Performance appraisal methods: origins, dates and evidence

When each method began, who built it, and what the research says about its accuracy.

Institute for Performance Management. Published 26 October 2026. Last reviewed 26 October 2026.

Most lists of appraisal methods give a definition and stop. This page adds the part writers and students usually have to dig for: when each method first appears in the record, who built it and why, and what the research says about how well it measures performance. Every date links to a source, listed at the end.

The methods at a glance

Method	First documented	Origin	How it scores people	Main weakness
Graphic rating scale	1920	Walter Dill Scott and the Scott Company, United States	A mark on a line or numbered scale for each trait	Halo and leniency; ratings say more about the rater than the ratee
Forced choice	1946 to 1948	US Army Personnel Research Section	Rater picks which of several equally favourable statements fits best	The rater cannot see how choices translate into a score, so the result is hard to explain to the employee
Critical incident technique	1954 (wartime work from the 1940s)	John C. Flanagan, US Army Air Forces Aviation Psychology Program	A record of specific effective and ineffective behaviours	Slow to collect; depends on how diligently incidents are logged
Management by objectives	1954	Peter Drucker, The Practice of Management	Results against goals agreed in advance	Rewards what is easy to measure
Multi-source (360-degree) feedback	1950s	Esso Research and Engineering Company	Ratings from managers, peers, reports and sometimes customers	Raters still disagree heavily with one another
Behaviourally anchored rating scale (BARS)	1963	Patricia Cain Smith and Lorne Kendall	A scale whose points are described by real job behaviours	Costly to build; gains over simple scales are smaller than hoped
Forced distribution (forced ranking)	No single origin; best known from GE under Jack Welch	GE's "vitality curve" under Jack Welch	Fixed shares of staff placed in top, middle and bottom groups	Damages collaboration and invites legal challenge; many adopters have dropped it
Continuous check-ins without ratings	2012	Adobe; later Deloitte and GE	Frequent conversations, sometimes with a few forward-looking questions	Too new for long-run evidence

Ranking, paired comparison, checklists and written essays are older than any of these and have no single documented origin, so they are covered briefly below rather than dated.

Before the methods: rating people for office

Grading people against a fixed scale is far older than modern management. In 220 AD the Chinese state of Cao Wei adopted a proposal from the court official Chen Qun: a common scale of nine ranks, with controllers appointed to place candidates for office on it. In theory candidates were graded on moral probity, administrative ability and service to the state. In practice a father's grade influenced his son's, and later controllers from powerful clans promoted their own descendants. A saying from the period ran: "There are no poor people in the upper ranks and no powerful families in the lower ones."¹ The system lasted in some form until the Song dynasty.

That is worth knowing because it shows, very early, the problem every later method tries to fix: the grade reflects the grader and the person's connections as much as the work.

The industrial version starts with Robert Owen, who ran the New Lanark mills in Scotland from 1800 to 1824. Owen used a "silent monitor" for each worker: a square block of painted wood the overseer turned to show how the worker had behaved the previous day. Black meant bad, blue indifferent, yellow good and white excellent.²

The methods in order of origin

Graphic rating scales (1920)

The rating scale most organisations still use descends from work by the American psychologist Walter Dill Scott. During the First World War Scott designed a rating scale for selecting US Army officers. His first version rated traits such as appearance, military experience, influence over men and regard for authority,³ and by the end of the war the system was in use across the Army.⁴

Scott then took the idea into industry through the Scott Company, founded in 1919.⁴ The graphic rating scale itself, where the rater marks a point along a line for each trait, was introduced by Scott in a magazine article in November 1920 and set out in Scott and Hayes's 1921 book. Donald Paterson's 1922 paper described "The Scott Company graphic rating scale", and Max Freyd's 1923 review spread it through the research literature. A 2021 review of who invented it concludes that the scale was a joint achievement of Scott Company staff, but "if one still wants to name the sole inventor, Walter Dill Scott will be it."⁵

The trouble appeared almost immediately. In 1920 Edward Thorndike studied ratings of employees in two large industrial companies and of army officers rated by their superiors. Ratings of supposedly separate traits moved together far more than they should. "Obviously a halo of general merit is extended to influence the rating for the special ability, or vice versa," he wrote, giving the halo effect its name.⁶

First documented: November 1920 (Scott); army precursor 1917 to 1918.

Evidence: the halo effect was identified in 1920, in the same year the scale was published.

