explaining greater test use for selection: the role of …...explaining greater test use for...

Post on 05-May-2020

7 Views

Category:

Documents

1 Downloads

Preview:

Click to see full reader

TRANSCRIPT

Explaining greater test use for selection:the role of HR professionals in a world ofexpanding regulation

Alison Wolf, King's College LondonAndrew Jenkins, Institute of Education, University of LondonHuman Resource Management Journal, Vol 16, no 2, 2006, pages 193-213

In recent decades there has been a well-documented increase in organisations' useof formal tests for selection purposes. This article investigates whether tests'technical qualities and predictive validity are the critical determinants of take-up, orwhether other organisational and contextual factors are more important. Casestudies of organisations were used to identify the factors driving change andWorkplace Employee Relations Survey data were analysed to determine theconsistency of the findings with national data. The single most important factorappears to be the regulatory environment, with tests used as a precautionarymeasure to protect selection decisions from challenge. The growing formalprofessionalisation of HR departments is also important, and the influence of bothfactors is consistent with current theories of organisational decision-making.Continued growth in test use is likely, but the implications for efficiency in selectionare less clear.Contact: Professor Alison Wolf, Department of Management, King's CollegeLondon, 150 Stamford Street, London SEI 9NH, UK. Email:alison.wolf@kcl.ac.uk

The productivity and efficiency of organisations have always depended heavilyon the quality of their workforce, or their 'human capital', and there is generalagreement that its importance relative to fixed capital is steadily increasing

(see, e.g., Becker, 1993; Elias and Scarbrough, 2004). The recruitment and selection ofan effective workforce consequently can be viewed as central to the success of anenterprise and a key function of HRM. For example, Schmitt and Chan define thegoals of strategic HRM as 'to acquire, deploy and allocate human resources in waysthat provide the organization with a competitive advantage' (1998: 239, italics ours).

Success in attracting appropriate, able and skilled employees sets upper limits onpotential performance, however good the management and development of anexisting workforce may be. Not surprisingly, therefore, recruitment and selection arethe subject of a voluminous literature. The development and advance of psychologyas a discipline has been strongly associated with careful evaluation of the validityand utility of different selection tools in identifying the 'best' candidates for aposition (Schmidt and Hunter, 1998; Wood and Payne, 1998).

However, while such analyses of specific selection methods may be extremelyhelpful to managers or HR professionals, they do not, in themselves, explain fullyhow selection decisions are made, or why practices and procedures change over

HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 16 NO 2, 2006 193

© 2006 The Authors.Journal compilation © 2006 Blackwell Publishing Ltd., 9600 Garsington Road, Oxford, 0X4

2D0, UK and 350 Main St, Maiden, MA, 02148, USA.

Explaining greater test use for selection

time. We would agree with, for example, Johns (1993), Ramsay and Scholarios (1999)and Muchinsky (2004) that, taken as a whole, the selection literature underestimatesthe importance of both the locus of organisational decision-making and the politicaland regulatory environment within which organisations operate. This articleexamines the importance of such factors through a study of companies' increasinguse of formal selection tests. We use a combination of qualitative data fromorganisational case studies and quantitative data from the 1998 Workplace EmployeeRelations Survey (WERS 98), and ask what factors are important in explainingchanges in test use and whether these are likely to promote organisationaleffectiveness.

In adopting this focus we do not discount the importance of more conventionalconcerns. Selectors need to be aware of how well particular methods predict futureperformance (their validity) and of different methods' cost, relative to the increasedaccuracy they bring. The cumulative findings of selection research on validity havebeen examined in a number of important recent meta-analyses.

Schmidt and Hunter (1998) summarise the findings of 85 years of personnelpsychology research and conclude that, while general mental ability is the single bestpredictor of future performance, combining it with an integrity test, a work-sampletest and/or a structured interview adds significantly to validity. Schmitt and Chan(1998) reach similar conclusions; Robertson and Downs (1989) summarise evidenceon the validity of work-sample and trainability tests specifically, and conclude thatthey are generally effective. Assessment centres predict promotion in managementwith considerable, albeit variable, accuracy. (Gaugler et al. (1987) report a meanvalidity of 0.36 and Dayan et al. (2002) provide a useful up-to-date summary of theevidence.)

The use of personality tests for selection is highly controversial, and has beenattacked on the grounds of both fairness and utility (Mischel, 1968; Blinkhorn andJohnson, 1990; Paul, 2004). Earlier meta-analyses (see, e.g., Schmitt et al, 1984) tendedto conclude that they are clearly less effective than assessment centres, work samples,and cognitive ability tests in predicting performance at work. However, Robertsonand Kinder examine the validity of a range of personality variables in predictingvarious different aspects of job competence and, while agreeing that there 'are likelyto be some circumstances in which personality scales will not show validity',conclude that for some criterion variables - notably creativity, analysis andjudgement - 'personality scales produce practically useful criterion-related validity'of up to 0.33 and can add information to that provided by measured tests ofcognitive ability (1993: 240). Bartram (2004: 238) argues that 'academic literature tosupport the use of personality assessment for selection' has been a majordevelopment of the last decade.

The growing consensus around formal testing means that an informed use offormal measurement is seen as a major indicator of good HR practice in selection.For example, in their analysis of HRM and performance in a large sample of UKcompanies. Guest et al (2003) use test administration as one of their two majorindicators of HR practice in the area of recruitment and selection.' Selection test useis also identified as one component of a 'high-performance' HR model in the sameteam's analysis of data from the WERS 98 (Guest et al, 2000: 17). We return to thedata from WERS below as part of our own analysis.^

194 HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 16 NO 2, 2006

© 2006 The Authors.Journal compilation © 2006 Blackwell Publishing Ltd.

Alison Wolf and Andrew Jenkins

Practitioner-oriented literature treats test use as a valuable contribution to aselection process which is generally presented as self-contained and straightforward.An introductory textbook on HRM will typically describe how managers compareapplication forms, references and other selection data, including test results, with aclearly specified personnel specification, rank the applicants, and then make joboffers to those best qualified for the task (see, e.g., Graham and Bennett, 1995:183-186). More sophisticated HR textbooks (e.g. Wood and Payne, 1998; Torringtonet al, 2002) similarly treat the selection process as a largely technical, apolitical affair.This approach is, in fact, consistent with the dominant research paradigm (Schmittand Chan, 1998).

However, a number of researchers, including Becker (1989), Johns (1993), Guion(1998) and Muchinsky (2004), have argued that the focus on technical issues neglectscritical dimensions of the selection process. Becker (1989) has criticised researcherswho evaluate the utility and predictive power of different selection methods asthough they were completely context-independent. Similarly, Johns (1993) criticisesindustrial and organisational psychologists for viewing selection tools purely astechnical improvements that would be adopted by any rational manager whounderstood them. It is more useful, he argues, to analyse which tecltniquesare favoured (or rejected) exactly as one would other organisational decisions orinnovations, notably by examining the decision-making frame of managers,including both power relationships within the organisation and its externalenvironment. Guion (1998) emphasises that use (or non-use) of selection tests inpractice has rather little to do with their psychometric properties, and Muchinsky(2004: 176) is of the view that 'Time and money are far more salient organizationalcriteria in decision making than reliability and validity'.

