Implicit Logic in Managerial Discourse . A Case Study in Choice of Selection Criteria

A grande maioria das teorias de selecção têm dado pouco atenção ao modo como gestores como seleccionadores identificam e justificam critérios de selecção e até que ponto existem evidências para a sua consistência e lógica. Este artigo, na base de um quadro conceptual sócio-cognitivo, propõe uma reflexão sobre estas questões. Foi desenvolvido um código de análise para explicar o discurso utilizado por 22 gestores no que se refere à justificação de critérios de selecção para operadores de câmara numa empresa europeia de comunicação social. Os resultados do estudo evidenciam, ainda que se trate de uma função técnica, que os gestores com experiência na função para a qual os candidatos estão a ser seleccionados se manifestam mais preocupados com valores, crenças e personalidade; e que estes critérios, independentemente do nível de hierarquia na gestão e da experiência em entrevistas de selecção, todos os gestores estão mais preocupados com a articulação pessoa-organização para necessidades presentes e futuras da organização do que com a especificidade pessoa-trabalho para o desempenho imediato. A consistência dos resultados sugere a existência de uma 'lógica implícita' no modo como os gestores justificam critérios que resultam de uma aprendizagem implícita e conhecimento tácito de ambas as experiências organizacionais e operacionais.

Little attention has been paid in mainstream selection theory to how selectors choose to justify criteria and whether there is evidence of any consistency or logic in the manner in which they do so.This paper addresses these questions within a socio-cognitive framework.A newly developed coding system is used to analyse and explain the discourse of 22 managers in justifying selection criteria for technical operators in a European broadcasting company.It was found that, even for a very technical position, managers with experience of the job for which candidates were being selected were more concerned with the values, beliefs and personalities of candidates.It also was found that, independently of their different levels of seniority and experience of selection or interviewing, all managers are more concerned with Person-Organisation Fit for both present and future needs than with immediate Person-Job Fit.The consistency of the findings suggests that there is an 'implicit logic' in the manner in which managers as selectors adopt criteria derived from implicit learning and tacit knowledge of both operational and organisational experience.

