Drawing Basic Insights and Communicating Findings
What this domain tests
Items ask you to convert analysis into a clear, defensible summary: writing an insight statement, choosing the right chart for the question, and reporting change and uncertainty honestly.
Summaries in text
- Structure each insight as finding, evidence, implication: "Checkout abandonment rose from 62% to 71% (+9 percentage points) between Q1 and Q2, concentrated in mobile users."
- Always state n, the time window and the unit: "median order value, USD", not "average value".
- Use rates or per-capita figures when group sizes differ; raw counts mislead across unequal populations.
- Percent change = (new - old) / old x 100. Percentage points describe the gap between two percentages: 20% to 25% is +5 points but +25%.
- Round to the precision the data supports; long decimal tails imply false accuracy.
Chart choice
- Bar: compare categories; sort by value unless there is a natural order.
- Line: change over time, time on the x-axis and evenly spaced.
- Histogram: distribution of one numeric variable - state the bin count.
- Scatter: relationship between two numeric variables.
- Box plot: spread and outliers by group. Label both axes with units, give the chart a title that states the takeaway, and keep bar charts starting at zero.
Worked example
Monthly sales: Jan 120, Feb 150, Mar 135. Jan to Feb is (150-120)/120 = +25%; Feb to Mar is (135-150)/150 = -10%. A line chart fits because the categories are ordered time; a bar chart also works but the line emphasises the trend. "Sales fell 10% from February" is correct; "sales fell 15" is not, because the unit is missing.
Traps
- Confusing percent with percentage points, or dropping the baseline.
- Line charts over unordered categories; pie charts with many slices.
- Truncated y-axes or 3-D effects that exaggerate small differences.
- Claiming cause from a correlation or a single before/after comparison.
- Reporting a total without the denominator.