In summary, there is good reason to associate test use with effective selectionpractice. However, the current weighting of selection research towards technical testissues has been criticised for neglecting important influences on how selectionactually takes place. This may in turn produce an incomplete understanding of howto conduct selection effectively. Our research seeks to increase such understanding bylooking explicitly at a range of possible explanatory factors for test use - technical,organisational and contextual.

THE RESEARCH

Since 1980, there has been a well-documented trend towards greater use, in the UK,of testing for selection, including, quite specifically, increased use of psychometrictests (e.g. Shackleton and Newell, 1991; Mabey, 1992; Ryan et al, 1999). These may bedefined as tests which measure a psychological characteristic or construct, whichmay be cognitive, such as general mental ability or quantitative reasoning, or involvebehavioural or personality traits. As Bartram notes (Bartram, 2004: 242) almost all thesurveys of test use are concerned with large organisations. (Bartram et al. (1995) isunusual in investigating small companies.) They mostly rely on postal surveys, andsay little or nothing about why differences in practice occur. The surveys areconsistent, however, in the pattern they find, one borne out by test publishers' ownsales data. Work-sample or competency tests are also widespread, and their use alsoseems to have increased (Jenkins, 2001).

HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 16 NO 2, 2006 195

© 2006 The Authors.Journal compilation © 2006 Blackwell Publishing Ltd.

Explaining greater test use for selection

While the evidence for increasing levels of test use is compelling, the reliance oncross-sectional questionnaire data greatly reduces these studies' explanatory power(Bartram, 2004). To understand the change in selection practices, it is important tofocus on why an organisation might decide to introduce, or expand, test use. Basedon both the selection and the wider HR and organisational literature, wehypothesised that several factors might be important.

One possibility is that selectors' views on the most valid and reliable indicatorsof candidates' future productivity have changed for what may be termed technicalreasons. Tests' real or perceived utility as predictive instruments may have increasedin absolute terms because they are more accurate or more cost-effective.Alternatively, selectors' confidence in other measures may have diminished. TheQualification and Curriculum Authority, which regulates qualifications offered inEngland's public education and training system, was very concerned, in the late1990s, that public confidence in qualifications was falling and that tests weretherefore being used as a substitute (Murphy, 2000, and personal communication).

Change may also be a result of organisational factors. There is widespreaddiscussion in the HR literature of how far HR personnel have become or are likelyto become holders of strategic decision-making positions in organisations, but alsogeneral agreement on the growing professionalisation of the field (see, e.g., Legge,1995; Caldwell, 2001). HR personnel have an interest in increasing their involvementin key decisions, which include selection choices. Growing awareness of the potentialcontribution of HR practices to business strategy (Guest and King, 2004) mayincrease organisational receptiveness to HR involvement and this may in turnincrease test use.

Third, during the 1990s, the context for selection decisions changed because ofa steady increase in the regulatory activity of the state, including a number ofmeasures with direct implications for selection and recruitment, such as newformalised equal opportitnity requirements. These developments appear in turn tohave created significant changes in organisations' internal culture, notably a growthin risk aversion (Power, 1997, 2004; Hood et al, 1999; Moran, 2003). Such legal andregulatory factors may have a significant impact on selection practices.

Study design

In exploring what factors had driven change, we relied primarily on case studies oforganisations that had recently changed their level of test use. These case studieswere designed to test hypotheses drawn from the literature, as discussed above.Unlike cross-sectional survey data, case studies are well suited to clarifying causalmechanisms. However, they have corresponding disadvantages. The number of casestends to be small and it is difficult to know whether the sample is systematicallybiased in key respects.

We therefore employed survey data as an independent check on our results, usingthe very large data set provided by the WERS 98. This allowed us to introducecontrols for the large number of variables which the literature suggests tnay affectselection practices, and specifically test use. While WERS is cross-sectional,consistency between the two forms of data would tend to support the findings of thecase studies.

196 HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 16 NO 2, 2006

© 2006 The Authors.Journal compilation © 2006 Blackwell Publishing Ltd.

Alison Wolf and Andrew Jenkins

The case-Study sample Data were collected during the period 2001-2002 from asample of organisations approximately three-quarters of which were contactedthrough the Chartered Institute of Personnel and Development's Recruitment Forum(which has approximately 1,500 members who are professionally involved inrecruitment). The remainder were identified through personal contacts.

We sought organisations where there had been substantial changes in test use(whether an increase or a decrease), and especially in usage of commerciallyprovided tests. Given the lack of relevant population data, we could not draw asample of companies that had changed test use. Previous research indicated that testuse was much more prevalent in large companies. This is reflected in our sample, butmay have been further reinforced by the use of Chartered Institute of Personnel andDevelopment channels. Our respondents were also, by definition, people with aninterest in research. However, they did not, on their own, make their organisations'policies and we have no reason to suspect that this introduced any systematic biasinto the sample.

Information on test use and its recent evolution was obtained through interviews.Respondents all had direct responsibilities for the organisation and conduct ofrecruitment and selection and were asked to describe, in detail, what tests were used,when, and by whom, and how this had changed. Interviews were semi-structured.We obtained respondents' own interpretation, and then probed with respect tovariables and possible explanatory mechanisms which had not been mentioned. Werequested information on changes over as much of the last decade as possible, butover half the respondents had worked with the organisation for less than 10 yearsand no organisations appeared to have kept documentation stretching that far back.We therefore had to rely on individual knowledge, which, in 85 per cent of cases, wasfelt by the respondents to be reliable for at least the last five years.

Data were collected for 53 organisations: 36 private sector companies, 12 publicsector or non-profit organisations^ and five organisations which were eitherpublic-private collaborations or recently privatised companies in which practicesstill reflected (in part) their public sector inheritance. Two private sector companiesemployed between 100 and 200 people; all others were larger, and some public sectorworkforces numbered well over 10,000.

All but eight of the private sector companies made use of outside firms andconsultants for either some or a great deal of their recruitment and selection activity;only two public sector organisations did so. We conducted interviews with nineoutside consultancy organisations who provided such services. Some consultantscarried out all the relevant work for several companies, and provided (separate)information on each.** Our results are therefore presented in terms of organisationalpractice, not in terms of interviewee.

Survey data The case-study findings, as noted above, were examined for theirconsistency with WERS 98 data. WERS 98 is a large government-sponsored surveyof public and private sector workplaces in Great Britain, which includes informationfrom interviews with the manager at each workplace most responsible for personnelmatters. The survey obtained an excellent response rate of 80 per cent on themanagement questionnaire and contains data on 2,191 British workplaces. Whenappropriately weighted it is nationally representative of workplaces with 10 or more

HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 16 NO 2, 2006 197

© 2006 The Authors.Journal compilation © 2006 Blackwell Publishing Ltd.

Explaining greater test use for selection

employees within Standard Industrial Classification major groups D to G.' In 1998it included questions on use of personality and competency tests. Unfortunately,questions on test use were introduced only in the 1998 survey so no panel data areavailable.

