In an observational study, to address the concern of unmeasured confounding, instrumental variable (IV) analysis is often used when an IV can be found. When unmeasured confounding is thought to be possible but not severe, regression is often used as the primary and IV as the secondary analysis. Confirmation of the qualitative findings of the regression analysis provided by the secondary IV analysis is taken as confirmatory evidence for the findings of the regression. However, these two analyses are correlated and it is unclear how much independent evidence is provided by the IV analysis. We develop a statistically sound method of analysis to resolve this redundancy. The method is based on a new estimator, EX estimator, which extracts the part of the regression estimator uncorrelated to the IV estimator. The EX and the IV estimators form evidence factors - they provide independent evidence to test a hypothesis about the causal effect. The estimation method is extended to a general model of treatment effects that allows for heterogeneous and nonadditive effect. Using this method we study the effect of exposure to violent conflict on preferences for altruistic behavior, time and risk.