Bimodal Distribution vs Uniform Distribution

Bimodal describes the shape of a graph: two prominent high regions with a real dip between them. Prominent is the operative word, because real data are bumpy and nearly every histogram carries small wiggles that are not peaks. The Fall 2026 course fixes the vocabulary in topic 1.6, where one prominent peak is unimodal, two are bimodal, and frequencies that are all about the same with no prominent peak are approximately uniform.

Twelve quiz scores out of 10: 3, 3, 4, 4, 4, 5, 8, 9, 9, 9, 10, 10. A dotplot shows one clump in the low scores and another in the high ones with nothing between. The mean is $78/12 = 6.5$ and the median is $(5 + 8)/2 = 6.5$ as well, so both land in the empty middle. No student scored 6, 6.5, or 7. A single center is the one number these data most clearly reject.

The peaks do not have to be the same height. Students see a histogram with one tall hump and one shorter one, decide "there is only one mode," and call it unimodal. A second peak qualifies when it is a clear local high separated from the first by a genuine dip, whatever its height. Requiring equal heights would make bimodality almost unobservable, since two peaks in real data are never exactly level.

Bimodality is a judgment made from a graph, and the graph can be tuned. Too many narrow bins turn ordinary sampling noise into a row of spikes; too few merge two real groups into one hump. Look at the shape at more than one bin width before committing to the word. A boxplot is no help at all here, since it is drawn from five numbers that carry nothing about peaks, so a bimodal set and a single-peaked set can produce identical boxplots.

When two peaks do show up, ask which two groups got mixed together, then describe them separately rather than averaging them into a population that does not exist.

Uniform means probability is spread evenly, and the word covers two different objects. A discrete uniform distribution has $k$ listable outcomes, each carrying probability $1/k$. A continuous uniform distribution on the interval from $a$ to $b$ has a flat density of height $\frac{1}{b-a}$, so probability is area and every subinterval of the same width carries the same probability. Data can also be called approximately uniform, which is the Fall 2026 course's shape word in topic 1.6 for a graph whose frequencies are all about the same with no prominent peak.

One roll of a fair six-sided die is discrete uniform with $k = 6$, so each face has probability $1/6 \approx 0.167$ and $P(X > 4) = 2/6 \approx 0.333$. A continuous uniform on 0 to 10 has density $1/10 = 0.1$, so $P(3 < X < 7) = 4 \times 0.1 = 0.4$: four units of width out of ten. Both sit centered at their midpoint, 3.5 and 5.

"Every value is equally likely, so $P(X = 3) = 0.1$" is the standard error in the continuous case. The height of the density curve at 3 is 0.1, but the probability of landing exactly on 3 is 0, because a single point has no width and probability here is area. That is why $P(3 < X < 7)$ and $P(3 \le X \le 7)$ are both 0.4 for the continuous uniform, while for the die $P(X > 4) \approx 0.333$ and $P(X \ge 4) = 0.5$ are different numbers.

Flat also does not mean the values sit close together. The continuous uniform on 0 to 10 has standard deviation $10/\sqrt{12} \approx 2.89$, which is real spread. Uniform says every region is equally likely, not that the variable barely varies.

One caution about reading uniformity off a picture. Flatness is a claim about equal-width bins, so unequal bins make an evenly spread variable look bumpy and can hide a genuinely uniform shape. Check the bin widths before you use the word.