Regression diagnostics.

This chapter focuses on regression diagnostics. The development of a regression equation is only the first half of a regression analysis. The second, often overlooked part of a regression analysis is to make sure the assumptions underlying the analysis have been met. This is easily accomplished using the regression diagnostic procedures available in SAS® (Statistical Analysis System). Price per flat of strawberries and their availability on the open market were used as an example.

Mean amplitude of intraocular pressure excursions: a new assessment parameter for 24-h pressure fluctuations in glaucoma patients

Background Intraocular pressure (IOP) is important in the pathogenesis of glaucoma and its circadian fluctuations are important in the disease management; however, there are no adequate parameters to describe the fluctuations. This study investigates a new parameter, mean amplitude of intraocular pressure excursion (MAPE), and compares its ability in assessing 24-h IOP fluctuations with other ocular parameters. Methods Only the right eye was evaluated in each of the 79 healthy people and 164 untreated patients with primary open angle glaucoma (POAG). Each participant underwent 24-h IOP monitoring by measuring IOP every 2 h. IOP fluctuations were expressed as MAPE calculations and currently used parameters included mean IOP, standard deviation of IOP, max difference and area under the circadian IOP curve. Comprehensive ophthalmologic examinations were also performed. Associations between visual field deficits and IOP fluctuation parameters were investigated via partial least squares (PLS) regression. Diagnostic performance was evaluated with area under the receiver operating characteristic curves (ROC). Results Compared with healthy volunteers, the MAPE values in POAG patients were higher (4.16 ± 1.90 versus 2.45 ± 0.89, p < 0.01). In PLS regressions where visual field deficits were as dependent variable, MAPE had the highest score regarding variable importance in projection, and its standard regression coefficient was larger than other parameters. Diagnostic performance analysis showed the area under ROC of MAPE for glaucoma detection was 0.822 (0.768–0.868, p < 0.001). Conclusions MAPE might be an effective parameter in clinic to characterise IOP circadian fluctuations.

Performance: Empirical Evidence From Malaysian Banking Industry

Various theories and empirical studies have been applied and proposed to establish and explain how corporate governance practices are related to banks financial performance. This study concerns the relationship between corporate governance variables and bank performance in Malaysia. The data collected and analysed in this research is from quarter one year 2011 to quarter four year 2016. Various determinants have been identified namely return on equity(ROE) for bank performance measurement, CEO duality, board size, and board gender for corporate governance. Control variables are bank size and bank leverage. The methodologies adopted in this research includes descriptive analysis, correlation analysis, Pooled Ordinary Least Square (OLS) regression, Diagnostic Tests (Jarque-Bera Normality Test, Wooldridge Test and Variance Inflation Factor), Breusch-Pagan (BP) Lagrange Multiplier test, and Hausman test. In this study, the findings indicate that strong board composition and bank leverage were experience better performance.

Assessing Student Ability to Interpret Regression Diagnostic Plots

Simple linear regression analysis is an important component of introductory statistics courses, and testing the validity of its assumptions is an important part of regression modeling and analysis. The traditional approach in introductory statistics courses is to use residual plots to assess possible departures from the key assumptions. Confirmatory statistics are typically not used. But do undergraduate students in these introductory courses have the ability to assess residual plots correctly? This paper presents the results of an Institutional Review Board (IRB) approved study to address this question. The results are interesting and have implications for the teaching of regression in introductory courses as well as for textbook authors who cater to these courses.

Transformation model estimation of survival under dependent truncation and independent censoring

Truncation is a mechanism that permits observation of selected subjects from a source population; subjects are excluded if their event times are not contained within subject-specific intervals. Standard survival analysis methods for estimation of the distribution of the event time require quasi-independence of failure and truncation. When quasi-independence does not hold, alternative estimation procedures are required; currently, there is a copula model approach that makes strong modeling assumptions, and a transformation model approach that does not allow for right censoring. We extend the transformation model approach to accommodate right censoring. We propose a regression diagnostic for assessment of model fit. We evaluate the proposed transformation model in simulations and apply it to the National Alzheimer’s Coordinating Centers autopsy cohort study, and an AIDS incubation study. Our methods are publicly available in an R package, tranSurv.

CSF tau and β-amyloid predict cerebral synucleinopathy in autopsied Lewy body disorders

ObjectiveTo test the association of antemortem CSF biomarkers with postmortem pathology in Lewy body disorders (LBD).MethodsPatients with autopsy-confirmed LBD (n = 24) and autopsy-confirmed Alzheimer disease (AD) (n = 23) and cognitively normal (n = 36) controls were studied. In LBD, neuropathologic criteria defined Lewy body α-synuclein (SYN) stages with medium/high AD copathology (SYN + AD = 10) and low/no AD copathology (SYN − AD = 14). Ordinal pathology scores for tau, β-amyloid (Aβ), and SYN pathology were averaged across 7 cortical regions to obtain a global cerebral score for each pathology. CSF total tau (t-tau), phosphorylated tau at threonine181, and Aβ1-42 levels were compared between LBD and control groups and correlated with global cerebral pathology scores in LBD with linear regression. Diagnostic accuracy for postmortem categorization of LBD into SYN + AD vs SYN − AD or neocortical vs brainstem/limbic SYN stage was tested with receiver operating curves.ResultsSYN + AD had higher CSF t-tau (mean difference 27.0 ± 8.6 pg/mL) and lower Aβ1-42 (mean difference −84.0 ± 22.9 g/mL) compared to SYN − AD (p < 0.01, both). Increasing global cerebral tau and plaque scores were associated with higher CSF t-tau (R2 = 0.15–0.16, p < 0.05, both) and lower Aβ1-42 (R2 = 0.43–0.49, p < 0.001, both), while increasing cerebral SYN scores were associated with lower CSF Aβ1-42 (R2 = 0.31, p < 0.001) and higher CSF t-tau/Aβ1-42 ratio (R2 = 0.27, p = 0.01). CSF t-tau/Aβ1-42 ratio had 100% specificity and 90% sensitivity for SYN + AD, and CSF Aβ1-42 had 77% specificity and 82% sensitivity for neocortical SYN stage.ConclusionsHigher antemortem CSF t-tau/Aβ1-42 and lower Aβ1-42 levels are predictive of increasing cerebral AD and SYN pathology. These biomarkers may identify patients with LBD vulnerable to cortical SYN pathology who may benefit from both SYN and AD-targeted disease-modifying therapies.

Influence of Corporate Transparency Disclosures on Financial Performance of Listed Companies in East Africa

With the decline in the financial performance of listed companies in East Africa and the rising trend of corporate failure in both global and local perspective. Stakeholders are increasingly becoming more concerned of the financial performance of their firms. This study aimed to find out whether corporate transparency disclosure can be used to address the decline in financial performance and corporate failures. Therefore, the current study sought to examine the influence of corporate transparency disclosure on financial performance among companies listed in East Africa. Specifically, the study sought to examine the influence of financial transparency, risk transparency, social transparency and governance transparency on financial performance of companies listed in East Africa. The study adopted both descriptive and correlation design. Purposive sampling was used to select the 65 listed companies in Nairobi securities exchange in Kenya, 16 companies quoted in Uganda securities exchange, 7 companies which are quoted in Rwanda Securities Exchange as well as 24 companies listed in Daresalaam securities exchange from 2006 to 2015. Secondary data was collected through the use of document check index retrieved from annual audited financial statements. Regression diagnostic and panel data diagnostic tests were carried out. Results of the study revealed that there was a positive and significant relationship between financial, governance, risk, social transparency and financial performance of listed companies in East Africa.