JEL Classification: M12; M51
Throughout management studies there is an increasing concern that explicit, inferential paradigms are not capturing what really counts for managers in a world of increasing competitive pressures.In his Managers Not MBAs (2004), Henry Mintzberg claims that: 'Organizations are complex phenomena.Managing them is a difficult nuanced business, requiring all kinds of tacit understanding that can only be gained in context' (Mintzberg, 2004:9).Nonaka (1994;1998), giving examples from NEC, Sharp, Canon, Matsushita, Honda and other companies, had claimed that it is when tacit and explicit knowledge interact that: 'something powerful happens', and that: 'It is precisely this exchange between tacit and explicit knowledge that Japanese companies are good at developing… What's more, as new explicit knowledge is shared throughout an organization, other employees begin to internalize it -that is, they use it to broaden, extend and reframe their own tacit knowledge' (Nonaka, 1998: 29-31).Although not even mentioned by Michael Porter (1980;1990;1998), tacit knowledge already has a wide resonance in management theory as a key basis of competitive advantage (e.g.Baumard, 1999;Ambrosini and Bowman, 2001;Edmonson, Winslow, Bohmer and Pisano, 2003).And this is precisely because tacit knowing or know-how is not readily identifiable and replicable in the manner of explicit knowledge, such as a formula for a process, a patent, or licence.Fincham and Rhodes (2005), contrast explicit knowledge as formally expressed in some form that can be 'codified' and manipulated by logical reasoning, with tacit knowledge as a kind of knowing that cannot be easily transmitted or interpreted without a 'knowing subject'.Yet such 'knowing' by individuals or groups is central to Nonaka's (1998) case on interfacing the tacit and explicit.Does this then limit the operability of the concept of tacit knowledge?Or, are there methodologies that can offset Fincham and Rhodes' (2005) claims and that could explicate, classify and code such knowledge in a consistent manner?For instance, are Ambrosini and Bowman (2001) claim that identifying and enhancing tacit knowledge within organisations may not be as difficult as widely presumed since, in large part, knowledge, abilities or skills may be tacit simply because "people never thought of what they were doing, they never asked themselves what they were doing, and nobody else ever asked it either"?(2001: 816).This case study addresses such issues in one of the most vital areas for any organization: personnel selection.It introduces the still dominant 'normative' selection theory paradigm which claims that only explicit criteria should be adopted in selection (e.g.Guion, 1965Guion, , 1997;;Dipboye, 1994Dipboye, , 1997;;Anderson, 1997;Anderson andShackleton, 1990, 1993) and that managers should be trained and experienced in such techniques if they are to be able to interview rationally and well.It then considers the socio-cognitive basis of theories of tacit knowledge; suggests that insights into this may be gained from Ignacio Matte Blanco's (1975;1988) concept of 'unconscious logic', and introduces the concept of 'implicit logic' in discourse analysis.In explaining and advocating the methodology of a 'grounded theory' approach it relates this to the debate within management literature between those who espouse a 'critical realist' approach to discourse (e.g.Reed, 2005) and those following Derrida (1978) who claim that discourse is only a 'social construction' (e.g.Gergen, 1994Gergen, , 2000;;Contu and Wilmott, 2005).In introducing the case study, it describes the method of a newly developed coding system for analysis of managers' discourse concerning choice of selection criteria in a major service organization.It then presents and discusses the findings on which criteria managers implicitly prioritise and sets of criteria in their discourse on what is important for choice of candidates in selection before drawing conclusions on what may be the wider relevance of such findings to how organisations need to adapt to changing operational needs and competitive pressures.
The methodology of self-styled 'normative' selection theory is rational in the sense of being conscious, inferential and overt.Its rationale is that explicitly identifying and prioritising selection criteria ex ante, before an interview, will lead to a higher quality of selection decision-making and better predictors of performance.Yet the widespread lament of the advocates of such a normative approach (e.g.Guion, 1965Guion, , 1997;;Dipboye, 1994Dipboye, , 1997;;Anderson, 1997) is that managers rarely do this.Guion (1965) himself allows that choice in selection entails some final integrating judgement, and that 'the final integration represents either an implicit or an explicit prediction of how the candidate will perform' (Guion, 1965: 474;our emphasis), but also has observed that an advance weighting of 'the values that valid criterion measurement should reflect are rarely articulated and may not even be considered in making (candidate) choice.' (Guion, 1997: 276).Guion (1965) has been rare among the advocates of normative selection theory in claiming that interviewers should be concerned with the personality of candidates, whereas those favouring a normative approach claim that selectors are likely to be unduly influenced, or biased, by the personal characteristics or appearance of candidates during an interview rather than focussing on whether they meet explicit criteria for job-fit.Normative theory therefore claims that personality should be determined by pre-interview psychometric tests.By contrast, within selection literature, Ulrich andTrumbo, (1965), andThompson, Warhurst andCallaghan (2001) have claimed that assessing personality in interviewing is of paramount importance, while Robertson (1994) also has called for criteria concerning Personality in Context (Robertson, 1994), where the context is both the needs of the organisation as well as the needs of the job.Also, whereas normative methodology in selection presumes that explicit criteria have general validity, it may be that such criteria, like any proposition or expression, are meaningful only in the context of their use (Wittgenstein, 1953;Mintzberg, 2004).Yet, in such a context, managers as selectors also face the challenge of appointing to jobs which are becoming less clear cut, change rapidly and whose criteria may be difficult or impossible to know in advance (Bridges, 1996;Fletcher, 1997;Colenso, 2000).This reinforces the case, claimed by Schein (1980Schein ( , 1992Schein ( , 1994) ) and Schneider (1994), that mainstream models of criteria choice and predictive validity in selection need to be re-thought.As Fletcher (1997) already put the issue a decade ago, 'the organizational rules are changed so quickly and so often that the job analysis is outdated before the selection methods are in place' (Fletcher, 1997: 9).If this is the case, exclusive insistence on normative ex ante prioritisation of specific criteria for selection may be misplaced, while it may be mistaken to neglect qualitative judgement derived by managers from recent and current experience, at whatever tacit level.
A range of researchers in organizational decision-making (e.g., Cowan, 1991) and in selection (e.g., Altink, Visser and Castelijns, 1997) have claimed that organizational and operational culture will tend to affect the choice and application of criteria for selection.Yet, according to Tsoukas and Vladimirou (2001), such corporate culture is implicit rather than explicit.Like corporate 'ethos' (Schneider, 1994;Pichault and Schoenaers, 2003), it tends to 'underlie' corporate or group experience.It also may be that such culture will imply common rationalities on how things are to be done or should be done, or norms that are implicit in custom and practice rather than overt and 'up front' such as codes of practice or mission statements (Oliveira, 2002a).In the context of what counts in personnel selection, this suggests the research question: RQ1: Can analysis of managers' discourse reveal common rationalities that are implicit in their choice of selection criteria?If it were found that such rationalities are implicit rather than explicit, this could be consistent with the conceptual framework of tacit knowledge and implicit learning.Before WW2, Thorndike and Rock (1934)

Implicit Logic in Tacit Knowledge
without conscious intent to learn.The physicist Michael Polanyi then developed the concept of tacit knowledge (Polanyi, 1958;1962) and in so doing suggested that there was a 'tacit coefficient' which the mind correlated previous tacit knowledge.Reber (1967;1993) has claimed that implicit learning is an unconscious process of abstraction by which we categorise new experience on the basis of patterns of past experience.Following Nonaka (1994Nonaka ( , 1998)), Baumard (1999) has stylised individual and collective 'knowledge modes' which are either explicit or tacit, claiming that individuals assimilate tacit knowledge and interiorise collective knowledge at a less than conscious level, and that implicit learning is a two way organisational process, from groups to individuals and from individuals to groups (Baumard, ibid.).Herriot (2003), with Klimoski and Brickner (1987), has suggested that the 'natural' predisposition of managers in interviewing is to use 'implicit constructs' based in organizational culture.Ashforth and Johnson (2001) claim that a workgroup and its occupational identity will tend to be a sub-set of organizational identity.Ashforth and Johnson's (ibid.)claim is paralleled by the clinical analysis of Ignacio Matte Blanco (1975;1988) who has claimed that the unconscious mind is logical in terms similar to mathematical set theory, and correlates sets-within-sets of meaning for us without us having to follow how it does so.Bartlett's (1995) data driven analysis of recognition and remembering found that those participating in his experimental work tended to relate incoming cognition to 'schema' or 'schemata' of previously stored experience, and do so at varying levels of consciousness.Matte Blanco was unaware of Bartlett's work but his concept of sets-within-sets of stored experience and Bartlett's schema and schemata are identical, if inverse in their focus.Bartlett (1995) is concerned with how people relate the sense data of current experience to schema of previously organised experience, i.e. how people relate current to previous experience.Matte Blanco (1975;1988) is more concerned with how sets of previously stored experience unconsciously influence current perception.