Forced choice (1946 to 1948)

After the Second World War the US Army had a problem familiar to any HR department: inflated ratings. Its Personnel Research Section tested a "forced choice" format for officer efficiency reports. Instead of marking a scale, the rater picks which of a set of equally favourable-sounding statements best and least describes the officer. Because the statements look equally favourable, the rater cannot easily tell which choices raise the score. A 1946 Army comparison of the methods notes that an ordinary rating scale makes it possible for the rater "to raise or lower the final score at will," while the paired design reduces the chance he "can deliberately mark the blank to increase or decrease the total score."⁷ E. Donald Sisson described the approach publicly in 1948 in "Forced Choice: The New Army Rating."⁸

First documented: Army research reports from 1946; published 1948.

Evidence: designed to resist deliberate inflation; the trade-off is a score the rater cannot easily explain.

Critical incident technique (1954)

John C. Flanagan developed the technique in the US Army Air Forces Aviation Psychology Program during the Second World War, studying what separated effective from ineffective behaviour, including the causes of pilot errors.⁹ He published the method in 1954, defining it as "a set of procedures for collecting direct observations of human behavior."¹⁰ In appraisal, the manager keeps a running record of specific incidents of good and poor performance, which then supplies evidence for the review conversation.

First documented: 1954 publication; wartime development in the 1940s.

Evidence: grounds the review in observed behaviour rather than general impressions, but only as good as the manager's record keeping.

Management by objectives (1954)

Peter Drucker coined "management by objectives" in The Practice of Management in 1954.¹¹ The ideas were not new, and Drucker drew on earlier practice, but he assembled them into a system: agree objectives in advance, then judge results against them. His student George Odiorne developed it further in the 1960s.¹¹

Three years later Douglas McGregor published "An Uneasy Look at Performance Appraisal" in the Harvard Business Review. He argued that judging another person's worth forces managers into the uncomfortable position of "playing God," and that appraisal treats people like products passing an inspection line.¹²

First documented: 1954 (the term).

Evidence: clear where results are measurable; weak where the important work resists numbers.

Multi-source or 360-degree feedback (1950s)

One of the earliest recorded uses of surveys to gather feedback about employees was at Esso Research and Engineering Company in the 1950s.¹³ Popular histories often trace it further back, to German army officer selection around 1930. We could not find a primary source for that claim, so we do not date the method from it. The practice spread widely once online tools made collecting ratings from many people cheap.¹³

First documented: 1950s.

Evidence: more raters average out some individual bias, but raters still disagree heavily (see the research section).

Behaviourally anchored rating scales (1963)

Patricia Cain Smith and Lorne Kendall published the method in the Journal of Applied Psychology in 1963, in response to dissatisfaction with how subjective ordinary rating scales were.¹⁴ Each point on the scale is described by an example of real job behaviour, usually gathered using Flanagan's critical incident technique, so that "4" means the same thing to every rater. Building one is laborious: experts collect incidents, sort them into dimensions, reassign them blind and discard any that experts disagree on.

The payoff has been modest. A 1975 review found the format's strength may lie mainly in the careful definition of performance dimensions rather than in the behavioural anchors themselves, and later work found BARS can still suffer from unreliability, leniency and poor separation between dimensions.¹⁴

First documented: 1963.

Evidence: better defined than a plain scale, not reliably more accurate.

Forced distribution or forced ranking

Forced distribution requires managers to place fixed shares of their staff into groups, typically with around 10% labelled as underperforming.¹⁵ Its best-known form was GE's "vitality curve" under Jack Welch, nicknamed "rank and yank," in which the bottom 10% were let go. GE dropped formal forced ranking around 2005.¹⁶ Microsoft ended its version in November 2013, announcing there would be "no more curve," after long criticism that it damaged collaboration.¹⁵

The case for it is inflation. A 2020 US Army review notes that before the Navy introduced forced distribution, the majority of its officers were being rated in the top 1%.¹⁷

First documented: no single origin; GE dropped it around 2005 and Microsoft in 2013.

Evidence: stops grade inflation, at the cost of teamwork and legal risk.

