Right-align numeric columns and left-align text columns

What it is

Numbers in one column need their magnitudes stacked so they can be compared; words in one column need a shared starting edge so they can be read. Numeric right alignment puts ones, tens, and decimal points on one vertical line, so difference becomes position rather than speech. Text left alignment puts first letters on a rail the scan can ride down. Left-aligning prices or right-aligning names slows both jobs, because the eye loses the edge it was counting on.

Why it happens

Comparing numbers is a place-value operation. Right alignment (with tabular or monospaced figures) stacks the same digit on the same x, so a number with an extra digit sticks out left and magnitude is visible in the silhouette. Left-aligned numbers scatter the ones place, and people have to finish reading each value before they can compare; the preattentive “which row is larger” disappears. Text runs the other way. In left-to-right languages the left edge is the landing rail; a ragged right edge is just word length. Right-aligned text makes every first letter jump, and the scan cannot follow one edge. Alignment is not decorative symmetry. It is two perceptual tasks — place-value versus reading — each needing its own edge. Currency signs, percent marks, and minus signs have to join the place-value stack, or prefixes will knock the ones column off the line.

Studying it

Show the same numeric column left-aligned, right-aligned, and decimal-aligned, and run tasks such as “find the three largest” or “which of these two rows is larger.” For text columns, compare left versus right alignment on a target-word search. Independent variables: alignment, tabular figures or not, presence of currency and decimals. Dependent variables: comparison time, order-of-magnitude errors, landing stability during search. On eye tracking, a right-aligned numeric column should draw fixations down the right edge; if gaze jumps left and right inside the column, place-value never formed.

Where it stops holding

In right-to-left scripts the text landing rail is the right edge, so text columns should follow writing direction; numeric place-value still wants digits stacked, and many locales keep left-to-right numbers. Phone numbers, postal codes, and IDs look numeric but are not compared as magnitudes; they are scanned as strings and belong left-aligned. Mixed units (“12 hours” next to “3 minutes”) break place-value; what must align is a converted common magnitude, not the literal digits. Centering gives neither edge, hurting both jobs, and is tolerable only on short header or status labels.

Applying it

• Right-align quantities, money, percentages, and durations, and keep place-value with tabular or monospaced figures. Left-align names, descriptions, and statuses.
• Put currency signs and decimal points into the same stack; do not let a sign push the ones place out of the column. Keep minus signs or parentheses stable so the right edge does not break.
• Treat fields that look numeric but are read as codes as text.
• How to check: cover the headers and look at one column. A numeric column should let you point at the largest rows without reading aloud; a text column’s first letters should sit on a straight rail. If either fails, change the alignment.

Related

• Within the group: E4.03.1 A table is a surface for comparing attributes, not for reading a single record · E4.03.3 Column width should follow the type of content · E4.03.4 Narrow viewports need a restructured table, not a uniform scale
• Adjacent: E4.04 Sticky headers and horizontal scroll · E2.19 Currency, units, and magnitude input
• Search terms: numeric alignment · tabular figures · column alignment