-Implicit Logic
In terms of personnel selection, analysis of what is implicit in managers' discourse may illuminate otherwise casual reference to 'explicit' or 'implicit' in discussion of selection criteria.For instance, Dewberry (1998) has claimed that: 'The few available existing models of the relation between personality and job performance proposed, implicitly or explicitly, that there is a causal relationship' (Dewberry, 1998: 5).Similarly, Chan (1996) found that 'recruitment personnel are driven by questions (albeit sometimes implicitly) like will this person make a good organizational citizen?'In the same way, in looking at recruitment decisions, O'Reilly, Chatman and Caldwell (1991) found that 'recruiters made implicit judgements about a candidate on the basis of how congruent they perceived their values to be with those of the organization' (cit., Millward and Alexander, 1998).
Managers may explicitly justify selection criteria similarly or differently.If they are unwilling to rank them ordinally ex ante in an explicit manner, but it is found from ex post analysis that such a ranking is implicit in their discourse, this may reflect shared values and meanings (c.f., Schein, 1994) and a shared set of meanings and norms (c.f., Alvesson, 1996).Discourse analysis thereby should make it possible to explicate Klimoski andBrickner's (1987), andHerriot's (2003) suggestion that the 'natural' predisposition of managers as interviewers is to use 'implicit constructs' based in organizational culture, and Ashforth and Johnson's (2001) claim that a workgroup and its occupational identity will tend to be a sub-set of organizational identity.This would be consistent with Alvesson and Kärreman's (2001) stress on knowledge as a situated, community based set of meanings.If it were found to be the case, it should be evident in the manner in which managers identify selection criteria, and any such implicit rationale or 'implicit logic' (Oliveira, 2000;2001;2002a;2002b;2005) liable to ex post identification and coding or what Pichault and Schoenaers (2003) allow as coding or de-coding in context.Therefore, although mainstream normative selection theory presumes that only explicitly adopted criteria are rational, this presumption is not self-evident.They may be rational in the sense of inference and explicit logic but neglect a referential implicit logic by which managers are less than consciously drawing on schema (Bartlett, 1995) or sets, or sets-within-sets (Matte Blanco, 1975;1988) of previous experience.Also, in principle, discourse analysis should be able to identify the influence of such experience in managers' implicit prioritisation of criteria, and especially whether this prioritisation is influenced more by operational than organisational experience, or experience of interviewing, in what way, and in relation to which criteria.Therefore the research question: RQ2: Can analysis of managers' discourse on selection criteria reveal an implicit ranking of the importance in context of such criteria?Also, if Matte Blanco (1975Blanco ( , 1988) ) is correct in claiming that that the unconscious mind correlates sets-within-sets of meaning of previous experience, there is the further research question: RQ3: Are implicit priorities in managers' choice of selection criteria identifiable from sets-withinsets of meaning in their discourse?
Addressing such questions needs a methodology that can capture, classify and code both explicit and implicit meaning in managers' discourse.It is within such a conceptual framework that the methodology of the case study in this paper adopted a grounded theory approach within a socio-cognitive perspective.
-A Grounded Theory Approach As a method developed among others by Charmaz (1990;1994), Henwood andPidgeon (1995), grounded theory involves procedures such as interviews, observations and discourse analysis.It includes: (1) generation of low level categories, to interrelate the relevant features of the data corpus; (2) definitions and linkages between the categories at the different levels of abstraction; (3) comparisons between cases, instances and categories, in order to explore the complexities of a data domain, and (4) theoretical sampling of new cases, to seek data to support or disconfirm the emergent conceptual framework or theory.The case study applied this approach in generating, interrelating and comparing cases, concepts and categories.The newly developed coding system was designed to determine whether there was an implicit logic in managers' discourse concerning selection criteria.

-Social Construction of Discourse
A grounded theory approach to discourse analysis shares some features with the 'social construction' analysis of discourse (Derrida, 1978;Gergen, 1994Gergen, , 2000)).This has much to support it in claiming within an existentialist tradition that what matters is 'the self and the other', and that we construct our own perceptions of 'the other'.Davies andHarré (2001 [1990]) have reason to claim that discourse and life are not linear, and that people tend to organize dialogue in at least two modes: 'the 'logic' of the ostensible topic and the story lines which are embedded in fragments of the participants' autobiographies' (Davies and Harré, 2001: 264).Identifying implicit logic in such fragments within discourse and seeking to draw out their implications is central to the 'grounded theory' approach of this case study.
In this regard, therefore, we also support Bakhtin (1981Bakhtin ( [1935]]) in claiming that dialogue is fundamentally dialogic and that it creates meanings that previously 'were not there' : e.g. an interviewer may gain an insight into what is rhetoric in top managers' claims that they run a 'learning organization' while, through discourse, a middle manager may come to realise that he or she is not alone in having thought, at varying levels of consciousness, that something both 'should be done' without articulating and acting jointly on it with others.For such reasons we also therefore agree with Gergen (1994;2000) that the more capable people are in constructing and reconstructing narratives of themselves, the more broadly capable they may be in both thought and action.In a selection context, such discourse should be able to demonstrate whether Implicit Logic in Managerial Discourse: A Case Study in Choice of Selection Criteria Teresa Carla Oliveira

