Which statistical analysis is NOT possible on an interval scale?

On an interval scale, the most notable characteristic is that it allows for the measurement of differences between values as well as the computation of averages. For instance, in an interval scale such as temperature measured in Celsius, it is meaningful to say that 30 degrees is warmer than 20 degrees, and you can calculate the average temperature.

However, one limitation of interval scales is that they do not support the calculation of ratios. This is because there is no true zero point in an interval scale; for example, in the Celsius temperature scale, 0 degrees does not signify the absence of temperature—it’s simply a point on the scale. Therefore, statements such as "30 degrees is twice as hot as 15 degrees" do not hold true, as this type of ratio comparison assumes a true zero exists.

Consequently, while interval scales allow for averages and comparisons of magnitude, the inability to calculate meaningful ratios clearly differentiates this statistical analysis from the other types emphasized in the question.