students of tertiary institutions in Kwara State. The objectives of the study were to examine: (i) the perception of drug abuse; (ii) the difference in the perception of drug abuse based on age, gender, religion and class level; (iii) the prevalence of drug abuse; (iv)the difference in the prevalence of drug abuse based on age, gender, religion and class level; (v) if mental health problems, physical health problem, social health problems and poor academic performance are consequences of drug abuse; and(vi) the difference in the consequences of drug abuse based on gender, age, religion and class level among students of tertiary institutions in Kwara State.
The descriptive research design of survey type was used for the study. The population of the study comprised all 121,505 students of all tertiary institutions in Kwara State. A multi-stage sampling procedure was used to select 790 respondents. A researcher-developed questionnaire was used for the study. The reliability of the instrument was established using a test re-test method and a coefficient of 0.87 was obtained. Data collected were subjected to inferential statistics of t-test, ANOVA and Chi-square at 0.05 level of significance.
The findings of the study were that:
This study concluded that age and class level affected wrong perception of drugs, religion do not influence the prevalent drug abuse among students of tertiary institutions in Kwara State. The implication is that irrespective of perception of drugs, the prevalence and consequences of drug abuse were noticeable among students in tertiary institutions. The study recommended health education to influence students’ perception of drug abuse and health seminars to reduce the prevalence and consequences of drug abuse among students of tertiary institutions in Kwara State.
A modified procedure for detecting Heteroscedasticity in the presence of Outliers
BOLAKALE ABDUL-HAMEED
08/55EG036
May 2023
The presence of contamination often called outlier is a very common attribute in data. Among other causes, outliers in a homoscedastic model make the model heteroscedastic. Moreover, outliers distort diagnostic tools for heteroscedasticity such that it may not be correctly identified. Detection of heteroscedasticity in the presence of outliers presents a significant challenge for most of the existing methods. Therefore, this study aimed at a modified procedure for detecting heteroscedasticity in the presence of outliers. The objectives of the study were to:(i) investigate the performance of some selected existing procedures for the detection of heteroscedasticity; (ii) propose a modified procedure for detecting heteroscedasticity in the presence of outliers; (iii) investigate the properties of the proposed modified procedure; and (iv) compare the performance of the proposed method with some selected existing methods in the detection of heteroscedasticity in the presence of outliers.
The modified test was obtained by substituting non-robust components in the Breusch-Pagan test with robust procedures which makes the modified Breusch-Pagan test to be unaffected by outliers. Monte Carlo simulations and real data sets were used to investigate the performances of the newly proposed test and the selected existing tests at varying percentages (5%, 10%, 20% and 30%) of outliers and sample sizes (20, 40, 60 and 100). The performances were assessed using the probability value (p–value) and the power of all the methods considered.
The findings of the study were that:
The study concluded that outliers have a significant and detrimental impact on the efficiency of test procedures for detecting heteroscedasticity in linear regression. The existing procedures are therefore not efficient at detecting heteroscedasticity as the modified procedure mBP, in the presence of moderate or high percentage of outliers. The study therefore recommended that the proposed mBP test be used for detecting heteroscedasticity in linear regression diagnosis when the percentage of outliers in data sets is either known or unknown.
IMPUTATION MECHANISM FOR LONGITUDINAL STUDY WITH MISSING RESPONSE AND COVARIATES
KOLE, EMMANUEL
MATRIC NO.: 05/68EP011
FEBRUARY, 2023
Missing values are intended measurements not taken in data collection and imputation is the process of estimating or predicting such missing values either in response variable or the covariates. Missing values are common in longitudinal study where measurements on individuals are repeatedly taken over time. The existing imputation methods of solving this problem do not focus on longitudinal study; therefore this study aimed to develop appropriate procedure for imputing missing values in longitudinal study for the response variable and also examine such for the covariates. The objectives of the study were to: (i) determine when a missing data set from a longitudinal study should be imputed (ii) develop appropriate model for imputing missing values for the response and predictor variables (iii) examine imputation strategy in a longitudinal study for the response variable and the covariates, and (iv) validate the procedure by combining the response variable and the covariates using real life data.
