Main Page
Alphabetical Index

Hints for prints


Use of repertory grids in quality control
The processes used to assess the quality of manufactured goods, and to prevent the production of faulty products, are, one imagines, unproblematic. After all, manufactured goods are physical objects whose required dimensions, weight, etc. can surely be specified in straightforward objective  terms. Vehicle cylinder heads can be machined ‘to the nearest thou’ (thousandth of an inch) using a laser-controlled lathe; cardboard boxes filled with cornflakes by machinery that operates to the nearest hundredth of a gram. Personal judgement, one imagines, simply isn’t involved.
There are, however, many processes which are not as straightforward, and in which a surprising degree of expert judgement is required. The ‘finish’ of a garment is one example. The consistent mixture of differing types and grades of tea-leaf according to the seasonal availability of entirely different leaves in order to produce the same taste, consistently each time, is another. Typically, these judgements are made by experienced employees who may not be able to articulate the logic underpinning their judgements beyond a simple assertion that they know when they’ve got it right: ‘It feels good to me’ or ‘it isn’t right’ being typical statements.
One can assess the accuracy of these kinds of judgement very straightforwardly: when correct, the customer doesn’t complain, and when incorrect judgements have been made, the goods are returned and the customer complains. The real problem arises when one wishes to improve quality control accuracy, and particularly, when one wishes to train new controllers to replace staff who are leaving. One needs a device for making tacit knowledge explicit (Jankowicz 2001).
To what should the novice quality controller learn to attend? What, exactly, is meant by the ‘finish’ of a garment? Is it the precision with which button-holes have been stitched and the care with which buttons have been sewn on? Or is it rather, the feel of the material as it’s rubbed between the thumb and the forefinger, or put to the operator’s cheek? And in the latter case, what should the nature of the sensation be in the case of silk which has been woven properly or with faults during the process? Does it differ in the case of cotton, this distinction between the successful and unsuccessful weave? What constitutes ‘seconds’?
The repertory grid can be used to answer questions like these. A dozen elements, each being a physical sample of the product, the whole set varying in overall quality as determined by the experienced quality controller, is used. Elements are presented in triads and the controller is asked to indicate "in what way are two of these the same, and one different, in terms of quality? What exactly is it about them that makes you say that? What are you noticing, and how are you noticing it?"
The answers are in the form of constructs, the implicit pole being identified with as much care as the emergent pole to identify the nature of both ends of the quality scale. A set of constructs can be obtained from several quality controllers and aggregated or compared. The attributes which are particularly characteristic of a well-finished garment can be pointed to, and the specific faults which must be identified during the quality inspection demonstrated to the novice quality controller. Or, indeed, the exact faults to be avoided during initial manufacture can form part of the training of the operatives who produce the product in the first place, as part of a detailed task analysis of the job itself.  


  • Jankowicz, A.D. (2001). Why does subjectivity make us nervous? Making the tacit explicit. Journal of Intellectual Capital 2, 1, 61-73.

Devi Jankowicz

Establ. 2003
Last update: 15 February 2004