Methodology
candidates 'believe in themselves' in Riesman's (1954) sense of being 'self-directed', rather than are 'other directed' and likely to be dependent on direction or supervision by others.It also should be able to indicate whether they value the challenge of multi-tasking, multi-skilling, job variation and flexible methods of work operation (Womack, Roos and Jones, 1990) rather than routine job definition and task segmentation (Taylor, 1911).
-'Critical Realism' Yet there also can be problems within a social constructionist approach.As Reed (2005) stresses, some advocates of such a methodology insist not only that discourses but also organizations are discursive constructions and cultural forms that have no status or significance 'beyond their textually created and mediated existence' (Reed, 2005(Reed, : 1622)).This not only may intentionally go to extremes, such as Derrida's claiming that the (first) Iraq war 'did not happen' in the sense that there were different Iraq wars for different actors or observers.It also implies that no generalisations can be made about economic systems or modes of management that can be known to have a basis in reality rather than only as our construction of such reality.
In supporting a social constructionist rather than critical realist approach Contu andWilmott (2005: 1650) reasonably enough ask one of the longest standing questions of philosophy: 'How do you know the world is the way you say it is; and why should we believe you?'But it may be that analysis of one-to-one discourse with managers finds that they consistently perceive 'their reality' in the same or a closely similar way, which could imply that they are addressing the same reality rather than only a mutual social construction of such a reality.It also could be that if such managers independently identify and interpret 'their reality' in the same or similar ways, this would constitute an 'implicit verifier' of such a reality.
This study analyses the content and process of managers' identification and justification of criteria for selection of candidates for posts as TV operators in a European public broadcasting corporation.The tasks in question were not simply camera operation, but included also handling video, sound and lighting equipment, live studio and outside broadcasting, editing film or video tapes for entire programmes or programme inserts, graphic design and animation, staging studio programmes, selecting and assembling scenery, and selecting or storing archive material.The job therefore was less a job-in-itself, than multiple 'sets-within-sets' of job roles demanding multi-skilling, multi-tasking and flexible work practices of a kind which are typical of broadcasting but also increasingly demanded throughout the service sector, and in several front line sectors of manufacturing industry (c.f.Womack, Jones and Roos, 1990;Fletcher, 1997;Colenso, 2000;Womack and Jones, 2005).
The corporation offered 'laboratory' conditions for the research questions of this case study inasmuch as (1) although its overall selection process was modelled on that of the BBC (1996), with highly structured pre-interview screening, trainability and psychometric tests, managers as selectors still were free to adopt and deploy their own criteria for selection in giving information and asking it of candidates; (2) its managers were not trained in selection methods or interviewing techniques in the manner considered mandatory in normative selection theory; (3) the interview panels for selection included managers with a range of different operational and organisational experience.The selection procedures had proven highly successful over a period of some three decades in retaining selected personnel through their two year probationary period and in their later employment.

-Participants
The participants in the research interviews were (n=22) twenty two managers in the broadcasting company at different levels of organizational and operational responsibility, with differing degree of experience of the organisation, and seniority.Seniority was defined in the terms of practice in the organization.Thus senior managers were heads of departments and members of the board of the company.Middle managers were heads of units.Junior managers had narrower operational experience.
The managers had been within the organization for an average of fifteen years.Most of them (n = 16) had themselves been selected for and done the job of TV operators for which they were interviewing.Of the twenty two, 4 were senior managers (two of them with experience of having been a TV operator); 10 were middle managers (eight of them with experience of having been a TV operator), and 8 were junior managers (six of them with experience of having been a TV operator).Thirteen of the 22 either currently were involved in selection, including panel interviewing, or had been so.The other 9 either were managing or had managed TV operators.

Data Collection Procedures
Individual semi-structured interviews were carried out with the managers concerned.Permission to record the interview on audio-tape was given by the interviewee as the basis for later discourse analysis.The discourse analysed related to the following questions: • What in your view should be the selection criteria for TV Operators?
• How should suitable candidates identified and selected?
• What information is it important to give to and receive from the candidates during the selection process?
The research interviews with managers took place in their own offices on a one-to-one basis, and lasted on average 45 minutes.

