Folks conversant in the engineering aspects of radio frequency are typically ignorant or dismissive in the nonlinear effects of such radiation in biological life that deals in energy flows taking cues from the environment, which is why the conversation begins and ends with wattage/power exposure.

There's a biophysicist named Andrew Marino who explains the plight of deriving evidence of causation, or lack thereof, from apparently mixed results in studies focused on the biological effects of non ionizing radiation (and hence why the safety standards are set the way they are):

“A common assumption is that the numerical value of the measured effect is the requisite reproducible observation. If a putative effect is +50 units, “reproducibility” is taken to mean that +50 units must be observed when the putative cause is reproduced or, allowing for the apparent stochastic variability exhibited by living systems, at least something close to +50 units. Under this assumption, an observation of 0 units or −50 units counts as evidence against a causal link, and observation of −50, 0, and +50 in three independent trials would be interpreted as strong evidence against the reality of the effect, based on averaging. But adoption of this assumption in a one-size-fits-all manner emasculates our ability to understand nonlinear biological phenomena, for example those caused by EMF’s. To see this, suppose that +50 and −50 were each observed five times in ten independent trials, an entirely permissible result in a fully deterministic system governed by nonlinear laws (nonlinear differential equations). Under the common assumption, a strong inference against the existence of an effect would arise because the data averaged to zero. Nevertheless the results were reproducible in the sense that they were consistently non-zero.

Suppose the analytical method used to evaluate the data obtained in the ten trials returned a positive value each time the putative cause was applied, regardless of whether the measurement yielded +50 or −50. In this case the cause-effect relationship would be captured by the analysis and the scientific requirement of reproducibility would be satisfied. The biological effects of environmental EMFs must be viewed from this perspective of nonlinearity.

Otherwise the chronic error of the power and cell-phone companies—the false negative result—will continue to occur, with all the harm that the error entails. The nonlinear perspective would permit rationalization of causality based on the frequency of the kind of observation made (consistently non-zero), not on the arbitrary (and certainly wrong) assumption that a non-zero average of the magnitude of the observation is a necessary property of reproducibility.”