Goleman and Hersey–Blanchard: how sceptical to be

In Goleman's table (2000), the "authoritative" style correlates 0.54 with a team's working climate and "pacesetting" correlates −0.25 (Goleman, 2000, p. 81). A team leader who reads that may conclude: drop the second style. Hersey and Blanchard's model, for its part, tells you to adjust how much you direct and how much you support, person by person and task by task. This post asks what evidence stands behind each model and how far to trust it before taking either into your own team.

What does Goleman say, and where do the numbers come from?

Goleman's six leadership styles are coercive, authoritative, affiliative, democratic, pacesetting and coaching. Goleman (2000) compares them to the clubs in a golfer's bag: a good leader picks the club to suit the shot (Goleman, 2000, p. 80).

The numbers come from research by the consultancy Hay/McBer, on a random sample of 3,871 executives drawn from a database of more than 20,000 worldwide (Goleman, 2000, p. 79). The team measured the "climate" around each executive through six factors: flexibility, responsibility, standards, rewards, clarity and commitment (Goleman, 2000, p. 81). The article says the analysis "strongly suggests" that climate accounts for nearly a third of results (Goleman, 2000, p. 82). In the part I read, the calculation is not shown.

What does that table actually prove?

Not causation. The exhibit is titled the "impact" of leadership styles on climate, yet the sentence introducing it says the figures are correlations between each style and each climate factor (Goleman, 2000, p. 81). As correlation is not causation warns, the reverse direction is just as plausible: a team that is going well may make a warm style easy to use, and a team in disarray may push a leader towards coercion. The reverse reading is my own; the article does not discuss it.

On the source: the only footnote in the reprint says Goleman consults for Hay/McBer on leadership development (Goleman, 2000, p. 90). That is a disclosure, not an accusation. But author and data source are not independent, and I would want an outside replication before trusting any single figure. In the version I read, the article does not say how styles and climate were measured, how many people lie behind each coefficient, or whether any are statistically significant. The sentence "Many studies, including this one, have shown that the more styles a leader exhibits, the better" cites no study, and the reprint has no reference list (Goleman, 2000, p. 87).

The article also weakens its own "when to use" chart. Goleman writes that the best leaders "don't mechanically match their style to fit a checklist of situations – they are far more fluid" (Goleman, 2000, p. 87). And the main illustration of flexibility is one case: Joan, general manager of a division of a global food and beverage company, whose division had missed its profit targets for six years (Goleman, 2000, p. 88). Seven months later it beat the yearly target by $5 million, which the article calls "the ultimate proof" of her style (Goleman, 2000, p. 89). One case, no comparison group, no company named.

What does situational leadership say, and why are there so many versions?

Situational leadership says to choose how much to direct and how much to support according to the follower's readiness for a given task. According to Graeff (1997), it first appeared in the Training and Development Journal in 1969 as the "Life Cycle of Leadership", then ran through several editions of Hersey and Blanchard's textbook (Graeff, 1997).

The trouble is that the model has changed many times. Per Graeff, Blanchard and colleagues in 1985 turned "ability and willingness" into "competence and commitment" and renamed the four styles from "telling, selling, participating, delegating" to "directing, coaching, supporting, delegating"; Hersey and Blanchard's 1988 textbook kept a structure close to the 1982 one (Graeff, 1997). The result is that one follower can receive two different pieces of advice.

Graeff also raises a conceptual flaw that touches use. In his argument, SLII's "competence" is very close to the old "ability", and "commitment" combines confidence with motivation, so it is unclear how it differs from "willingness" (Graeff, 1997). If the two inputs are that blurred, then, in my reading, placing someone in a cell is the leader's judgement, not a measurement.

What do the tests show?

Little, and unevenly. Graeff (1997) counted five published studies of the original version. Hambleton and Gumpert (1982) concluded in favour, though Vecchio pointed to problems of method. Vecchio (1987) and Norris and Vecchio (1992) found support at low maturity, mixed results at the two middle levels and none at high maturity. Blank and colleagues (1990) and Goodson and colleagues (1989) found no support. Graeff's verdict: "very weak" support (Graeff, 1997).

Thompson and Vecchio (2009) tested three versions: the original 1972 statement, the revised 2007 theory and a third formulation built on the autonomy a leader gives followers and their job experience. The data are surveys of 357 employees and 80 supervisors at ten Norwegian financial institutions. According to the abstract, the 2007 revision predicted subordinate performance and attitudes worse than the original, and the third formulation was described as promising for further exploration, though it too added little to the variance explained (Thompson and Vecchio, 2009). In my reading, the part worth keeping is the intuition that people with different experience need different amounts of autonomy; the specific prescriptions are unconfirmed.

Another post on this site places the model inside a change programme: situational leadership during a change.

How do you use both with the right amount of doubt?

This section is my inference from the sources above; no study has tested it. The two models share an untested input: the leader must read the follower correctly. Goleman's "when to use" row speaks of employees who are not competent enough (Goleman, 2000, p. 85), who are self-motivated and highly skilled (Goleman, 2000, p. 86), or who want to be coached (Goleman, 2000, p. 87). That is the variable Graeff finds blurred. The pairing is mine.

Use them if you treat them as a list of questions before handing out a task or holding a one-to-one, and write down what happens. Do not use them to explain why someone "failed" or to label people by "style". Four questions:

1. Which number do you want to move, how will you count it, over how long? With nothing countable, no model can be refuted.
2. For each person and task, how do their ability and their willingness differ? Write the two separately and ask the person.
3. Which version are you using, and do its labels match the text you read?
4. What else changes during the trial: workload, tools, new joiners?

A test on a hypothetical team: 12 people in customer care, names invented by me, new task refund handling. Hà is fast at refunds but bored; Tuấn has just joined, is keen and often gets them wrong. Write two columns for each person (can do it yet, wants to do it) and ask them to fill in the same grid about themselves. Where the two grids differ is where the model's input is weakest, and a conversation should come before choosing a style. Then pick one style per person for four weeks and record the handling time per case and the number of cases redone. With 12 people this proves nothing statistically; it only tells you whether the model's advice matches what you see.

Read Goleman's table the same way. Pacesetting scores −0.25, yet Goleman himself says it works with a highly skilled, self-motivated team (Goleman, 2000, p. 86); an average from a database does not forbid you using it for one particular person. And if what you meet is resistance to a change, not missing skill or interest, the diagnosis differs: see Kanter's ten reasons for resisting change.

Going deeper: what did I read, and what did I not?

I read Goleman (2000) in full through a Harvard Business Review reprint posted on a website other than the publisher's. Page numbers follow those printed in the reprint's footers; the opening page carries none, so "page 79" is inferred from the order. I read Graeff (1997) in full from a PDF, also not from the publisher, with no journal page numbers, so I cite it without pages. For Thompson and Vecchio (2009) I read only the abstract, through aggregator pages, because the publisher's page refused access; I do not know how they measured effects.

I have not read the Hersey and Blanchard textbook; every statement about what the versions say comes through Graeff. I read no independent test of Goleman's six styles, no research on situational leadership after 2009, and no source on Vietnam. Two places where the site's library differs from what I read: the framework page calls Goleman's data a survey, whereas the article says the team "studied data about or observed" executives (Goleman, 2000, p. 81); and the situational-leadership page summarises the evidence as "inconsistent", whereas Graeff, on data up to 1997, is harsher. The customer-care team and its people are hypothetical.