Continuous check-ins without ratings (2012 onward)

Adobe dropped annual reviews in 2012, replacing them with regular "check-in" conversations. The old process had consumed around 80,000 manager hours a year. After the change, voluntary attrition fell 30% while involuntary departures rose 50%.¹⁸

Deloitte followed publicly in 2015. Its old process took close to 2 million hours a year, and its redesign replaced ratings of the person with four statements about what the team leader would do: give this person the largest raise if it were their own money, always want them on the team, see them at risk of low performance, consider them ready for promotion today.¹⁹ GE began moving its 300,000 staff off annual reviews in 2015 to an app called PD@GE, with frequent "touchpoints" and peer feedback.¹⁶

First documented: 2012.

Evidence: company-reported results only so far; no long-run independent studies.

Ranking, paired comparison, checklists and essays

Simple ranking (ordering staff from best to worst), paired comparison (comparing each person with every other in turn), checklists of behaviours and free-written essays are all older than the methods above and have no single documented starting point. Comparison methods remove the problem of everyone being rated "above average," because someone has to come last, but they still allow halo errors, and they say nothing about whether the best person in a weak team is any good.²⁰

What a century of research says about accuracy

Psychologists have studied appraisal ratings for about as long as the rating scale has existed, and the findings are consistent enough to state plainly.

The rater matters more than the person rated. A 2000 study of two large samples of managers, each rated by seven people, found that the individual rater's idiosyncrasies accounted for 62% and 53% of the variance in ratings. The ratee's actual performance accounted for 21% and 25%.²¹ This is the study Deloitte cited when it scrapped ratings.¹⁹

Two managers rating the same person often disagree. A 1996 meta-analysis put the average interrater reliability of supervisors' ratings of overall job performance at .52.²² Two supervisors rating the same employee agree only moderately.

Redesigning the form has done less than expected. In 1980 Frank Landy and James Farr reviewed decades of work on rating formats and called for a moratorium on format research, arguing that changing the scale was not where the gains lay.²³ Researchers have since returned to the question, with some evidence that format design does affect rating quality.²⁴

Feedback can make performance worse. A 1996 meta-analysis of 607 effect sizes found feedback improved performance on average, but over a third of feedback interventions decreased it.²⁵ How the conversation is handled matters at least as much as which method produced the rating.

A 2017 review of 100 years of appraisal research in the Journal of Applied Psychology, co-written by Kevin Murphy of the University of Limerick, recommended that future work concentrate on performance management as a whole rather than on rating methods alone.²⁶

Timeline

Year	Event
220 AD	Cao Wei adopts Chen Qun's nine-rank system for grading candidates for office¹
1800 to 1824	Robert Owen's silent monitors at New Lanark²
1917 to 1918	Walter Dill Scott's officer rating scale used across the US Army^{3, 4}
1920	Scott introduces the graphic rating scale; Thorndike names the halo effect^{5, 6}
1946 to 1948	US Army develops and publishes forced-choice ratings^{7, 8}
1950s	Esso Research and Engineering uses multi-rater surveys¹³
1954	Flanagan publishes the critical incident technique; Drucker coins management by objectives^{10, 11}
1957	McGregor's "An Uneasy Look at Performance Appraisal"¹²
1963	Smith and Kendall publish behaviourally anchored rating scales¹⁴
1980	Landy and Farr call for a moratorium on rating format research²³
1996	Meta-analyses on rating reliability and on feedback effects^{22, 25}
2000	Study finds rater idiosyncrasy explains over half of rating variance²¹
Around 2005	GE drops formal forced ranking¹⁶
2012	Adobe replaces annual reviews with check-ins¹⁸
2013	Microsoft ends stack ranking¹⁵
2015	Deloitte publishes its redesign; GE begins moving off annual reviews^{16, 19}
2017	DeNisi and Murphy review 100 years of appraisal research²⁶

How to cite this page

Institute for Performance Management (2026). Performance appraisal methods: origins, dates and evidence. Retrieved from https://www.certifiedperformance.org/performance-appraisal-methods

If you spot an error or have an earlier source for any date here, contact us and we will correct the page and credit you.

The Institute's performance appraisal training covers how to run these methods well in practice.

