DDBA 8307B Week 3 Chart Brief Example

Flat by Calendar Month, Two Peaks by Month of Membership: A Chart Brief for a Fitness Club Chain Whose Steady Average Cancellation Rate Concealed When Members Actually Leave

Student Name

Walden University

DDBA 8307B: Quantitative Business Data Analysis Using SPSS

Instructor Name

Month Day, Year

Flat by Calendar Month, Two Peaks by Month of Membership: A Chart Brief for a Fitness Club Chain Whose Steady Average Cancellation Rate Concealed When Members Actually Leave

The Table That Concealed the Pattern

A composite chain of nine fitness clubs in the Columbus, Ohio, area with about 14,600 members reports a monthly cancellation rate to its owners. For two years the rate has hovered between 3.8 and 4.4 percent of active members a month, with no seasonal pattern worth noting. The owners read the table as a sign of stability. The table is accurate. But it sorts cancellations by calendar month, and members do not join in the same calendar month. A member who joined in January and one who joined in July are at very different points in their membership in any given month, and the calendar-month average mixes them together.

The Chart

I recoded the membership file in SPSS so that each cancellation is assigned to the member's month of membership at the time, month 1 being the first month after joining, and computed the share of members still active at the start of each membership month who canceled during it. The chart plots that rate for the first eighteen months.

Figure 1

Monthly Cancellation Rate by Month of Membership, Members Who Joined January 2023 to June 2025

Rates computed from 11,240 memberships; each point is cancellations in that month divided by members active at its start.

Two sharp peaks appear. In the third month of membership, the cancellation rate jumps to 9.8 percent, more than three times the months around it. In the thirteenth month, it jumps to 11.6 percent. Between the peaks, the rate is low and steady, under 3.5 percent.

What the Peaks Mean

Both peaks match features of the chain's pricing. New members pay a discounted introductory rate for two months, then the regular rate begins in month three. Annual members commit for twelve months, and their commitment ends at the start of month thirteen. The chart suggests that many members leave at precisely the moments their price rises or their commitment lapses. The calendar-month table hid this because each calendar month contains members at every stage, and the peaks of some are averaged with the quiet months of others.

Why a Line Chart

Cleveland and McGill (1984) found that people read position on a shared axis more accurately than they read lengths, angles or areas, which is why a line chart with points on a single axis suits this comparison: the reader sees immediately that two points sit far above the rest. Tufte (2001) urged designers to remove anything on a chart that does not show data; the chart has no gridlines, no legend and no color, because there is only one series. The vertical axis starts at zero so the peaks are not exaggerated.

Limits of the Chart

The chart shows when members cancel, not why. The timing fits the price changes, but members may also leave at month three because the novelty of a new gym fades after eight weeks, and at month thirteen because annual reflection prompts people to review subscriptions. Field (2018) cautions that patterns in observational data invite causal readings they cannot support. A test, such as extending the introductory rate to three months at some clubs and comparing cancellations, would be needed to separate price from novelty.

How the Rates Were Computed

Computing a rate by month of membership required restructuring the data. The membership file holds one row per member with a join date and, where relevant, a cancellation date. In SPSS I computed each member's length of membership in whole months, then built a second file with one row per member per month of membership, flagging the month in which each cancellation occurred. The rate for each month is the number of cancellations in that month divided by the number of members who reached it still active. Members who joined recently and had not yet reached later months were counted only for the months they had completed, so newer members do not distort the later points. Without that step, the thirteenth-month point would rest on fewer members than it appears to, since anyone who joined in the past year could not yet contribute to it.

Checking the Pattern

The two peaks appear in every club and in both years of joiners, which makes chance an unlikely explanation. They are slightly lower at the two clubs that waived the introductory discount, a detail that fits the pricing explanation and is worth a closer look.

The Point for the Owners

The owners' stable average conceals two predictable moments when members are most likely to leave. Retention efforts, such as a check-in call or an offer, could be timed for the end of month two and month twelve, when they are most needed.

References

Cleveland, W. S., & McGill, R. (1984). Graphical perception: Theory, experimentation, and application to the development of graphical methods. Journal of the American Statistical Association, 79(387), 531-554. https://doi.org/10.1080/01621459.1984.10478080

Field, A. (2018). Discovering statistics using IBM SPSS Statistics (5th ed.). SAGE Publications.

Tufte, E. R. (2001). The visual display of quantitative information (2nd ed.). Graphics Press.