Compare economic data sets without confusing association and cause

Learning goal: Compare two related economic data sets, identify a pattern and explain what the evidence does and does not establish. Check units, time intervals and population boundaries before drawing a conclusion.

Economists use data to investigate production, exchange, work and choices. A data set records observations under specified definitions. Numbers become useful evidence when readers know what was counted, where, during which period and in what units. A table of sales without a time period cannot reliably answer whether daily or monthly activity has changed.

The diagram presents invented records for two workshops in one fictional region. Across four months, Workshop A produces forty, fifty, sixty and seventy repaired bicycles. Workshop B produces thirty, forty-five, sixty and seventy-five. Each count includes completed repairs during that month, with the same counting definition for both workshops. These are not real economic statistics for a named country.

Both workshops increase output across the recorded months. A has higher output in the first two months, both have sixty in the third, and B has higher output in the fourth. The claim “A always repairs more” fails when all four months are considered. The claim “B's increase is greater over this interval” is supported: A increases by thirty repairs from first to last month, while B increases by forty-five.

Be precise about totals and changes. A completes 220 repairs across the four months, and B completes 210. Thus A has the larger four-month total even though B has the larger first-to-last increase and the larger final-month output. Different questions use different comparisons. A responsible report does not choose whichever statistic makes a preferred workshop appear best without naming the question.

A second invented table records paid worker-hours. A uses one hundred hours each month; B uses seventy-five, one hundred, one hundred twenty-five and one hundred fifty. A's output rises while its recorded hours stay fixed. B's output and hours both rise. These relationships raise questions about organization, equipment, repair difficulty and staffing; they do not by themselves identify a single cause.

Comparing repairs per worker-hour can add information, but the measure has limits. In month four, A completes seventy repairs in one hundred hours, or 0.7 repair per hour. B completes seventy-five repairs in one hundred fifty hours, or 0.5 repair per hour. B has more completed repairs in month four, while A has more per recorded worker-hour. If B handled much harder repairs, these rates alone would not establish greater efficiency or quality at A.

A relationship between two variables does not prove that one caused the other. More worker-hours and more repairs can occur together because demand rose, because extra workers were hired or because another condition changed. To investigate causes, ask what other evidence is needed and whether a comparison can distinguish competing explanations. Do not replace an unresolved question with a confident causal story.

Check the visual presentation. Both workshop lines use the same output scale, and the vertical axis starts at zero. A graph with a shortened axis can make modest differences look dramatic if the scale is overlooked. A worker-hours graph would need its own unit label; repairs and hours cannot be placed on one unlabeled scale as though they were the same quantity.

Bias can also come from what a data set includes and how it is organized. A chart that shows only months three and four would hide A's lead in months one and two and its larger four-month total. Starting that chart's vertical axis at fifty-five instead of zero would stretch both rises, sixty to seventy for A and sixty to seventy-five for B, so that one month's growth looks dramatic compared with the workshops' whole output. Ask which records were left out, why the display begins where it does and whether another arrangement of the same numbers supports a different impression.

Data quality matters as much as arithmetic. A workshop might count repairs begun while another counts repairs completed. Missing records can make an apparent decline misleading. Ask how data were collected, whether definitions changed and whether every relevant month is included. A precise calculation cannot repair an inconsistent measurement definition.

A well-supported conclusion identifies the comparison, cites particular values and acknowledges relevant limitations. Separate direct findings from proposed explanations. “B's recorded output grew more between months one and four” is a finding. “New equipment caused that growth” is a hypothesis unless additional evidence about equipment and alternative causes is supplied.

Worked comparison: A's four-month total is forty plus fifty plus sixty plus seventy, or 220. B's is thirty plus forty-five plus sixty plus seventy-five, or 210. A has ten more completed repairs in total. This does not contradict B's larger final-month value because a total and one month's output answer different questions.

Worked qualification: In month four, A's 0.7 repair per recorded worker-hour exceeds B's 0.5. Before claiming A is more productive in every meaningful sense, investigate repair complexity, quality standards and whether worker-hours were recorded consistently.

Data inquiry: Recreate both workshop tables, verify sums and plot the repair records on shared axes. Write three different questions about the data: one about a single month, one about change and one about a total or rate. Answer each with the appropriate calculation and units.

Independent performance: An adult supplies a small, dated public economic table with source definitions, such as output, employment or trade data for two places or periods. Record the source, units, population boundaries and any missing values. Compare two related data sets and write a claim with two numerical pieces of evidence, one limitation and one alternative explanation. Keep the real source record separate from the fictional workshop example.