Effect Of Corporate Social Responsibility On Profitability Of Listed Deposit Money Banks In Nigeria

3.4    Variables Definition and Measurement
NPM = Net profit margin as a proxy of profitability is measured by net profit divided by sales (interest income) as used by Maisaje (2015).
ROTA = Return on total assets as a proxy of profitability is measured by net profit before interest and after taxes divided by total assets (Joliet & Muller, 2013; Al-Taani, 2013; Visic, 2013; Mitani, 2014; and Maisaje, 2015).
ROE = Return on equity as a proxy of profitability is measured by net profit after taxes divided by equity (Pouraghajan et al., 2012; Visic, 2013; and Nirajini & Priya, 2013).
CSR = Corporate Social Responsibility, measured by CSR Expenditure divided by total asset (Bessong & Tapang, 2012; Bolanle, Olanrewaju & Muyideen, 2012; Folajin Ibitoye & Dunsin, 2014; Iya, Magaji & Bawuro, 2015; Odetayo, Adeyemi & Sajuyigbe, 2014).
α = Constant term
β1-4 = Beta parameters to be estimated
SIZE = firm size measured by natural logarithm of total assets (Smith et al., 2012 and Soumadi & Hayajneh, 2012).
LEV = Financial leverage measured by debt to equity (Firer & Williams, 2003; Shiu, 2006 and Chan, Watson, & Woodliff, 2014).
INTR = Interest rate as a measure of the general business environment (Chatta, 2016). ε = error term
i = participating firms (i = 14 firms) t = time variable (t = 10 years)
3.5    Methods of Data Collection
This study uses secondary data sourced from the financial statements of the sampled banks. Data on CSR expenditure and profitability were obtained from the annual reports of the sampled deposit money banks while data on interest rate was taken from the CBN Statistical Bulletins (2006-2015).
3.6    Data Analysis Techniques
The statistical techniques that would be employed to analyze data collected for this study are descriptive statistics and inferential statistics. The descriptive statistics consist of mean, standard deviation, minimum and maximum. Descriptive statistics would be used in this study because they help to describe the basic features of the data in a study. They provide simple summaries about the population and sample. The descriptive statistics used in the study are mean, standard deviation, minimum and maximum. Descriptive statistics do not, however, allow conclusions beyond the data analyzed or reach conclusions regarding any hypotheses the study made. They are simply a way to describe data.
The inferential statistics include correlation and regression analyses. Multiple regression analysis was used to achieve the study objectives and to test their hypotheses. A multiple regression analysis is a statistical tool for the investigation of relationships between variables. Usually, the investigator seeks to ascertain the causal effect of one variable upon another. Multiple OLS regression models are used in this study because they easily measure the degree of confidence that the true relationship is close to the estimated relationship. The F-statistic in the analysis of variance (ANOVA) which shows the overall model effect was used to test for significance for the overall model fitness. Also, the p-values in the regression coefficients were used to test for significance  of individual  variables  in the model. The decision criteria in both cases are that if the calculated F-sig value is greater than 0.05 (in most of the cases, significance level is 5%), then reject the null hypothesis  for the model; however, if the calculated F-sig value is less than or equal to 0.05, then accept the null hypothesis and conclude that there is no significant  difference between the dependent and independent variables in the model.
3.7    Diagnostic/Post Estimation Tests
Diagnostic/post estimation tests such as serial (auto) correlation, normality, heteroskedasticity, stationarity, Hausman specification (fixed effect and random effect), effect size, granger causality and multicollinearity were carried out. These tests were carried out to establish the quality of the data.