Data Analysis and Coding Procedures
Data analysis was undertaken for all interviews on the basis of the classification system for coding managers' discourse.A provisional system was defined after listening to, transcribing and analysing five randomly chosen interviews.In order to check the reliability of the provisional system, five interviews were given to a senior independent assessor, who was asked to code the interviews on this basis.The assessor found the provisional coding robust in relation to the main individual categories, but recommended one modification: the introduction of 'context' as a separate code.This took account of the degree to which managers were concerned to stress the nature of the post in terms of changing operational and organizational needs.The initial interviews then were re-coded, and the rest of the interviews also coded on this basis.All the audio-taped interviews were transcribed verbatim.Responses to questions which related directly to the research topic -identifying and justifying selection criteria in terms of what managers needed from prospective TV operators -were then analysed and coded.
The selection of three mutually exclusive categories for the content analysis was both 'data driven' and 'theory driven'.It was 'data driven' in the sense that the codes of the category in relation to identifying and justifying the choice of selection criteria (Category 1) were grounded in the managers' discourse.It was 'theory driven' in the sense that the codes for the other categories (Categories 2-3) were derived from the conceptual framework of Bartlett's (1932Bartlett's ( , 1995) ) schemata and Matte Blanco's (1975Blanco's ( , 1998) ) sets-within-sets of meaning, while the ex post discourse analysis was concerned to understand the meaning of managers' statements and answers in the context (Wittgenstein, 1953) both of the interview with them, and their implicit prioritisation of the same criteria in different operational or organizational contexts.None of the managers was prepared to give an explicit ranking of the importance to them of specific criteria rather than affirm that something was 'vital', 'important', 'really important' or 'must be taken into account'.
The three categories used in coding analysis of managers' discourse in the research interviews are summarised below.Each of the categories includes sets and sub-sets of meaning.Each category also is complementary to the others, and therefore may interrelate different meanings in the sense of Polanyi's (1958Polanyi's ( , 1962) ) 'tacit coefficient' which, in turn, may be made explicit by the discourse analysis.But within each category the codes are mutually exclusive.It is in this sense that each statement or sequence of meaning in discourse is coded in relation to the three main categories of the coding system.
Category 1: Specific criteria identifiable from ex post discourse analysis.
Within this category explicit or implicit use of criteria by managers is identified in terms of eight different codes.
(1)Values: moral, ethical, aesthetic principles or dispositions underlying and influencing behaviour; (2) Beliefs: criteria concerning something believed in and conveying conviction; (3) Personality: the character and characteristics of a person, such as motivation, commitment, readiness to learn, adaptability, flexibility, aptitude for team-working, sociability.
(4) General Knowledge: criteria concerning knowledge that is not specifically related to job-fit, but is considered necessary to perform the job well.
(5) Specific Knowledge: criteria concerning knowledge that relates directly to job-fit, and is necessary to perform that job.
(6) Abilities: criteria concerning the capacities and capabilities to be able to perform given tasks.
(7) Skills: being able to think and act creatively above a given level of competence or ability, and including cognitive, social and life skills rather than simply technical job capability.
(8) Context: The lifestyle, conditions and working environment of the job, including the nature and changing context of multiple job roles; objective conditions rather than subjective attributes of candidates.
Category 2: How criteria relate to sets concerning Person-Organisation Fit or Person-Job Fit.
Within this category, managers' discourse is analysed in relation to two different codes: (1) Person-Organisation Fit regards criteria concerning the candidate's understanding of and potential to fulfil a range of roles within the organisation or part of the organisation which serve, enhance or safeguard its mission.These requirements are based on managers' discourse concerning organisational needs.
(2) Person-Job Fit regards criteria concerning the candidate's understanding and/or potential to perform a particular task.These requirements are based on managers' discourse concerning operational needs.
Category 3: How the discourse relates to sets of criteria such as Values, Beliefs and Personality (VBP) or Knowledge, Abilities and Skills (KAS).
The first three codes above within Category 1 form a set of Values, Beliefs and Personality (VBP).Codes 4 through 7 form another set of Knowledge, Abilities and Skills (KAS), including the further sub-set distinguishing general from specific knowledge.The use of Knowledge, Abilities and Skills or KAS inverts the more familiar sequence in selection literature of Knowledge, Skills and Abilities on the grounds that progress normally is not from knowledge through skill to ability but from knowledge through ability to skill, i.e. knowledge in knowing something, ability in being able to do it, and skill in being able to do it well.In practice, skill does not come before ability.This inverted sequence informed the newly developed coding system for discourse analysis of Knowledge, Abilities and Skills in the case study.
Managers' discourse was analysed in terms of all 3 complementary categories rather than one category.Within each of the 3 categories, the codes are mutually exclusive.
In applying the coding system to examine the selection criteria explicit or implicit in managers' discourse, there was an overall 84% (Cramer's V=.825; p<.001) of inter-coder reliability between the initial coding of the researcher and that of the independent assessor after mutual agreement to introduce the category of Context.Both the researcher and assessor independently recoded a further sample of interviews after the adoption of the criterion of Context in Category 1 of the Coding System.Inter-coder reliability for Category 1 was 81% agreement (Cramer's V=.826; p<.001); reliability for Category 2 was 86% agreement (Cramer's V=.822; p<.001); for Category 3 it was 87% agreement (Cramer's V=.838; p<.001).

