Newspapers, websites and other sources of information routinely describe the evolution of the total number of Covid-19 deaths using graphs that are either on a log scale or on a linear scale. We hypothesize that people will react differently to these graphs depending on the scale used, even if the data plotted is the same. The reason is twofold. First, linear scale looks "steeper" than the log scale. Second, even experts have a limited understanding of logarithmic scale (Menge et al 2018). We believe that this has a clear policy relevance. If the graph used affects people behavior and preferences, then the choice of the scale is not neutral. Newspapers and other sources of information must be aware that by choosing a given scale they are likely to generate a certain set of reactions among the general public. To test our main hypotheses, we show to one group the data on Covid-19 deaths in the United States using a log scale and to the other group we show the same data plotted on a linear scale. We ask people questions that capture their understanding of the graph. Moreover, we ask people questions on how worried they are about the crisis and on their policy preferences in relevant domains.