CASE-STUDY FINDINGS '

Patterns of test use

All the organisations were currently using tests for selection but patterns differedsomewhat by ownership type. In the private sector, all of the 36 companies studiedhad increased test use within the period studied. We were unable to locate anyprivate sector organisations that had significantly reduced testing in the last fiveyears. All but four of the private sector companies used both types of psychometricinstrument. In two private companies, the only psychometric tests used werepersonality ones, though both also used work-sample tests. Two companies usedaptitude/ability tests, combined with skill tests, but no personality tests. Just undera quarter reported increased personality testing but not increases in use of othertypes; just over 10 per cent had increased other forms of testing but not personalitytesting; and the remainder (i.e. two-thirds) had increased their use of bothpersonality and other forms of testing.

Among the 12 public sector and not-for-profit organisations, and using the samefive-year reference point, nine had increased use. Three had maintained a fairlyconstant (though substantial) level over the period and noted that, at some pointpreviously, there had been a move to substantially greater test use.

Organisations which were privatised (successors to public sector organisations) orpublic-private partnerships showed the most mixed pattern. All had changed theirpractices in the last five years, all reported high levels of use, but two reported lowerlevels than during (parts of) their public sector existence. However, the number oforganisations in this third category was small (n = 5).''

The vast majority of organisations (over 90 per cent of the total sample) used sometests that require users to be trained and licensed. Tests published by SHL Group picdominate this market in both personality and aptitude/ability testing. A few usedother commercially produced psychometric tests (e.g. Psytech) which are availablefor immediate use without such training. Just under two-thirds of organisations usedsome form of skill or competency test, which might be commercially purchased ordesigned in-house. As with the licensed psychometric tests, which test or tests wereused would depend on the type of vacancy.̂ Examples included literacy andnumeracy tests, clerical batteries, group exercises, in-tray exercises or otherwork-sample exercises for use as part of an assessment centre.

Licensed users were invariably found in personnel or HR departments and thetests were kept there (usually now, in electronic form). The number of individualstrained to interpret tests was generally very small, though more might be allowedto administer them. Only 11 organisations had an explicit policy of maintainingnumbers of licensed practitioners over time. Moreover, among those cases wheretesting (using licensed tests) was an entirely in-house concern, almost half theorganisations had only one person with the relevant full level A and B training (seefootnote 8). However, a substantial number of private sector companies, including

198 HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 16 NO 2, 2006

© 2006 The Authors.Journal compilation © 2006 Blackwell Publishing Ltd.

Alison Wolf and Andrew Jenkins

some very large ones, out-source a large part of recruitment activity, includingtesting. From the company's point of view, this ensures stable access to licensedprofessional staff.

Respondents' own categories did not always align with the standard definition ofwhat constitutes a psychometric or other type of test (see above). A clear distinctionwas generally drawn between commercially produced and home-grown tests, andalso, critically, between those which could be used freely and those requiring alicense and training for use." Personality tests were clearly separated from others.Conversely, 'ability' and 'aptitude' psychometric tests, on the one hand, andcommercially produced skill tests (e.g. of literacy or numeracy) were not alwaysdifferentiated. Our discussion nonetheless, as far as possible, distinguishes clearlybetween use of skill tests and ability/aptitude testing.

As noted above, we hypothesised that use of psychometric tests for selectionmight be increasing for one or more reasons: changes in the net value of test dataas indicators of future productivity (technical reasons); changes in the balance ofpower and influence within organisations (organisational reasons); and changes inthe external environment (legal and regulatory reasons). Respondents' ownexplanations for change could, it transpired, all be grouped within these majorcategories and are therefore discussed below using this structure.

Technical factors

Test data might become more desirable to organisations because such data areperceived to be more valid and reliable than in the past; or because the reliability andvalidity of other information are perceived to have fallen; or because the costs oftesting have fallen, providing more information for a given level of expenditure(Jenkins and Wolf, 2005). Of these, only the third appears to be of any importance.

The more popular tests marketed by the major test manufacturers have changedrather little since the late 1980s/early 1990s (personal communications with companypersorvnel) although individual items are regularly updated.' None of ourrespondents referred to changes in test quality, although several noted thatpreviously sceptical managers were now convinced of the accuracy and value ofpersonality testing.'"

The director of finance was the most sceptical and now he won't makeappointments without (personality) testing. (NHS Trust PersonnelManager)

Cross-sectional data indicate consistently that tests are used more often for sometypes of position than others (Jenkins and Wolf, 2002; Wilk and CappeUi, 2003).However, not one of our respondents mentioned substantive changes in theirorganisation's skill requirements as a reason for change. Nor were other sources ofinformation perceived to have 'lost' quality. Although we probed directly, there wasno indication of a general loss of faith in formal educational diplomas. Fourrespondents commented negatively on quality in GCSEs and/or degrees, but onlyone stated that selection policy had been directly affected by this, with a move togreater use of ability testing.'^

In contrast, changing relative costs - or, more accurately, perceptions of cost - doappear to have been important. All but one of the outside agencies in the sample

HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 16 NO 2, 2006 199

© 2006 The Authors.Journal compilation © 2006 Blackwell Publishing Ltd.

Explaining greater test use for selection

noted that test use had been 'sold' to companies as a way of reducing the expensivemanagerial time required for interviews. This reduction in internal company activity,and increased use of agencies for large-scale candidate testing, are only possiblebecause of the quite recent arrival of computerised marking and output, whichlowers turnaround time.

Organisational factors

While technical factors appear of limited importance, organisational issues havecertainly influenced the increasing uptake of formal, psychometric tests. As alreadymentioned, there is a large literature discussing the evolving role of HRprofessionals, plus clear evidence that the management of employee relations hasincreasingly become the province of specialists rather than generalists and thatpractitioners are much more often formally qualified, and to a higher level, than inthe past (Cully et al, 1999; Millward et al, 2000).

The formalisation and professionalisation of HR mean that familiarity with testshas grown. Many of our interviewees emphasised their role, as HR professionals, orof their HR predecessors and colleagues, in encouraging uptake of psychometric andother formal tests. In so doing (as might be predicted) they emphasised thesubstantive value of test use to the organisation rather than organisational dynamicsand the internal balance of power. The latter is nonetheless likely to have been ofinterest to them. We may note that HR professionals will generally play nosubstantive role in job interviews outside their own department (Lupton, 2000),whereas when tests are used they are the experts who interpret the results.

The role of active HR professionals is especially clear in the private sectororganisations from our sample which did their own in-house testing. Inthree-quarters of these, systematic test use was the result of HR professionals havingpromoted it, and in each case this was as part of a more general attempt to tightenup the recruitment process and make procedures and criteria fit clear job and personspecifications. The following comments were typical:

The HR department was the motivating force for change. Test use wasjust part of the drive .. . we've had to move to proper job and personspecs, and are working towards competencies. (Personnel Manager,large privatised company)

Managers aren't actually sure what they are looking for. There's nomatrix of information wanted against method to be used. So if they dosay they want a test, it will be on the lines of 'Oh, I've used them before- I'll have one of them'. When HR ask 'Why?' they haven't a clue. (UKManager for Quality Assurance, international consultancy company)

Public sector practice was less easy for professionals to influence, and only aquarter related similar HR-driven reforms. The widespread use of specialistrecruitment companies by private sector organisations is also testimony to aformalisation of selection processes, but in these cases our interviewees were not ableto discuss the internal dynamics of their client companies.