Working Hypotheses
In addressing the research questions the case study undertook a content analysis of managersd iscourse in terms of the following research hypotheses.
H1 Discourse analysis may reveal common rationalities in managers' implicit prioritisation of specific criteria for candidate selection.
H1 would be supported by finding that managers implicitly prioritise the same criteria for selection despite their reluctance to rank criteria in order of importance in their research interview discourse.Finding such common rationalities would be consistent with tacit knowledge implicitly learned from mutually shared experience (Polanyi, 1958(Polanyi, , 1962;;Reber, 1967;Nonaka, 1994Nonaka, , 1998;;Baumard, 1999;Mintzberg, 2004).Consistency in such findings would suggest that there is a referential unconscious logic (Matte Blanco, 1975Blanco, , 1978) ) in their prioritisation of criteria and thereby challenge the claim of normative theory (e.g.Guion, 1965Guion, , 1997;;Dipboye, 1994Dipboye, , 1997;;Anderson, 1997;Anderson andShackleton,1990, 1993) that only the inferential adoption of explicit criteria can assure rationality in selection decision-making.Common rationalities would enhance the case of Reed (2005) that discourse is not only a 'social construction' (Gergen, 1994(Gergen, , 2000;;Contu and Wilmott, 2005) but relates to a shared reality underlying the implicit derivation by managers of the same criteria from similar experience.
It also may be found that the individual criteria commonly prioritised by managers are the personality and skills needed from candidates for their job roles in organisational and operational context.Prioritising personality would be consistent with the claims of Ulrich and Trumbo (1965) that assessing personality in interviewing is of paramount importance, while prioritising personality and context would be consistent with what Robertson calls the need for criteria concerning 'personality in context' (Robertson, 1994).Such a finding would contest the counter claim of normative theory that selectors should prioritise candidate's knowledge, abilities and skills in relation to their suitability for job-fit.A high degree of concern for both context and skills would indicate support for Fletcher's (1997) claim that selectors increasingly need to be concerned with candidates' skill in adapting to multiple job roles and more flexible organisational and operational needs.
H2 Discourse analysis may reveal that managers implicitly prioritise sets of criteria for candidate selection.
H2 would be supported by finding from discourse analysis that managers implicitly prioritise interrelated criteria and sets of criteria for selection.This could confirm the claim of Bartlett (1995) that the mind at varying levels of consciousness relates current cognition to schemata of previously stored experience.It also would be consistent with Matte Blanco's (1975Blanco's ( , 1988) ) claim that the mind interrelates sets (schemata) of previously stored experience which unconsciously influence current perception.For instance, it may be found in terms of Category 2 that managers implicitly prioritise sets of criteria concerning Person-Organisation Fit rather than Person-Job Fit and, in terms of Category 3, do the same for sets of criteria which prioritise Values, Beliefs and Personality (VBP) rather than Knowledge, Ability and Skills (KAS).In which case, the higher implicit ordinal ranking of criteria concerning VBP than KAS would suggest that, even for a highly technical job, managers are more concerned not only with individual criteria (H1) but also wider ranging sets of such criteria concerning who people are in terms of their values, beliefs and personality, rather than only what they may be able to do in terms of their current knowledge, abilities and skills.
H3 Managers' discourse concerning Values, Beliefs and Personality (VBP) may correspond with sets of criteria concerning Person-Organisation Fit, and their discourse concerning Knowledge, Ability and Skills (KAS) with sets of criteria concerning Person-Job Fit.
If H3 is supported, it would further support Matte Blanco's (1975;1988) claim that the unconscious mind interrelates sets-within-sets of meaning in a logical manner, which also is consistent with the claim of Polanyi (1958;1962) that there is a 'tacit coefficient' within knowing and understanding that may interrelate present and past knowledge without our being aware of it.It also would demonstrate support for the claims of Charmaz (1990;1994), andHenwood andPidgeon (1995), that grounded theory not only (i) can generate low level categories, and (ii) interrelate the relevant features of the data corpus, but do so (iii) with linkages between categories at their different levels of abstraction.
H4 Analysis of implicit logic within managers' discourse may demonstrate that their common rationalities concerning criteria for selection derive from their experience of the organisation and of the job rather than from experience of interviewing.
If H4 is supported in the context that the managers concerned had achieved a high success rate in selecting appropriate candidates, it would contest the claim of 'normative' selection theory (e.g.Guion, 1965Guion, , 1997;;Dipboye, 1994Dipboye, , 1997;;Anderson, 1997;Anderson andShackleton, 1990, 1993) that managers should be trained to interview and suggest that the expertise needed for successful selection is less in general interviewing technique than based in and implicitly derived from the grounded reality (Charmaz, 1990(Charmaz, , 1994;;Henwood and Pidgeon, 1995) of their experience of the organisation and of the job.
A total of 513 statements and sequences of discourse were extracted and coded from the 22 research interview transcripts.Where the data permitted, a Chi-Square test was performed.Moreover, cross-tabulated data for differences in (a) seniority, (b) experience of the organisation and of the job and (c) experience of selection and interviewing was also analysed to permit a higher degree of disaggregation of linkages between categories and codes.
H1 Discourse analysis may reveal common rationalities in managers' implicit prioritisation of specific criteria for candidate selection.
For H1 there was an overall highly significant difference (Chi-Square=329.5, p<.001) between the different criteria explicitly or implicitly identified by all managers, with clear priority for Personality, Context and Skills, in that order.
Example of Personality: 'TV operators not only need to be able to learn, but also be able to act.They need to be people with a sense of show and for the spectacular.' Junho '06 / (53/71) 62 63 1 Taking into account that the nature of the data for this paper is categorical, the statistical analysis is based on non-parametric tests.The statistical test used is Chi-Square (c.f.Bryman and Cramer, 1997).On the few occasions where the data is not Chi-Square tested, this is because the frequencies in the cells had an expected count of less than 5.

Findings 1
Example of Context: 'It is vital when we select people to describe the context of what they will do.They need to know what sort of company they are coming to work for.It is important to tell them that they need to travel a lot, they need to work during week-ends and evenings, and we need to find out their availability for such a kind of work.' Example of Skill: 'A TV operator needs a basic training to be able to perform a specific function which has a technical component of some kind.But, more than technical ability, he also needs creative skills.This is important, and the selection process needs to reflect this.' Within the limits of the sample, the results support H1.
H2 Discourse analysis may reveal that managers implicitly prioritise sets of criteria for candidate selection.
For H2 nearly three-quarters (73.5%) of the coded discourse of managers concerned sets of criteria in terms of VBP or KAS (of 376 statements 216 concerned VBP).It was found that managers in general implicitly prioritised criteria concerning VBP rather than KAS (Chi-Square=8.340;p<.01).
Of the 513 identified selection criteria in terms of P-Org Fit and P-Job Fit, 396 concerned P-Org Fit and that, therefore, managers in general are more concerned with criteria concerning P-Org Fit than P-Job Fit (Chi-Square=151.737;p<.001).

Example of Person-Organisation Fit:
'If I want to select TV operators, I'm not going to give tests to see if the candidate would be good as a producer.This is obvious.But now, in a company needing to rejuvenate itself and move, I believe that it is a tremendous mistake to select an operator on the basis that he is going to do one job for ever.'