Sources

Wilkinson, E. (2012). Chinese History: A New Manual. Harvard University Asia Center, pp. 265-266; Wang, X. (2007), Wen Shi Zhe 2007(6), 62-69; Zhang, X. (1999), Journal of Zhengzhou University 32(11), 24-30. As summarised in "Nine-rank system", Wikipedia.
Vassar College Archives and Special Collections, Robert Owen exhibit: communities.
Lynch, E. C. (1968). Walter Dill Scott: Pioneer industrial psychologist. Business History Review, 42(2), 149-170. doi:10.2307/3112213
"Walter Dill Scott", Wikipedia, citing the WWI biography of Scott at angelfire.com/biz/pottershouse.
Shiina, K. (2021). Frontiers in Human Neuroscience, 15, 711691. doi:10.3389/fnhum.2021.711691. Primary sources it discusses: Scott (1920), American Magazine 90, 44-47; Scott and Hayes (1921), Science and Common Sense in Working With Men; Paterson (1922), Journal of Personnel Research 1, 361-376; Freyd (1923), Journal of Educational Psychology 14, 83-102.
Thorndike, E. L. (1920). A constant error in psychological ratings. Journal of Applied Psychology, 4, 25-29. PDF
Rundquist, E. A. (1946). Comparison of the Rating Check List (RCL) and Forced Choice List (FCL) Methods of Obtaining Ratings. PRS Report No. 717, Personnel Research Section, Adjutant General's Office, US Army. DTIC
Sisson, E. D. (1948). Forced choice: The new Army rating. Personnel Psychology, 1. Cited in Evans, L. A. and Robinson, G. L. (2020), Evaluating our evaluations, Military Review 100(1).
"Critical incident technique", Wikipedia, citing APA Division 19 and the American Institutes for Research.
Flanagan, J. C. (1954). The critical incident technique. Psychological Bulletin, 51(4), 327-358. PubMed 13177800
Drucker, P. F. (1954). The Practice of Management. Harper. Antecedents and Odiorne's development as summarised in "Management by objectives", Wikipedia, citing LaFollette and Fleming (1977), Academy of Management Proceedings 1977(1), 2-5.
McGregor, D. (1957). An uneasy look at performance appraisal. Harvard Business Review, 35(3), 89-94. Argument as summarised in Journal of Business and Management.
Bracken, D. W., Dalton, M. A., Jako, R. A., McCauley, C. D. and Pollman, V. A. (1997). Should 360-degree feedback be used only for developmental purposes? and Atkins, P. W. B. and Wood, R. E. (2002), Personnel Psychology 55(4), 871-904. As cited in "360-degree feedback", Wikipedia.
Smith, P. C. and Kendall, L. M. (1963). Retranslation of expectations. Journal of Applied Psychology, 47(2), 149-155; Schwab, D. P., Heneman, H. G. and DeCotiis, T. A. (1975), Personnel Psychology 28(4), 549-562; Kingstrom, P. O. and Bass, A. R. (1981), Personnel Psychology 34(2), 263-289. Via "Behaviorally anchored rating scales", Wikipedia.
Olson, E. G. (18 November 2013). Microsoft, GE, and the futility of ranking employees. Fortune.
Nisen, M. (13 August 2015). GE had to kill its annual performance reviews after more than three decades. Quartz, republished by Government Executive.
Evans, L. A. and Robinson, G. L. (2020). Evaluating our evaluations: Recognizing and countering performance evaluation pitfalls. Military Review, 100(1).
Chartered Management Institute (11 May 2018). Why Adobe killed off the annual performance review.
Buckingham, M. and Goodall, A. (2015). Reinventing performance management. Harvard Business Review, April 2015, 40-50. HBR
Comparison methods and halo error as summarised in "Performance appraisal", Wikipedia, citing Howes and Muchinsky (2022).
Scullen, S. E., Mount, M. K. and Goff, M. (2000). Understanding the latent structure of job performance ratings. Journal of Applied Psychology, 85(6), 956-970. doi:10.1037/0021-9010.85.6.956
Viswesvaran, C., Ones, D. S. and Schmidt, F. L. (1996). Comparative analysis of the reliability of job performance ratings. Journal of Applied Psychology, 81(5), 557-574. doi:10.1037/0021-9010.81.5.557
Landy, F. J. and Farr, J. L. (1980). Performance rating. Psychological Bulletin, 87(1), 72-107.
Gorman, C. A. and Meriac, J. P. (2015). It's okay to do rating format research again. Symposium, Society for Industrial and Organizational Psychology, Philadelphia.
Kluger, A. N. and DeNisi, A. (1996). The effects of feedback interventions on performance. Psychological Bulletin, 119(2), 254-284. doi:10.1037/0033-2909.119.2.254
DeNisi, A. S. and Murphy, K. R. (2017). Performance appraisal and performance management: 100 years of progress? Journal of Applied Psychology, 102(3), 421-433. doi:10.1037/apl0000085