Formalisation of recruitment, including test use, is often promoted byprofessionals on two grounds: higher quality of selection but also protection of the

200 HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 16 NO 2, 2006

© 2006 The Authors.Journal compilation © 2006 Blackwell Publishing Ltd.

Alison Wolf and Andrew Jenkins

organisation in a potentially hostile environment. The legal and regulatory contextfor increased test use is discussed next, but the symbiotic relationship between thiscontext, the general growth of formalised HR, and the arguments for selection reformwhich are put forward by HR professionals should be emphasised.

The external environment: legal and regulatory factors

While organisational factors are significant, our data suggest that the single mostimportant factor explaining the rise in psychometric test use is a change in theexternal legal and regulatory environment. The rise of the regulatory state hasapparently encouraged organisations to use tests as a defensive measure. Thiswas especially evident for the public sector and privatised companies, wherethree-quarters volunteered defensive use as a major driver of change; but it was alsocited, unprompted, by half the private sector companies who carry out their owntesting and by five of the nine specialist recruitment companies dealing withmultiple clients.

Respondents referred explicitly either to the increased risks now attetidant onmaking hiring decisions in an inappropriate way, or (even more frequently) to theimportance of being able to defend hiring practices as fair and objective should therebe a tribunal or court challenge. In all these cases, tests were seen as an importantsafeguard and potential defence. The following quotations are typical.

The design of the selection process was driven very much by the needto show faimess. (Head of HR and Training, medium-sized privatecompany)

Psychometric tests are certainly seen partly as a safeguard againstaccusations of unfair practice. (Senior Selection Adviser, large publicsector organisation)

Fairness is always an issue. We've given up using role plays becausesome roles are easier or more sympathetic to play than others. (AssistantHead, Corporate Personnel, public sector organisation)

Using personality tests has pre-empted the trouble experienced by otherlocal authorities. On the contrary, we've been praised for objectivity andthe rigour of our testing procedures. People very seldom opt to go beforetribunals because it is so easy to refer back to our systems. (HRDManager, local government)

This does not mean that respondents were not genuinely committed to equalopportunities policies: on the contrary. However, the dominant reason given forusing tests, in this context, was not their substantive contribution to equalopportunities, but the potential risks of not being seen to promote 'equal opps' in avisible, documented fashion. One senior respondent was formalising selectionprocedures, including test use, in a multi-site professional organisation that hadoperated without any clear job descriptions or person specs. 'What I found here', sheinformed us, 'was potential legal dynamite.'

Use of tests for this reason may at first seem somewhat anti-intuitive. A numberof highly publicised American court cases in the early 1980s ruled agaitistcompanies' selection and promotion practices on the grounds of inappropriate test

HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 16 NO 2, 2006 201

© 2006 The Authors.Journal compilation © 2006 Blackwell Publishing Ltd.

Explaining greater test use for selection

use (Russell, 1984; Kleiman and Faley, 1985). Some major tribunal cases in the UKalso found against companies because tests were used which were inappropriate tothe job and resulted in defacto discrimination against minorities. Two key cases in theEuropean Court helped codify the doctrine of 'adverse impact''^ which has sincebeen incorporated into law by the UK along with Italy, Holland and Ireland(Higuera, 2001). One might have expected a consequent drop in test use for selection.

On the contrary, use has continued to increase in the USA (Paul, 2004) as well asthe UK and Europe. Our respondents emphasised the need for recruitment andselection procedures to be demonstrably relevant and demonstrably valid andreliable, and were unanimous in seeing use of tests in selection as powerful evidencefor the presence of properly thought out and objective procedures.

The impact of test use remains a matter of great controversy among academicsand with the general public (Mischel, 1968; Blinkhorn and Johnson, 1990; Paul, 2004),and in every case where comment on this issue was not volunteered, we probed.However, only four of our respondents felt that there was any risk of adverse impactas a result of test use. In the case of 'restricted' tests (requiring licences) thevalidation statistics, and the guarantees of lack of bias which the companiesconcerned provide, were seen as central to the potential value of the tests in casesof legal challenge; and all but the same four respondents accepted these statisticsand statements without reservation. A senior professional in a specialist companystated that psychometric tests are of 'incontrovertible objectivity', while to anotherinterviewee.

High calibre test publishers ensure that tests cannot reflect either genderor ethnicity. In any case, any good assessor would know if the test wasimpartial or not. (Recruitment Consultant in multi-branch recruitmentagency)

Alongside increased test uptake, changes in the legal and regulatory environmenthad led to a declining use of qualifications for selection. As noted earlier, there islittle evidence from our sample that qualifications are seen as less valid and reliablethan in the past. However, a number of respondents commented that qualificationsshould be used less, for equal opportunity reasons; while six of the public sectororganisations had reduced or eliminated formal use of qualifications altogether as abasis for recruitment decisions because of concern that this could leave them opento challenge on equal opportunity grounds. As one respondent explained, indescribing the decision of a major government department to bar any use ofqualifications for selection.

We were really concerned about adverse impact, for older women andminorities especially. The Civil Service Commission wants recruitmentto be fair and open. So we thought we were far better off using testsand competencies. (Occupational Psychologist, Central GovernmentDepartment)

THE RESULTS IN NATIONAL CONTEXT

As noted above, case studies are well suited to clarifying causal mechanisms, but itis difficult to know whether the sample is systematically biased in key respects, and

202 HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 16 NO 2, 2006

© 2006 The Authors.Journal compilation © 2006 Blackwell Publishing Ltd.

Alison Wolf and Andrew Jenkins

we therefore employed survey data from the WERS 98 to check our results. A findingthat organisational and regulatory factors remain significanfly associated with levelsof use in a large data set and in the presence of multiple controls will tend to confirmtheir status as drivers of change.

The WERS management questionnaire included the following two questions:

'When filling vacancies at this workplace do you ever conduct any typeof personality or attitude test?'

and

'When filling vacancies at this workplace, do you ever conduct any typeof proficiency or competency test?'

As discussed earlier, these types of test are very different not only in theconstructs being measured but also in the conditions attached to use. The surveydata also show very different patterns of uptake: 47.9 per cent of Britishestablishments with more than 10 employees make some use of proflciency orcompetency test compared to 19.4 per cent of establishments in the case ofpersonality tests. Because of these differences we analysed uptake separately.

The data were analysed using techniques of ordered logistic regression for surveydata. We constructed separate dependetit variables for personality testing andcompetency testing, each taking the value one if the establishment was using thatform of selection test at the time of the survey and zero otherwise. The dependentvariables were then regressed on a range of explanatory variables, derived from boththe hypotheses examined in the case-study research and the wider literature (seeJenkins and Wolf (2002) for a full discussion of the analysis).

The case-study data indicated that activist HR professionals played an importantrole in increasing test usage, and that this formed part of a general push for theformalisation of selection processes. WERS asks specifically about the nature ofpersonnel management, and, as one measure of formalisation, we distinguished thepresence of an 'HR specialist' from the presence of a 'personnel specialist'. Thisfollows Hoque and Noon (2001)" who found the difference to be associated with theprevalence of high-performance HR practices. As other indicators of formalisedprocedures, we included whether or not the establishment had a formal grievanceprocedure, and whether it was an 'Investor in People', since it is difficult to securethis status without highly formalised personnel, training and record-keepingsystems.