Example of Person-Job Fit:
'There is a need for continual job up-dating without effort.' Within the limits of the sample, the results support H2.
H3 Managers' discourse concerning Values, Beliefs and Personality (VBP) may correspond with sets of criteria concerning Person-Organisation Fit, and their discourse concerning Knowledge, Ability and Skills (KAS) with sets of criteria concerning Person-Job Fit.
For H3, the finding that managers are more likely to associate VBPs with P-Org Fit and KAS with P-Job Fit is highly significant (Chi-Square=40.07;p<.001).Managers' are more concerned with sets of criteria relating to VBPs than KAS in relation to Person-Organisation Fit by a factor of 2:1.

Example of P-Org Fit in relation to VBP:
'We like to select people who are motivated to work in the TV business, and we take the selection process seriously because we want successful candidates to do well in the training course and stay with the company.We also need to remember that the training course takes a lot of time, is demanding in resource terms and expensive'.

Example of P-Job Fit in relation to KAS:
'A cameraman directly influences the product with his own initiative and decision about which image the producer wants, and even may suggest the image.Therefore this operator has a direct influence on the outcome.He fulfils himself as an operator through his own capacity and creativity for the benefit of the final product which is the programme'.
Within the limits of the sample the findings very clearly support H3.
H4 Analysis of implicit logic within managers' discourse may demonstrate that their common rationalities concerning criteria for selection derive from their experience of the organisation and of the job rather than from experience of interviewing.
For H4, Chi-Square tests also show that there is no significant difference in the implicit prioritisation of specific criteria (Category 1) that can be attributed to different seniority, experience of the organisation and of the job and whether or not managers have been engaged in interviewing.This implies that there are common rationalities between managers in their prioritisation of the criteria which are important for selection.However, there are differing rationalities in terms of the degree to which managers assign relative importance to specific criteria, which can be attributed to whether or not they have experience of the job for which candidates are being interviewed.
Chi-Square tests suggest that there is no significant difference in terms of implicit prioritisation of VBP rather than KAS (Category 3) in relation to managers' seniority, whether or not they have done the job in question, and whether or not they have interviewed and selected candidates for the job.This indicates a common rationality between managers in their implicit prioritisation of VBP as more important than KAS in the sets of criteria which are important for selection.However, as shown in Table 1, disaggregating the sets of criteria by cross tabulation reveals that there are differences in the degree to which managers implicitly assign relative importance to VBP rather than KAS, and that this varies in relation to different levels of seniority.
Table 1 reveals a descending order of importance for implicit ranking of VBP of 13.5 for senior managers, to less than 11 for middle managers, and 6.5 for junior managers.A similar descending scale obtains for senior managers whose implicit ranking of the importance of KAS is 10.5 whereas for middle managers it is just over 8 and for junior managers only 4.
Table 2 illustrates the cross-tabulation of managers' discourse in terms of VBP and KAS with their different experience of selection and interviewing.Chi-Square tests for Table 2 show that there is no significant difference in their discourse in terms of their preference for VBPs rather than KAS (Category 3) which can be attributed to whether or not they have been engaged in interviewing.
This implies that there are common rationalities between managers with and without experience of interviewing in their identification of VBPs as more important than KAS of the criteria which are important for selection.
Table 3 illustrates the cross-tabulation of managers' discourse in terms of P-Org and P-Job Fit with their different degrees of seniority.
A Chi-Square Test shows that there is a highly significant difference (Chi-Square=24.1;p<.001: Cramer's V=.217; p<.001).Averaging the sets of criteria by managers' seniority, gives the following ratios of their higher preference for criteria in terms of P-Org Fit than P-Job Fit: senior managers: over 6:1; middle managers 4:1; junior managers well under 2:1, i.e., again, a consistent and marked descending scale.Chi-Square tests show that there is no significant difference between managers' justification of criteria concerning P-Org and P-Job Fit and whether or not they have been engaged in interviewing.
Within the limits of the sample, the findings very clearly support H4.