The proposed procedure of imputation for the response is determined by the relationship that the response variable has with the independent variable(s) from the subjects for which the data are complete. The imputation for the predictor is done from the probability density function or probability mass function that the independent variable is judged to have. The proposed procedure is compared with existing imputation methods at various percentages of missing values and different levels of correlations among the predictor variables. The histogram and stem diagram were used to show normality.
The findings of the study were that:
The study concluded that the proposed imputation procedure is plausible for both continuous and categorical variables when the percentage of missing value is not large. The full conditional specification (FCS) was equally used to validate the procedure by combining the response variable and the covariate and to examine the goodness of the imputation methodology. It was found that the relative efficiency is at least ninety percent which means that values missing could be imputed with high certainty. The study therefore recommended the proposed imputation mechanism should be applied to scenarios with missing response and covariates for optimal result.
BAYESIAN CHANGE-POINT TECHNIQUES FOR GAUSSIAN BASED MODELS
ADEGOKE, Taiwo Mobolaji
In statistics, a change point is a time at which a change occurs in a set of observations which include a change in the mean, variance, correlation, scale and location parameters of a distribution. However, existing studies in Bayesian change-point analysis mostly focused on the use of non-informative prior for the location and conjugate prior for the scale parameters, especially under the assumption of unequal variances across sub-samples. This study, therefore, investigated Bayesian change-point techniques for Gaussian based models. The objectives of the study were to: (i) develop Bayesian change-point models for the means and variances of multivariate Gaussian variables under equal and unequal variances of different sub-samples; (ii) determine the performances of the developed change-point models for both the single and multiple change points; and (iii) validate the performances of the developed models on real-life data.
Consider a series of observation drawn from a multivariate Gaussian distribution
for which possible change(s) occurring at the points
. Thus, the new sample data become
with
for a single change point. Informative priors were adopted for the mean and variance structure of the models to elicit the prosteror estimates. Bayesian hypothesis testing techniques was employed to ascertain the points of changes under different prior distributions considered in Monte-Carlo study using Gibbs sampling.
The findings of the study were that:
The study concluded that the Bayesian Change point models based on informative priors for both the single and multiple change points situations are relatively more efficient than those based on non-informative priors under the equall and unequal varainces. The study therefore recommended that whenever it is desirable to detect the points at which changes occure in any manufacturing or service processes, the proposed Bayesain change point models based on informative priors should be employed for optimal results.
MODIFIED GENERALIZED POISSON REGRESSION-BASED MODELS FOR TO COUNT DATA WITH HETEROGENEOUS DISPERSIONS
OYEKUNLE, Janet Olufunmike
The conventional model for count data is the Poisson regression model with a single parameter under the assumption that both the variance and the mean are the same (equi-dispersion). In many real-life cases, the above assumption is violated in which case; the variance may be greater or less than the mean. However, several models have been proposed to handle each of these dispersion cases but model that accounts for a combination of any of the cases has not received much attention in the literature. Therefore, this study aimed to develop generalized Poisson based-models that would account for the heterogeneous dispersion in count data. The objectives of the study were to: (i) develop modified Generalized Poisson Regression (mGPR) and modified Poisson Quasi-Lindley Regression (mPQLR) for modeling count data with heterogeneous dispersions; (ii) compare the performances of the proposed models with the classical GPR and PQLR models in Monte-Carlo studies; and (iii) validate the results of the Monte-Carlo studies on real life datasets.
The classical GPR and PQLR models were modified by incorporating additional dispersion parameter into their original formulations. The parameters of both the new mGPR and mPQLR models were estimated by method of maximum likelihood. Count data sets were simulated at various sample sizes n = 200, 300, 500 and 1000 each at different ratios of dispersion combinations of 1:1, 1:3, 2:3, 3:1 and 3:2. The performances of the various models were examined using Akaike Information Criteria (AIC), Bayesian Information Criteria (BIC), loglikelihood (-2Loglik), and Deviance.
The findings of the study were that:
The study concluded that, the proposed models were more efficient than the existing models for modeling data with heterogeneous dispersions. The study therefore recommended that the proposed mGPR and mPQLR models should be adopted for modeling count data that demonstrated evidence of the presence of more than one dispersion structures.
© 2025 University of Ilorin