The case studies also highlighted sensitivity to regulatory issues. These wereharder to measure using WERS variables, but we attempted to do so using a numberof variables. We included whether or not there was a formal equal opportunitiespolicy; and created an index of the number of diversity-promoting practices adoptedby the organisation, with a range of 0 through 6. We also included whether or notthere had been any complaint to a tribunal by an employee during the previous 12months, since that experience was likely to have increased regulatory awareness.

Table 1 summarises test use patterns in terms of these variables, and confirmsthat, at national level, both formalised HR procedures and an apparent sensitivity toHR-related legislation are associated with higher levels of test use. However, thesevariables may be highly intercorrelated not only with each other but with other

HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 16 NO 2, 2006 203

© 2006 The Authors.Journal compilation © 2006 Blackwell Publishing Ltd.

Explaining greater test use for selection

TABLE 1 Relationship between test use and individual measures of HR professionalisationand sensitivity to regulation

Specialist HR managerInvestor in People statusNon-IiPFormal grievance proceduresNo formal grievance procedureFormal equal opportunities policyNo formal equal opportunities policyScore of 5 or 6 on 6-point diversity

promotion index"0 or 1 on diversity indexTribunal complaint in last 12 monthsNo tribunal complaint in last 12

months

° The establishment scored one point each for keeping employee records with ethnic origin identified;collecting statistics on posts held by men and women; monitoring promotions by ethnicity, gender, etc.;reviewing selection procedures to identify discrimination; reviewing the relative pay of different groups;making adjustments at the workplace to accommodate those with disabilities.Note: At this level of analysis, using a Pearson chi-squared test, the relationship between test use andpresence of the relevant variable is significant (p < 0.01) in every case other than between IiP status andcompetency test use.Source: WERS 98.

Percentage usingpersonality tests(nationally 19.4%)

60.425.017.020.69.923.212.032.2

14.232.818.3

Percentage usingcompetency tests(nationally 47.9%)

70.849.247.850.527.952.539.668.9

41.366.046.4

organisational variables - notably those identified in other literature as associatedwith test uptake.

Most surveys have found test use to increase with establishment size andunionisation: both were included. So was age, as some authors (e.g. Johns, 1993)suggest that green-field sites are more likely to adopt up-to-date formalised HRpractices. There is considerable evidence that an establishment's skill mix will affectthe likelihood of test use. We therefore included both industrial sector and theproportion of the workforce falling into different categories (professional,managerial, operative, etc.). We also included whether an establishment largely hiredinternally and a training variable (percentage in the largest occupational groupreceiving training in the last 12 months) as indicators of whether the establishmenthad identified, and needed, specific skills.

Economists argue that companies using performance-related pay are able toreward employees differentially after hiring, according to their measuredproductivity, and so have less need of accurate predictive data at the time of hiring(Lazear, 1998: 66). We therefore included the existence of performance-related paycontracts as a variable, but would emphasise that the WERS data do not include

204 HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 16 NO 2, 2006

© 2006 The Authors.Journal compilation © 2006 Blackwell Publishing Ltd.

Alison Wolf and Andrew Jenkins

proportions or types of workers on such contracts. Finally, we included variablesrelating to company vacancies (types of vacancy in the last 12 months) and locallabour market conditions (existence of short-term contracts, unemployment rate inthe establishment's travel to work area), to indicate whether or not establishmentswere actively in need of particular types of skills, and how easy it was to find them.Both are likely to affect the resources allocated to individual selection procedures.

Because of the large numbers of variables to be investigated, the statisticalmodelling proceeded in a number of stages. Initially the effects of workplace sizealone were examined, as this had been the most strongly supported determinant oftest use in previous studies (see Jenkins, 2001). In the early stages of the analysis thisrelationship indeed appeared strong. However, once the other variables wereincluded, establishment size ceased to show any significant impact on test use.

Variables relating to the type of HR expertise and the skill mix in theestablishment were added, followed by indicators of local labour market conditions,level of commitment to diversity, grievance procedures, tribunal cases and sector. Ateach stage, variables which did not show significant or near-significant levels ofimpact were dropped from the model.

As emphasised above, our objective was to determine whether the factorsidentified in the case studies remained important in the presence of controls, not toprovide an exhaustive explanation of when establishments use tests. Table 2summarises the main results for use of personality and attitude tests and providesclear support for an association between personality testing and the presence of HRspecialists and of formalised procedures, as indicated by Investor in People status.Personality testing is also strongly related to the number of diversity-promotingpractices (which we have used as an indicator of sensitivity to regulatoryrequirements but may also be seen as a reflection of formal HR practices). Thepositive relationship between workplace age and personality test use is not what theliterature predicts, but may be related to progressive formalisation of the HRfunction as a workplace becomes more established.

Conversely, the measures of sensitivity to regulation are not significant. Neitherexistence of a formal grievance procedure nor recent tribunal experience issignificantly associated, in the multivariate model, with personality test use. As weshall see, this is not the case for competency test uptake.

As Table 2 indicates, a number of the skill mix and labour market variables werealso signiflcantly associated with personality test use. Managerial and professionalvacancies (as opposed to their level of representation in the workforce) are positivelyrelated to use, operative and unskilled vacancies negatively so. This is consistentwith Wilk and Cappelli's findings on relationships between job requirements and testuse. Wilk and CappeUi (2003) find that the higher the skill level, the more likelytesting is to occur. Higher levels of training are strongly associated with personalitytesting. Local labour market variables are harder to interpret. Testing is less likelywhen local unemployment is very high. If employers find it relatively easy to replaceworkers, they may well spend less time and money on selection. Conversely, therelationship with short-term contract use is positive, though the variable itself is notvery satisfactory (measuring simply whether or not any such contracts are used).

Finally, there are only two significant relationships with measures of skill mixbased on percentages of the workforce in different positions. In both cases the

HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 16 NO 2, 2006 205

© 2006 The Authors.Journal compilation © 2006 Blackwell Publishing Ltd.

Explaining greater test use for selection

TABLE 2 Variables associated with uptake

HR specialist presentInvestor in People statusNo. of practices to promote diversityManagerial vacancies in last 12 monthsProfessional vacancies in last 12 monthsRoutine unskilled vacancies in last

12 monthsUnemployment in travel to work area

(base 0%-2%) more than 7.75%Workers on short-term contracts (<1 year)

of personality

Coefficient

1.3670.5370.1970.5360.637

-0.521

-1.931

0.476

tests

Standard terror

0.3380.2150.0590.2180.2840.221

0.621

0.241

Training: % in largest occupational group received training (base, noHigh (80%-100%)Medium (40%-79%)% operatives in workplace% professionals in workplaceWorkplace age (base, 0-4 years)5-9 years10-24 years25 or more years

0.9841.3880.024

-0.028

0.9610.3300.986

0.3830.4290.0060.008

0.3460.3190.316

4.042.493.332.462.24

-2.36

-3.11

1.97

training)2.573.243.72

-3.57

2.781.033.12

Notes: The results are from logistic regressions: for complete specification of the models used.and Wolf (2002). Results include controls for workplace size (not significant in

p > l t l

0.000*»*0.013»»0.001***0.014**0.025**0.019**

0.002***

0.049**

0.010***0.001***0.000***0.000***

0.006***0.3010.002***

5ee Jenkins

the full model) and forindustrial sector. Variables dropped from the final specification were grievance procedures, complaints totribunals in last 12 months, union recognition. internal labourmanagerial, technical, clerical, sales, routine unskilled, % earning*** 1% significance level, ** 5%.N = 1,978.Source: WERS 98.

markets, %>£22k.

in occupational groups

coefficients are very small, although the negative relationship between test use andhigh proportions of professionals may reflect the finding, in case studies carried outby Scholarios and Lockyer (1999), that small professional consultancies are unlikelyto use personality measures, focusing instead on specific professional skills.

