<p>If an investigator uses survey sampling methods, it might happen that they will not be able to acquire information that is both precise and complete at the same time. Therefore, non-response procedures may occur simultaneously and affect the estimated value. This problem motivated us to suggest a new class of estimators for estimation of population mean under stratified random sampling based on non-response scenario. We suggest a new class of exponential-type estimators that take advantage of the correlation, skewness, standard deviation, and kurtosis that are known about the population parameters of the auxiliary variable. Up to the first order approximation, properties of the suggested estimators are derived, including efficiency and minimum mean square error. We have used an auxiliary variable to estimate the population mean, assuming that in situation-I the non-response is detected only on the study variable and in situation-II, the non-response occur on both the study and the auxiliary variables. The proposed estimators outperform the traditional non-response mean estimators that relies non-response only on the study variable, and non-response happens on both the study and auxiliary variables, as shown by their reduced mean square errors; this is supported by numerical data that verifies the theoretical findings.</p>