It is. If you gathered data for a thing and the values were 2, 2, 35, 36, 38, 40, 41, 42, 44, 46, and 48. The mean would be 34. The median would be 40. The mode would be 2. The mode clearly seems like the least representative statistic of the dataset.
Not to start a math tangent but if anyone seriously tried to defend the statistic as being the mode, it would be incredibly weak. Even if true, it would still be very misleading. (And of course I know you’re not doing that.)
If you’re looking at a 2-dimensional statistics graph, you can have an M-shaped curve. Visualise it as two hills to the left and right of the middle. The relevance of this curve to the Politics and Elections forum is that with relatively even partisan politics, opinion polls will have clusters to the left and right of centre, but the median and mean will be close to the centre. For example, suppose a survey of Americans were asked how satisfied they were with Biden’s efforts to reduce gasoline prices, rating his efforts on a scale of 1-7 with 1 being extremely dissatisfied and 7 being extremely satisfied. The most common response might be 2 - dissatisfied, low responses for 4 - neither dissatisfied or satisfied, and the second most common response being 6 - satisfied. You could have the mean as something like 3.7, the median at 3, and the mode at 2. The most relevant conclusion for the survey could be that lots of people were dissatisfied, but looking at the mean and median would show that people were only slightly dissatisfied or somewhat dissatisfied.