Table 3 shows results from the parallel modelling exercise carried out withcompetency test use as the dependent variable. Use of competency tests is muchmore widespread than use of personality tests, and fewer of the independentvariables showed significant relationships with test use. However, the results onceagain lend support to the case-study findings that both HR formalisation and theregulatory environment are driving change. The number of diversity-promotingpractices used and the existence of formal grievance procedures are bothsignificantly and independently associated with use (although the existence of an HRspecialist is not). So, too, is recent tribunal experience, although the cross-sectional

206 HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 16 NO 2, 2006

© 2006 The Authors.Journal compilation © 2006 Blackwell Publishing Ltd.

Alison Wolf and Andrew Jenkins

TABLE 3 Variables associated with uptake

ConstantNo. of practices to promote diversityFormal grievance proceduresComplaints to tribunals in last 12 monthsWorkers on short-term contracts (<1 year)% routine unskilled workersCraft/skilled vacancies in last 12 monthsTechnical vacancies in last 12 monthsOperative vacancies in last 12 monthsWorkers on performance-related pay

of competency

Coefficient

-0.9610.2140.6750.4450.478

-0.0110.5770.461

-0.7280.505

tesfs

Standarderror

0.4910.0560.3450.2230.2550.0050.2080.2580.2970.247

t

-1.963.821.961.991.87

-2.382.781.79

-2.452.04

Note: The results are from logistic regressions: for complete specification of the models usedand Wolf (2002). Results include controls for workplace size (notindustrial sector. Variables dropped from the final specificationreceiving training, liP, local unemployment rate.professional, technical, clerical, sales, operative.market.**» 1% significance level, ** 5'%, * 10%.N = 2,101.Source: WERS 98.

workplace age, "i

significant in thewere presence> in occupationa

p>\t\

0.05**0.000***0.050**0.046**0.061*0.017**0.006***0.074*0.014**0.041**

, see Jenkins

full model) and forof HR igroups

pecialist, %managerial.

% earning >£22k, union recognition, internal labour

nature of the WERS data means that we cannot determine how far greater use ofcompetency testing was actually a response to such experiences.

Apart from these factors, only a limited number of other variables provedsignificant. As with personality testing, usage was positively related to the existenceof short-term contracts and also to the use of performance-related pay for at leastsome workers. This second relationship suggests that, at least in the UK, those firmsthat can measure individual productivity with accuracy prefer to do so before hiring(through use of skill tests) rather than by hiring and then paying differentiallyand/or firing, as the US literature suggests. This may in turn reflect regulatorydifferences that make it harder to dismiss employees in the UK. Finally, use ofcompetency testing is positively related to craft vacancy levels and negatively tooperative vacancy levels and to the proportion of unskilled workers in theworkforce. All these relationships are intuitively plausible, given that competencytests are concerned with measurement of specific job-related skills.

CONCLUSION AND IMPLICATIONS FOR PRACTITIONERS

Although the literature on testing and selection is dominated by issues of testvalidity, a number of authors have argued that this can be restrictive and misleading(see, e.g., Johns, 1993; Ramsay and Scholarios, 1999; Carlson and Connerley, 2003;Muchinsky, 2004). The results reported here support this position. Test use is not, it

HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 16 NO 2, 2006

© 2006 The Authors.

Journal compilation © 2006 Blackwell Publishing Ltd.

207

Explaining greater test use for selection

appears, primarily dependent on either technical characteristics of particular tests orthe structural characteristics of a subsection of establishments (e.g. large size orgreen-field sites). Their use, and more specifically recent increases in testing, arerather to be explained by ongoing changes in the workplace, notably theprofessionalisation of HR and the growing influence of the regulatory state, and bythe resulting institutional dynamics.

These findings imply that further increases in test use are very likely, but alsothat there is no reason, a priori, to expect that these will necessarily increasethe effectiveness of the selection process. In their classic discussion of institutionalisomorphism, DiMaggio and Powell (1983) argue that organisations tend to becomeincreasingly similar in structure for reasons that have little to do with efficiency. Twoof the mechanisms they identify as important are evident here, and suggest both thatthere will be continued convergence in organisational selection practices and that theresults for organisational effectiveness are unclear. The first mechanism is a commonlegal environment (creating what the authors term 'coercive isomorphism'). Thesecond is professionalisation, including not only shared education but also thegrowth of professional networks along which ideas and reconunendationsdisseminate and which create 'normative pressures'.

The regulatory factors identified in our research are a clear example of how thelegal environment may create 'coercive isomorphism'. Awareness of and worriesabout the regulatory environment and its risks push unrelated organisations towardsformal testing for selecfion; but such general forces, while promoting organisationalconvergence, are neutral with respect to efficiency (DiMaggio and Powell, 1983: 152).In our research we found an example of the growing tendency, in public and privatesectors alike, to formally identify and 'manage' risks to the organisation; but asPower points out, there is an 'overwhelming tendency for the new risk managementto exacerbate process' as part of a new culture of audit and accountability (Power,2004: 13, italics in original).

If increased use of formal testing indeed prevents organisations from facingexpensive legal actions, then this is potentially a benefit, though one that needs tobe evaluated alongside the overall impact on costs and the quality of the selectiondecisions made. It is important to note, however, that HR personnel are likely toview the relative costs and benefits of such defensive testing rather differently thanthe organisation as a whole.

Modern organisational theory emphasises that, very often, 'a disparity can arisebetween the costs and beneflts that an individual bears versus those that accrue tothe organization as a whole' (Roberts, 2004: 119). As a result, people may undertakemore, or less, of an activity than is optimal for the organisation as a whole. In theregulatory context, it seems likely that HR personnel will be highly inclined to 'riskminimise' in the selection procedures they advocate. If a selection process leads toa formal complaint, it is the HR department that will be blamed. However, it isextremely difficult to measure and track the effecfiveness and productivity of mostindividual employees. So if recommended procedures, such as more testing, fail toimprove (or even reduce) the effecfiveness of selection decisions, this may well neverbe detected.

Professionalisafion plays a similar role to laws and regulations in advancing'institutional isomorphism' - in this case, test use - without any presumption that

208 HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 16 NO 2, 2006

© 2006 The Authors.Journal compilation © 2006 Blackwell Publishing Ltd.

Alison Wolf and Andrew Jenkins

this will necessarily increase efficiency. As professionalisafion of HRM continues toadvance, so too will familiarity with testing theories and test batteries. Quite apartfrom regulatory issues, more and more HR professionals will be keen to deployselecfion tests either because they believe in their effecfiveness or because their useincreases the importance and influence of HR departments, or both. HR professionalswill also increasingly work with managers who obtained their own jobs partlythrough test performance. This is likely to mean that advocacy of test use will wingreater support, since people are inclined to think that procedures at which theythemselves were successful must be robust and desirable.

