As random noise is introduced in the study which would help estimate the responses in the sample, the result is more error will be accounted in the study. In effect, some modifications of the true responses may not estimate statistical values accordingly, which then, as explicated earlier, leads to erroneous development of findings/conclusions in the study. And since we are just estimating the actual responses, the results become less efficient and less reliable. It becomes less efficient because the difference to the true effect of being studied will be greater (greater standard error), and data results become less reliable that the results generated using the random noise will not have a truthful description of the population.

The data will still be useful for the public health purposes. Identifying this white noise would help the researcher control these variables in such a way the true effects of the relationship will be shown. The researcher must understand the relationship of the white noise in the effect being studied and the outcome to truly control the white noise and able to determine the relationships accordingly.

One can still estimate the proportion of the population at risk. For example is for the particular disease that is known to be related to specific behaviors, smoking and lung cancer. Most papers studying the relationship of smoking and lung cancer showed that smokers are really high risk of developing lung cancer. Yet no one can conclude that smokers will definitely develop lung cancer due to the other factors intertwined with it. It implies that there are various factors that can possibly cause lung cancer aside from smoking. Smoking, while being exposed to these other factors, might...

This may lead to serious miscalculations of estimates. Thus, to truly estimate the population at risk, one must identify all possible factors that may cause lung cancer. For some factors that cannot be measured, using white noise might be helpful given that the white noise that will be used can give accurate estimates of the factor to be measured.
If adding random noise to data makes statistical estimates less efficient, the problem can be fixed by studying this random noise and knowing their relationships with the factor being studied and the outcome. By understanding their relationships, one can control the random noise.

One way of controlling random noise is by specifying the group of this factor that may be considered homogenous. For example, in the lung cancer and smoking study, the risk of lung cancer is higher in male than in female, so grouping both males and females to one analysis may skew the results. It is either we stratify the group and have two results accordingly, or collect all male respondents only, or possibly, use an adjusted estimates to estimate the effects of smoking.

If it is known the risk of one group is greater than the other, stratifying per group may give more useful results than one estimate yet not applicable for either. Collecting only all male respondents would give clearer results for the respondents, making the sample exclusively homogenous and will no longer need further stratifications. Lastly, using adjusted estimates would account the effects of both the groups and can be interpreted accordingly -- that is, descriptive of the characteristics of the sample or group being studied.