Part of a whole

Diverging stacked bar chart

Positions ordered response components around a shared centre, often for survey comparisons.

  • Chart
  • Composition
  • Deviation

What it answers

How do positive and negative responses balance?

Shape
Stacked segments around a centre
What it shows
Balance of favourable and unfavourable responses.

Example

The question

How did staff rate each part of the new system?

Staff survey, share of answers by item

Staff survey, share of answers by itemSpeed: strongly disagree 4%, disagree 10%, neutral 16%, agree 48%, strongly agree 22%; Search: strongly disagree 12%, disagree 20%, neutral 18%, agree 36%, strongly agree 14%; Reports: strongly disagree 18%, disagree 26%, neutral 20%, agree 26%, strongly agree 10%; Support: strongly disagree 3%, disagree 7%, neutral 14%, agree 44%, strongly agree 32%.Strongly disagreeDisagreeNeutralAgreeStrongly agreeSpeedSearchReportsSupport50%25%0%25%50%75%100%4822122036141826264432
Staff survey, share of answers by itemSpeed: strongly disagree 4%, disagree 10%, neutral 16%, agree 48%, strongly agree 22%; Search: strongly disagree 12%, disagree 20%, neutral 18%, agree 36%, strongly agree 14%; Reports: strongly disagree 18%, disagree 26%, neutral 20%, agree 26%, strongly agree 10%; Support: strongly disagree 3%, disagree 7%, neutral 14%, agree 44%, strongly agree 32%.Strongly disagreeDisagreeNeutralAgreeStrongly agreeSpeedSearchReportsSupport50%25%0%25%50%75%100%
Speed: strongly disagree 4%, disagree 10%, neutral 16%, agree 48%, strongly agree 22%; Search: strongly disagree 12%, disagree 20%, neutral 18%, agree 36%, strongly agree 14%; Reports: strongly disagree 18%, disagree 26%, neutral 20%, agree 26%, strongly agree 10%; Support: strongly disagree 3%, disagree 7%, neutral 14%, agree 44%, strongly agree 32%.

Illustrative data. It belongs to no company.

When to use it

Use it when

Comparing several Likert-style survey items.

Avoid it when

Unordered components or unclear neutral/denominator rules.

How to read it

Read ordered response categories on each side of the centre; inspect neutral placement.

Note

State how neutral responses are positioned.

The data you need

Layout
Long table
Fields
Item: text; Response: ordered category; Count or Share: nonnegative number
One row is
One response-level count or share for an item. What is the grain?
Rules
Specify response order, shared denominator and neutral-centre rule. Aggregate individual responses first if necessary.
Optional
Raw alternative: RespondentID,Item,Response. Define reverse-coded questions.
Preparation
Validate the input grain and missing values. Specify response order, shared denominator and neutral-centre rule. Aggregate individual responses first if necessary.
Units
Counts or percentages of eligible respondents per item. Declare whether neutral/missing responses are in the denominator.
Example of the input layout: Long tableA long table: one row per observation, few columns. New data adds rows.ItemResponseCount or ShAABBC
A long table: one row per observation, few columns. New data adds rows.

Worked example

Data disagree=12, neutral=8, agree=30, total=50. Shares are 24%,16%,60%. Centre neutral consistently; favourable responses dominate.

ItemResponseCount
ServiceDisagree12
ServiceNeutral8
ServiceAgree30