Two-factor theory and the bias that produces it

Frederick Herzberg published in the late 1950s a study based on interviews asking engineers and accountants to recount particularly satisfying and dissatisfying work episodes.

He distinguishes motivation factors, tied to the content of work, responsibility, recognition, achievement, from hygiene factors, tied to context, pay, conditions, relationships, whose absence produces dissatisfaction without their presence producing satisfaction.

The thesis is more radical than it looks: satisfaction and dissatisfaction would not be the two ends of a single scale, but two independent dimensions. Removing a source of dissatisfaction would therefore not make anyone satisfied, only not dissatisfied.

The result is obtained through retrospective narrative. Yet work on attribution shows a systematic tendency to attribute favourable episodes to oneself and unfavourable ones to circumstances.

That tendency mechanically produces the observed separation: internal factors appear on the satisfaction side, external ones on the dissatisfaction side, without any structure of needs having to exist. The result would then be a property of the narrative, not of the work narrated.

Studies using questionnaires rather than narratives do not find the separation into two families. Several syntheses in the 1960s and 1970s conclude that the result depends on the method used. The same factor, pay for instance, appears in them as a source of satisfaction or of dissatisfaction depending on the case.

Herzberg and his defenders replied that other methods do not capture what narratives capture, an argument hard to settle since it amounts to defining the phenomenon by the instrument.

The theory founded the job enrichment movement, which proposes broadening the content of the job rather than altering its conditions. It is in that sense the ancestor of many later schemes.

Its spread owes much to an article Herzberg published in 1968 in a management journal, presenting a simplified version aimed at practitioners. It is that version, more than the original study, that circulates in training.

It also supplied a convenient argument: classing pay among hygiene factors allows the claim that a raise does not motivate, a proposition whose empirical validity is uncertain.

The idea that the content of work matters independently of its conditions is supported by other better established work, notably Hackman and Oldham's job characteristics model. It describes the job through measurable dimensions, such as task variety, autonomy or feedback, and tests their effects by questionnaire.

It is a frequent configuration: a fragile theory opens a line that is later taken up with better methods, and the name attached remains that of the initial version.