A direct measurement determines the value of a quantity of interest through a measurement equation that calculates the sought measure straight from the data. Examples include the readings of voltage, current, or resistance by a multimeter, or the value of DC electrical power obtained by multiplying the voltage and current measures. To account for the uncertainty, we must encode the information about the measurement into a probability distribution of the attainable results and model the specific result obtained (which was unpredictable before its observation) as a random sample of the possible results. Therefore, for the measured value to be meaningful, information about the sampling distribution, such as the variance or the covariance matrix if multidimensional, must be provided. Eventually, we must address the problem of propagating the uncertainty from the observed data to the calculated measure.