Case Study 2 — Three Charts, Three Stories: How the Same Inflation Data Tells Different Tales

In 2022 and 2023, U.S. inflation was the dominant economic story. The CPI rose to 9.1% year-over-year in June 2022 — the highest reading since the early 1980s. Politicians, journalists, central bankers, and ordinary households were all worried about it. By mid-2024, inflation had fallen back to roughly the Federal Reserve's 2% target, and the conversation shifted.

But during those two years, the charts used to depict inflation in news media varied dramatically. Different framings, different time windows, different y-axis ranges produced wildly different impressions — sometimes from the same underlying data on the same day. This case study walks through three of those framings to illustrate exactly how charts can shape (and sometimes deceive) the reader, even when the data is honest.

The data we will use is the year-over-year percent change in the CPI for All Urban Consumers (CPIAUCSL). It is real, public, and sourced from FRED. You can reproduce every chart in this case study yourself in about five minutes.

Chart 1 — The "Inflation Is Crushing Us" Chart

Consider this version of the chart: y-axis ranges from 7% to 10%; x-axis spans March 2022 to August 2022 (six months); the line is bright red with a sharp upward slope, peaking at 9.1% in June. Title: "Inflation Surges to 40-Year High."

What this chart shows: - The peak of the inflation episode (June 2022) clearly marked - Six months of data, all at very high inflation rates - A y-axis that compresses the visual into a narrow band, making each tick look enormous - A title that emphasizes the historical comparison ("40-year high")

What this chart leaves out: - The pre-2022 inflation history, which would show that 9.1% was indeed unusual but that 1970s inflation was worse - The post-June 2022 trajectory, which (as we now know) showed inflation falling rapidly - The distinction between headline and core inflation (this is headline; core was lower) - The breakdown across categories (some prices were rising fast, others were not)

This chart was widely used in mid-2022 by news outlets that wanted to convey alarm. It is not lying — every number is correct, the y-axis truncation is mathematically defensible. But it tells one specific story: inflation is high, getting higher, and historically extreme. By selecting the time window narrowly, the chart deprives the reader of the context that would put the alarm in perspective.

Chart 2 — The "Inflation Is Coming Down" Chart

Now consider an alternative version of the same data, used in mid-2023: y-axis ranges from 0% to 10%; x-axis spans January 2022 to June 2023 (18 months); the line is bright green for the second half (declining); title: "Inflation Has Fallen Sharply From Its Peak."

What this chart shows: - The peak of the inflation episode (June 2022) at 9.1% - The decline through 2023, ending around 3% in June 2023 - A y-axis that starts at zero, giving honest visual weight to the changes - A title that emphasizes the recent decline

What this chart leaves out: - The longer historical context (2% was actually still above the Fed's target, and core inflation was still above 4%) - The fact that month-over-month inflation was still positive (prices were rising slower, but not falling) - The wage and labor market numbers that the Fed was using to assess underlying inflation pressure

This chart was used in mid-2023 by outlets that wanted to highlight the success of monetary policy. It is also honest. Every number is correct. The y-axis is more honest than Chart 1's. The title is defensible. But it tells a different story: inflation peaked, then fell — the system is working.

The same data. Different framings. Different visual stories.

Chart 3 — The "Long-Run Perspective" Chart

Now consider a third version, using a much longer time window: y-axis 0% to 16%; x-axis January 1948 to December 2023 (76 years); the line shows the year-over-year inflation rate over the entire postwar period.

What this chart shows: - Multiple inflation episodes: the Korean War spike, the Great Inflation of 1965–1982, the 2008 commodity spike, the 2020 deflation scare, and the 2021–2023 surge - The Volcker disinflation of 1981–1984, when inflation fell from 13.5% to about 4% - The Great Moderation period of 1985–2007, when inflation was unusually stable - The 2021–2023 surge, which was the largest since the early 1980s but smaller than the 1970s peaks - The post-2023 fall back toward the Fed's target

What this chart shows that neither of the others did: the 2021–2023 inflation surge was significant, but it was not unprecedented in U.S. history. The 1970s and early 1980s were worse. The Volcker disinflation showed that determined monetary policy could break inflation, though at the cost of a deep recession. The 2022 episode happened to end with a much milder recession (or no recession at all) — a result that was widely considered impossible by most forecasters.

This chart is the most informative of the three, and it is also the least often shown in news media — because the 76-year history doesn't fit into a tweet, doesn't drive immediate emotional reaction, and doesn't support a clean narrative on either side.

What this case study shows

All three charts use the same underlying data. All three are honest in the sense that no numbers are fabricated. All three could be cited in a news article without anyone being able to say "you lied." And all three tell different stories.

The differences come from three choices: the time window, the y-axis range, and the framing of the title. Each of these choices is defensible in some contexts and misleading in others. None is "neutral." Every chart involves choices, and the choices shape the reader's interpretation.

The lesson is not that chart-makers are dishonest. The lesson is that every chart is an argument — even when the data behind it is true. The skill of reading charts critically is the skill of asking, every time you see a chart: what choices did the chart-maker make? What story is the chart trying to tell? What story would the data tell if the choices were different?

A practical exercise

Open FRED right now. Find the CPIAUCSL series. Toggle it to "Percent Change from Year Ago" using the chart's edit controls. Now make the three charts described above:

  1. Chart 1 — Set the time window to March 2022 through August 2022. Set the y-axis to range from 7% to 10%. Save or screenshot.
  2. Chart 2 — Set the time window to January 2022 through June 2023. Set the y-axis to range from 0% to 10%. Save or screenshot.
  3. Chart 3 — Set the time window to January 1948 through December 2023 (or whatever the latest data is). Use the default y-axis. Save or screenshot.

Now look at all three charts side by side. Notice how different the same data looks in different framings. This is why being literate about chart construction matters: even honest data can be presented in ways that lead the reader to different conclusions.

Doing this exercise once will change how you read economic charts forever. Do it.

Discussion questions

  1. Which of the three charts is "most honest"? Is there even a single answer to this question?
  2. The narrow y-axis of Chart 1 is sometimes called "deceptive truncation." But sometimes truncating the y-axis is necessary to show meaningful variation. When is truncation justified, and when is it deceptive?
  3. Chart 3 (the long-run view) is more informative than the other two, but it is also the least likely to appear in a news article. Why? What are the structural pressures on news media that favor Charts 1 and 2 over Chart 3?
  4. Suppose you were assigned to write a news article about inflation in mid-2022. Which chart would you use? Would your answer change depending on whether your assignment was "explain why people are worried" vs. "explain why the Fed is acting" vs. "give historical context"?
  5. Find a recent economic chart in a news article and try to identify what choices the chart-maker made. Was the time window selected? Was the y-axis truncated? Was a particular measure chosen over alternatives? Could the same data be framed differently to tell a different story?