In advocating use of tests, pracfifioners are therefore increasingly likely to pushon an open door. Further adoption of tests will be perfectly rational (Ramsay andScholarios, 1999: 68). It will not, however, necessarily be effective in improvingselection. Academic evidence demonstrates the predicfive validity of appropriatetesting, but it also underlines the crifical importance of context, includingorganisational culture and management philosophy, and of selecting the correct testfor correctly identified traits (Mischel, 1968; Robertson and Kinder, 1993; Townley,1999). The danger to which professionals must be sensifive is that tests may refiectthe safety of common pracfice rather than careful analysis of specific organisafionalneeds.

Acknowledgements

The research described was carried out under a grant received from the Centrefor the Economics of Education, funded by the Department for Educafion andSkills. Their assistance, the collaboration of the Chartered Institute of Personneland Development (CIPD), and comments on earlier drafts from Anna Vignoles,David Guest and two anonymous referees are gratefully acknowledged. Theviews and conclusions expressed are, of course, entirely the authors' own.

Notes

1. Three items are included iti the 'recruitment and selection' category when compiling anindex of HRM (Guest et al, 2003: 298), but one of the three is whether or not recruitmentcurrently generates good candidates. The other 'active' HR practice measured is whetherpreviews of work in the organisation are provided to candidates.2. The difficulty of establishing clear causal links between bundles of 'best' HR pracficesand organisational performance, let alone between the latter and a single strand, such asselection procedures, is of course well rehearsed in the research literature (see, e.g., Beckerand Gerhard, 1996; Purcell et al, 2003). Terpstra and Rozell (1993) looked specifically atthe relationship between selection practices and organisational performance in the US privatesector, and found that use of practices which tend to improve predictive accuracy - use offollow-up studies, use of validafion studies, use of structured interviews, use of cognitiveability tests, use of biographical or weighted application blanks - was associated withgreater profitability, hut not greater sales growth; they also caution that their study iscross-sectional so causality cannot be inferred.3. These include one agency which operated entirely in the public sector and one whosefunction was to provide particular recruitment/selection activities for a group of over 20small councils. Since the same procedures are used by all of these, they are counted as onlyone case for statistical purposes.

HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 16 NO 2, 2006 209

© 2006 The Authors.Journal compilation © 2006 Blackwell Publishing Ltd.

Explaining greater test use for selection

4. In some cases we interviewed both an external consultant and an internal manager onpractices in the same company; and in some organisations we interviewed one, and in othersmore than one, internal professional.5. Agriculture, forestry and fishing, mining and quarrying are excluded.6. Strictly, four privafised and one newish public-private collaboration.7. See Chartered In.stitute of Personnel and Development (2002) for some detailed results onthe use of different types of test for different types of vacancy.8. Respondents generally referred to requiring/having 'level A' or 'level B' training. Thisrefers to standards and cerfificates originally developed by the Brifish Psychological Societyand the Chartered Institute of Personnel and Development (previously Institute of Personneland Development). As a simplified rule of thumb, level A training can be seen as enablingholders to use ability tests, whereas most personality tests, including the most popular andwell-known tests, can be interpreted only by those certified at level B.9. The major exception to this statement is that the Occupafional Personality Quesfionnairewas (re)launched in 1999 a.s OPQ32, assessing personality using 32 characterisfics, andreplacing previous more restricted versions.10. Respondents who described managers as changing their views generally ascribed this tothe fact that there had been 'no disasters' since personality tesfing was introduced. Two alsoadded that managers who had been tested and then appointed themselves tended to see thetesting process as both desirable and accurate.

11. The finding that test data are not seen to have become 'better' needs qualifying in onerespect. A few of the organisations visited (mostly in the public sector) had moved to anorganisafion-wide competency framework. One respondent, a psychologist with an unusuallylong period of involvement with his (large) privafised company's selecfion procedures, notedthat test output is reported in a way which aligns easily with competencies (e.^. analyficalreasoning), whereas other sources of information, such as formal qualifications, do not. Thismay explain .some increases in use, but only this one respondent mentioned it explicitly as afactor.

12. The key concepts are that 'Any procedure or test that produces in a selection process anadverse impact against women or other groups of applicants is potentially discriminatory'and that measures which are 'meant to be formally neutral' but produce an unfavourableeffect against a subgroup 'have to be screened against criteria of technical validity' (Higuera,2001: 105-106).

13. A personnel specialist was present at the workplace if the job fitle of the respondent wasHR manager, personnel manager or employee relations manager, if the respondent spent 50per cent or more of their fime on personnel matters, or if there was someone at a higherlevel in the organisation who spent a major part of their fime on personnel matters. Thisdefinition was adapted from Cully et al (1999). An HR specialist was defined as present atthe workplace if the Job fitle of the respondent was HR manager (see Hoque and Noon,2001).

REFERENCES

Bartram, D. (2004). 'Assessment in organisations'. AppUed Psychology, 53: 2, 237-259.Bartram, D., Lindley, RA., Marshall, I. and Foster, J. (1995). 'The recruitment and selecfion

of young people by small businesses'. Journal of Occupational and OrganizationalPsychology, 68: 4, 339-358.

210 HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 16 NO 2, 2006

© 2006 The Authors.Journal compilation © 2006 Blackwell Publishing Ltd.

Alison Wolf and Andrew Jenkins

Becker, B.E. (1989). 'The influence of labor markets on human resources ufility esfimates'.Personnel Psychology, 42: 531-546.

Becker, B. and Cerhard, B. (1996). 'The impact of human resource management onorganizational progress and prospects'. Academy of Management Journal, 39: 4, 779-801.

Becker, G.S. (1993). Human Capital A Theoretical and Empirical Analysis with Special Referenceto Education, 3rd edn, Chicago, IL: University of Chicago Press.

Blinkhorn, S.F and Johnson, CE. (1990). 'The insignificance of personality testing'. Nature,348: 671-672.

Caldwell, R. (2001). 'Champions, adapters, consultants and synergists: the new changeagents in HRM'. Human Resource Management Journal, 11: 3, 39-52.

Carlson, K.D. and Connerley, M.L. (2003). 'The staffing cycles framework: viewing staffingas a system of decision events'. Journal of Management, 29: 1, 51-78.

Chartered Insfitute of Personnel and Development (2002). Recridtinent Survey Report,London: Chartered Insfitute of Personnel and Development.

Cully, M., Woodland, S., O'Reilly, A. and Dix, G. (1999). Britain at Work: As Depicted bythe 1998 Workplace Employee Relations Survey, London: Routledge.

Dayan, K., Kasten, R. and Fox, S. (2002). 'Entry-level police candidate assessment center:an efficient tool or a hammer to kill a fiy?' Personnel Psychology, 55: 827-849.

DiMaggio, P. and Powell, W. (1983). 'The iron cage revisited: institutional isomorphismand collective rafionality in organizational fields'. American Sociological Review, 48: 2,147-160.

Elias, J. and Scarbrough, H. (2004). 'Evaluating human capital: an exploratory study ofmanagement practice'. Human Resource Management Joumal, 14: 4, 21-40.

