In my current study on employee engagement, I have r-square of 0.702 whereby, i have examined all the eight notable predictors of it. Will it be considered appropriate to have such high r-square in social sciences?
0.10 is minimum required in social science settings. When there are many predictors in one model it often results in high r-square. It is a matter of concern if you only one or two predictors. In your case i think its not the issue.
First I must admit that I am not an expert in Statistics though I have some proficiency. Your eight variables which are independent variables have a joint or combined impact on the variability of your dependent variable that is employee engagement by 70 percent. In other words the total explanatory power of the regression model is 70.2%. The eight predictors were not able to account for about 30 percent of the variation in employee engagement. Other variables not considered for your study should be the variables, which will account for the unexplained variance in the construct employee engagement. There can be many variables (independent) which will contribute to behaviour of the dependent variable-employee engagement. The eight variables you have chosen are more likely major predictors and they can be appropriately manipulated (assuming that there are no moderators) to enhance employee engagement. If the regression model has major powerful antecedents the r-square will get high. I think your model is a good model if not very good explaining the major dynamics of employee engagement. . A model can have an unmanageable number of factors making a complex research framework. This is not appropriate according to the principle of Parsimony (simplicity in explaining the phenomena or problems that occur, and in the application of solutions to problems) (Sekaran,1992). I think your model is in line with Parsimony. Also if a stepwise regression is performed, it is possible to find out to what extent each independent variable contributes to R square. Sometimes some of the independent variables do not become significant predictors (have no power to influence the dependent variable separately or individually although they can influence jointly).