Discussion of Findings
The findings support the case that while there may be an explicit logic in the ex ante adoption of selection criteria which is conscious, inferential and overt, there also may be an implicit logic in the manner in which managers less than consciously draw on tacit knowledge derived from experience of the organisation and the job in prioritising selection criteria, and that this is of more importance than whether or not they have previous experience of interviewing.The findings also support the case that implicit logic is wider ranging than inference alone in interrelating multiple criteria and sets of criteria in varying operational and organisational contexts, and thereby suggest that when managers 'think that they know' what they need from candidates, yet cannot explicitly rate or rank it, they have multiple reasons at varying levels of consciousness for knowing well what they need and prioritising it in their discourse.
The evidence in these regards supports the conceptual framework of Bartlett's (1932) analysis of the processing of memory, and Matte Blanco's (1975Blanco's ( , 1988) ) claims for an 'unconscious logic' interrelating past and resent experience.It is consistent with the claims for the pervasiveness of tacit knowledge of Polanyi (1962), Baumard (1999), andAmbrosini andBowman (2001), and supports Reber's (1967;1993) principle of implicit learning.By contrast with the reluctance of managers as selectors to rank the importance of criteria ex ante in the manner recommended by normative selection theory, or during discourse, the case study demonstrates that ex post coding and classification of their discourse can do so (Oliveira, 2000;2001;2002a;2002b;2005).It also confirms the claims of Charmaz (1990;1994), Henwood andPidgeon (1995), concerning the scope of a grounded theory method, and the viability of the newly developed coding system.
The implicit logic in managers' ranking of criteria reveals a consistency in the manner in which, even for a job demanding a high level of technical qualification, they are more concerned with sets of criteria concerning Values, Beliefs and Personality than Knowledge, Abilities and Skills, and with Person-Organisation Fit than with Person-Job Fit.There is a constant stress in their discourse on the need for personality, creativity, imagination, and capacity to adapt to changing organisational needs, rather than only concern with current job competence.
Granted that none of the managers in the research sample had been formally trained in selection techniques, the findings indicate that such common rationalities were derived from implicit learning (Reber, 1993) and tacit knowledge (Polanyi, 1958;1962) which they had gained from operational and organizational experience.This supports what Baumard (1999) has stylised as individual and collective 'knowledge modes' by which individuals interiorise collective knowledge at a tacit level.The consistency with which they tacitly share such knowledge modes is reflected in their implicit prioritisation of individual criteria and sets of criteria.It tends to confirm Alvesson and Kärreman's (2001) case that knowledge constitutes a 'situated, community based set of meanings' and the claim of Tsoukas and Vladimirou (2001) that the 'givens' of corporate practice 'underlie' group experience.They also tend to support the presumption of Herriot (2003), with Klimoski and Brickner (1987), that the 'natural' predisposition of managers in interviewing is to use 'implicit constructs' based in organizational culture, and Ashforth and Johnson's (2001) claim that a workgroup and its occupational identity will tend to be a sub-set of organizational identity.
The evidence justifies the claims of Guion (1965), Ulrich and Trumbo (1965), Robertson (1994) and Thompson, Warhurst and Callaghan (2001) concerning the importance of personality and skills in organizational context.It also supports the findings of Cable and Judge (1997) and Schneider, Kristof-Brown, Goldstein and Smith (1997) that managers are more concerned with sets of criteria concerning Person-Organisation Fit rather than with Person-Job Fit.That managers develop criteria with concern for both operational and organizational context is consistent with the views of Pichault and Schoenaers (2003).It supports the view of Altink, Visser and Castelijns (1997) that criterion development competencies and achievements are communicated through the organization and the claim of Schein (1980;1992) and Schneider, Kristof-Brown, Goldstein and Brent Smith (1997) that efficient organizations select staff to fit the company culture and ethos.
-Adapting to Change The findings illustrate the case expressed by Bridges (1996) and Fletcher (1997) that jobs are becoming less clear cut and changing rapidly with the implication that managers increasingly may be concerned more with Person-Organisation Fit than with Person-Job Fit .This also may be a reason why managers are reluctant to take time to pre-define selection criteria explicitly in the manner recommended by Anderson (1997), Dipboye (1994;1997) and others.This is suggested by managers' statements such as 'Today we have one job.Next year we have others.And if we stop adapting, 'goodbye''.And, 'There is a need for continual job up-dating without effort.'The presumption that job up-dating needs be done 'without effort' suggests also that managers assume and need implicit learning (Reber, 1967;1993) and that such learning, whether less than conscious, or unconscious, is taking place according to an implicit operational rationale or logic.
This article suggests that analysis of managers' discourse reveals common rationalities in their choice of selection criteria (RQ1); that there is an implicit logic in their ranking of the importance of such criteria (RQ2); that implicit priorities in managers' choice of selection criteria are identifiable from sets-within-sets of meaning in their discourse (RQ3).
Although the selection process in the organization in question was modelled on one of the most sophisticated in Europe (that of the BBC), the findings are based on one service organisation study.Its outcomes may or may not be replicated in other cultures.Nonetheless, some general implications are suggested by the results.
Overall, the findings tend to support Nonaka's (1994;1998) claims for interfacing tacit and explicit knowledge, and what Mintzberg (2004) claims should be the primacy of explicating tacit knowledge derived from actual experience.They indicate that Fincham and Rhodes' (2005) claim that tacit knowledge cannot be 'codified' and manipulated by logical reasoning is overdrawn, and suggest that this can be achieved by identifying an implicit logic within managers' discourse.
They support the principles of the social construction of discourse, but indicate that there are identifiable realities within such discourse (e.g.Reed, 2005), rather than only a social construction (Gergen, 1994;2000).In response to Contu andWilmott's (2005: 1650) question 'How do you know the world is the way you say it is; and why should we believe you?', the findings that managers consistently finds perceive 'their reality' in the same way, indicates an 'implicit verifier' of such a reality.
The implications are that a socio-cognitive analysis within a 'critical realist' (Reed, 2005) perspective can better identify and profile what really counts for managers at varying levels of consciousness than normative theory claiming what they ought to do according to explicit behaviour paradigms, and do so within the social context in which perceptions are mediated (Augoustinos and Walker, 1995).They tend to confirm that specific meaning needs to be contextualised in terms of the organizational and the personal agendas of the individuals involved (Harré and Gillett, 1994), such as front line operational managers.The coding system confirms the claim of Wittgenstein (1953) that words or expressions need to be understood in the specific context of their use and that such use and meaning may be identified within the implicit logic of managers' discourse.This tends to confirm related findings on implicit logic in discourse in actual interviewing of candidates (Oliveira, 2000;2001;2002a;2005) and that identifying this may be relevant to assessment of whether or not there has been procedural justice in selection (Oliveira, 2002b).The methodology also may be found in further case studies to have wider relevance in the context in which there is increased attention to the claim that tacit knowledge is a key factor in competitive advantage of companies.
Implicit Logic in Managerial Discourse: A Case Study in Choice of Selection Criteria Teresa Carla Oliveira

2. Normative Selection Theory
had identified the process of learning without awareness of how we have learned, or

Table 4
illustrates the cross-tabulation of managers' discourse in terms of P-Org and P-Job Fit with their different experience of selection and interviewing.