The numbers are accurate, but the impression is distorted—the blind spot of 'thickness' in election reporting graphs

Did you see the Nikkei Shimbun from 1/29/2026 regarding the House of Representatives election situation survey?

LDP 198 seats, Centrist 167 seats—. The numbers themselves are likely valid, based on the survey. However, the moment I saw that graph, I felt an indescribable sense of discomfort.

“Why is it so thick?”

The bars in the bar graph are unusually thick. And I felt that the visual impact created by that thickness reinforced the impression of a 'landslide victory for the LDP' more than the difference in numbers suggested.

Is this just my imagination? No, in fact, this is a phenomenon that has long been known in visual psychology.

As a certified public psychologist with an interest in information literacy, today I would like to delve a little deeper into this issue of 'thick graphs'.

Our brains react to 'area'

First, let me confirm the basics. A bar graph is essentially a tool for comparing numerical values by 'length'. If the LDP has 198 and the Centrist has 167, the ratio of the lengths of the bars should be 198 to 167, or approximately 1.19 to 1.

However, what happens when you make the bars thicker?

Our brains unconsciously end up comparing them by 'area'. A thick bar has an overwhelmingly stronger presence than a thin bar. Even with the same length, the thicker it is, the more it gives the impression of being 'larger,' 'stronger,' or 'dominant.'

Look at the Nikkei Shimbun graph this time. The bars for each party are drawn with considerable thickness. The thick, dark-colored bar for the LDP and the slightly lighter-colored bar for the Centrist. The difference in numbers is 31 seats, but don't you feel that the visual impression is much wider?

I don't mean to assert that the newspaper is intentionally manipulating impressions. However, I think it is worth knowing the fact that the design of a graph—whether intentional or unconscious—affects the reader's cognition.

Alberto Cairo's warning—the 'Area Trap'

Alberto CairoAlberto Cairoa world authority on data visualization, repeatedly warns about this issue in his book“The Truthful Art: Data, Charts, and Maps for Communication” (Japanese title: 'Techniques for Seeing Through Graph Lies').

What he points out is the fact that when one-dimensional data (in this case, the 'length' of the number of seats) is expressed in two-dimensional or three-dimensional shapes, perception is distorted. For example, if you double the height and width of an icon to show that a certain value has doubled, the area becomes four times larger. The number is doubled, but visually it gives four times the impact.

The same principle applies to bar graphs. The thicker the bars, the more the 'area difference' between a long bar and a short bar expands beyond the 'difference in length.' The difference between 198 and 167 is about 19%, but when drawn with thick bars, that difference feels much larger.

Our brains intuitively grasp the size of shapes before reading the numbers. That is precisely why the 'shape' of a graph leaves a stronger impression than the numbers themselves.

I also want to draw attention to how the 'majority line' is presented

There is one other point that concerned me.

The graph includes three reference lines: the majority line (233 seats), a stable majority (243 seats), and an absolute stable majority (261 seats). This in itself is useful information. It helps readers understand the significance of the seat counts.

However, the way the Liberal Democratic Party's thick bar is approaching the majority line seems visually highly emphasized. The impression that 'it is almost there' is stronger than the actual figure of '35 seats away from a majority'.

To borrow Cairo's words, graphs always 'simplify' something and 'hide' something. What this graph might be hiding is the uncertainty inherent in the figure of '35 seats away.' Even though it is reported that 50% of single-seat constituencies are close races, that tension is not conveyed by the graph. The 'sense of stability' emitted by the thick bar seems to mask the instability of the close races.

How would the impression change if it were a thin line graph?

Let's try a thought experiment here.

If the same data were represented with thinner bars—or perhaps simple lines or dots—what kind of impression would it create?

I suspect that the difference between the LDP and the center would, if not quite a 'close race,' at least leave the impression that 'it's a closer gap than I thought.' The figure of 198 to 167 itself should be conveyed more neutrally.

I am not saying that newspapers are manipulating impressions with malicious intent. Thick bar graphs may also have practical reasons, such as increasing visibility on the page. There may also be consideration for the fact that thin lines are hard to see when viewed on a smartphone.

But if that 'consideration' results in distorting the reader's perception—is that not a problem, regardless of intent?

The pitfall of the 'bandwagon effect'

Let me add one point from a psychological perspective here.

In social psychology, there is a concept called the 'bandwagon effect.' People have a tendency to want to join the dominant side. It is the psychology of 'backing a winner.'

In election reporting, a graph that makes a specific party look like it has 'overwhelming momentum' has the potential to amplify this effect. The psychology of 'if the LDP is going to win anyway, I might as well vote for them.' Or conversely, the resignation that 'if they can't win anyway, there's no point in going to vote.'

If the impression of 'overwhelming dominance' created by a thick graph influences voters' behavior—that is no longer just a matter of design. It is a matter that concerns the very foundation of democracy.

Of course, I don't think newspapers have that intention. But that is precisely why we, as readers, need to be sensitive to how graphs are 'presented'.

For dialogue, not criticism

Finally, I would like to say something to avoid any misunderstanding.

In this article, I do not intend to criticize the Nikkei newspaper. The Nikkei's election reporting clearly states its data collection methods and provides detailed analysis for each single-seat constituency. In terms of information transparency, I actually consider it to be reliable reporting.

What I want to say is that no matter how sincere the reporting is, once it passes through the medium of a graph, some form of 'translation' inevitably occurs. And being conscious of that translation process is important when engaging with information.

Don't just 'look' at a graph; 'read' it. Pay attention to elements like thickness, color, placement, and omitted information. Just doing that will change the quality of the information you receive from the same graph.

Let's take another look at the graphs before going to the polling station

There are about 10 days left until the voting day on February 8th.

From now on, various media outlets will likely release even more situation surveys. When that happens, please try to remember the 'thickness' we talked about today.

'Why is this graph so thick?' 'What kind of impression would it give if the lines were thinner?' 'Where is the gap between the numbers and the impression coming from?'

By asking questions like these, we can move from being 'consumers' of information to 'dialogue partners'.

An election is not a game of predicting the outcome. It is a place where each of us thinks for ourselves and participates with our own will. Let's head to the polling station without being misled by graphs, looking at the reality beyond the numbers.

As a voter myself, I will also verify things with my own eyes before I go to vote.