Workplace Reality Index methodology

Instruments and question wording

Each instrument is twelve statements on a five point agreement scale, four dimensions of three statements. The exact wording of every statement is on the tool page itself, shown to every respondent in the same order.

Sponsor

Leaders ADAPT LLC, a leadership coaching and development firm, designs and publishes the index. The instruments sit on the firm's own website and the firm sells leadership programmes. That interest is stated here so it can be weighed.

Population and how respondents are reached

Respondents are self selected: people who found one of the ten free tools, through search, a link or a share, and chose to complete it. They are not a random sample of workers and nothing here should be read as describing the workforce as a whole. They are not leaders unless the cut says so. Respondents answer an optional question before each test, whether they manage people themselves, and every instrument reports that cut separately.

Unit of measurement

The unit is a completion, not a person. Capture is anonymous and cookieless, so a person who takes a tool twice is counted twice. Ad blockers prevent some completions from being recorded, so every count is a floor.

Scoring

Each instrument has twelve statements on a five point agreement scale, three per dimension. Reverse worded statements are stored as six minus the response. A dimension score is the stored sum minus three, divided by twelve, times 100. The overall score is the stored sum minus twelve, divided by 48, times 100. Each tool page states in plain words what a high score means for that instrument, and how the result is assigned.

What is recorded

One anonymous event per completion: the instrument, its version, the score, the result, the four dimension scores, the twelve answers as digits, the optional routing answer, the optional region, the completion time and whether the visitor arrived from a forwarded link. Never a name, an email address, an employer, the wording of an item or any free text. An email address exists only if a respondent asks for their result by email, and it never enters the index.

Reporting floor

No cut is published with fewer than 30 completions, per instrument, per cut and per window. Below the floor a cut is shown as a count with no figures. The floor is 30 because below it a single respondent moves a percentage by more than three points. It is set before the data and never lowered. Every cut by role, region, device or referrer is itself subject to the floor.

Window and editions

Statistics pages show a rolling 90 day window ending on the refresh date, refreshed on the 12th of each month by a scheduled job that recomputes every figure from the raw events. A quarterly edition of record is published on the 5th of January, April, July and October, as its own edition file and page, never shared with the leader indexes. The first named release waits until one instrument reaches 385 completions, the minimum for a margin of error of five points at 95 percent confidence on a large population; a press summary waits for 1,000.

What is never published

• Any percentage of bosses, managers or workplaces who are a label. Figures describe what respondents report experiencing.
• Any share of respondents who say they would quit, intend to leave or expect to be fired. Stated intent moves too fast to be a statistic: one survey series on return to office mandates saw stated quit intent fall from 51 percent to 7 percent within a year (HR Dive).
• Any cut that names an employer, an industry small enough to identify one, or a country below the floor.
• Any free text. None is collected.
• Any figure from a window in which the site's own test runs have not been identified and removed.

Separation from the leader indexes

The Leadership Readiness Index and the AI Leader Readiness benchmark are built from leader facing instruments and describe leaders. The Workplace Reality Index is built from the ten instruments listed above and describes whoever took them. The two are never merged. Where an instrument here has a manager cut, it may be compared with a leader index in a paragraph, with both sources named, and never added to it.

Limitations

• Self selection: people who search for a toxic boss test are more likely than average to have a difficult boss. Figures describe the people who took the tool.
• Self report: every answer is the respondent's perception. The instruments measure experience, not fact.
• Instrument wording: these are coaching instruments, not validated psychometric scales, and they are not clinical instruments. Results are not diagnoses.
• Counts are floors because of ad blockers, and repeated completions by one person cannot be removed.

Corrections

If a published figure is wrong, the statistics page carries the corrected figure and a dated line in its change log saying what changed and why. Nothing is deleted silently.

Change log