Gaugler, B.B., Rosenthal, D.B., Thornton, G.C, III and Bentson, C (1987). 'Meta-analysesof assessment center validity'. Joumal of Applied Psychology, 72: 3, 493-511.

Graham, H.T. and Bennett, R. (1995). Human Resources Management, 8th edn, London:Longman.

Guest, D. and King, Z. (2004). 'Power, innovafion and problem-solving: the personnelmanagers' three steps to heaven?' Journal of Management Studies, 41: 3, 401-423.

Guest, D., Michie, J., Sheehan, M. and Conway, N. (2000). Employment Relations, HRM andBusiness Performance: an Analysis of the 1998 Workplace Employee Relations Survey,London: Chartered Insfitute of Personnel and Development.

Guest, D., Michie, J., Conway, N. and Sheehan, M. (2003). 'Human resource managementand corporate performance in the UK'. British Journal of Industrial Relations, 41: 2,291-314.

Guion, R.M. (1998). Assessment, Measurement and Prediction for Personnel Decisions,Mahwah, NJ: Eribaum.

Higuera, L.A. (2001). 'Adverse impact in personnel selecfion: the legal framework and testbias'. European Psychologist, 6: 2, 103-111.

Hood, C, Scott, C , James, O., Jones, G.W. and Travers, T. (1999). Regulation insideGovernment: Waste-watchers, Quality Policy and Sleaze-busters, Oxford: Oxford UniversityPress.

Hoque, K. and Noon, M. (2001). 'Counfing angels: a comparison of personnel and HRspecialists'. Human Resource Management Joumal, 11: 3, 5-22.

Jenkins, A. (2001). 'Companies' use of psychometric testing and the changing demand forskills: a review of the literature'. Discussion Paper 12, London: Centre for theEconomics of Education.

Jenkins, A. and Wolf, A. (2002). 'Why do employers use selection tests? Evidence fromBrifish workplaces'. Discussion Paper 27, London: Centre for the Economics ofEducation.

HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 16 NO 2, 2006 211

© 2006 The Authors.Journal compilation © 2006 Blackwell Publishing Ltd.

Explaining greater test use for selection

Jenkins, A. and Wolf, A. (2005). 'Employers' selecfion decisions: the role of qualificafions',in S. Machin and A. Vignoles (eds). What's the Good of Education? The Economics ofEducation in the UK, Princeton, NJ: Princeton University Press.

Johns, G. (1993). 'Constraints on the adopfion of psychology-based personnel practices.Lessons from organizational innovation'. Personnel Psychology, 46: 569-592.

Kleiman, L.S. and Faley, R.H. (1985). 'The implicafions of professional and legalguidelines for court decisions involving criterion-related validity: a review andanalysis'. Personnel Psychology, 38: 803-^33.

Lazear, E.P. (1998). Personnel Economics for Managers, New York: Wiley.Legge, K. (1995). Human Resource Management - Rhetorics and Realities, Basingstoke:

Macmillan.Lupton, B. (2000). 'Pouring the coffee at interviews? Personnel's role in the selection of

doctors'. Personnel Review, 29: 1, 48-64.Mabey, B. (1992). 'The growth of test use'. Selection and Development Review, 8: 3, 1 ^ .Millward, N., Bryson, A. and Forth, J. (2000). All Change at Work? British Employment

Relations 1980-1998, as portrayed by the Workplace Industrial Relations Survey series,London: Routledge.

Mischel, W. (1968). Personality and Assessment, New York: Wiley.Moran, M. (2003). The British Regulatory State: High Modernism and Hyper-innovation,

Oxford: Oxford University Press.Muchinsky, P.M. (2004). 'When the psychometrics of test development meets

organizafional realifies: a conceptual framework for organizafional change, examples,and recommendations'. Personnel Psychology, 57: 175-209.

Murphy, R. (2000). 'Public confidence in examinafions and qualifications'. Unpublishedreport to the Qualifications and Curriculum Authority, London.

Paul, A.M. (2004). The Cult of Personality. How Personality Tests are Leading us to Miseducateour Children, Mismanage our Companies and Misunderstand Ourselves, New York: FreePress.

Power, M. (1997). The Audit Society: Rituals of Verification, Oxford: Oxford University Press.Power, M. (2004). The Risk Management of Everything: Rethinking the Politics of Uncertainty,

London: Demos.Purcell, J., Kinnie, N., Hutchinson, S., Rayton, B. and Swart, J. (2003). Understanding the

People and Performance Link: Unlocking the Black Box, London: Chartered Insfitute ofPersonnel and Development.

Ramsay, H. and Scholarios, D. (1999). 'Selecfive decisions: challenging orthodox analysesof the hiring process'. International Journal of Management Reviews, 1: 1, 63-89.

Roberts, J. (2004). The Modern Firm. Organizational Design for Performance and Growth,Oxford: Oxford University Press.

Robertson, l.T. and Downs, S. (1989). 'Work-sample tests of trainability. A meta-analysis'.Joumal of Applied Psychology, 74: 3, 402^10.

Robertson, l.T. and Kinder, A. (1993). 'Personality and job competences: thecriterion-related validity of some personality variables'. Journal of Occupational andOrganizational Psychology, 66: 3, 225-244.

Russell, J.S. (1984). 'A review of fair employment cases in the field of training'. PersonnelPsychology, 37: 261-276.

Ryan, A.M., McFarland, L., Baron, H. and Page, R. (1999). 'An international look atselecfion practices: nafion and culture as explanafions of variability in practice'.Personnel Psychology, 52: 359-392.

Schmidt, FL. and Hunter, J.E. (1998). 'The validity and ufility of selecfion methods inpersonnel psychology: pracfical and theoretical implications of 85 years of researchfindings'. Psychological Bulletin, 24: 2, 262-274.

212 HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 16 NO 2, 2006

© 2006 The Authors.Journal compilation © 2006 Blackwell Publishing Ltd.

Alison Wolf and Andrew Jenkins

Schmitt, N. and Chan, D. (1998). Personnel Selection: A Theoretical Approach, ThousandOaks, CA: Sage.

Schmitt, N., Gooding, R.Z., Noe, R.A. and Kirsch, M. (1984). 'Meta-analysis of validitystudies published between 1964 and 1982 and the invesfigation of studycharacteristics'. Personnel Psychology, 37: 407-422.

Scholarios, D. and Lockyer, C (1999). 'Recruifing and selecting professionals: context,qualities and methods'. International Joumal of Selection and Assessment, 7: 3, 142-156.

Shackleton, V. and Newell, S. (1991). 'Management selecfion: a comparative study ofmethods used in top British and French companies'. Joumal of Occupational Psychology,64: 1, 23-36.

Terpstra, D.E. and Rozell, E.J. (1993). 'The relafionship of staffing practices toorganizafional level measures of performance'. Personnel Psychology, 46: 27-47.

Torrington, D., Hall, L. and Taylor, S. (2002). Human Resource Management, 5th edn,Harlow: Prenfice-HaU.

Townley, B. (1999). 'Pracfical reason and performance appraisal'. Journal of ManagementStudies, 36: 3, 287-306.

Wilk, S. and CappeUi, P. (2003). 'Understanding the determinants of employer use ofselection methods'. Personnel Psychology, 56: 1, 103-124.

Wood, R. and Payne, T. (1998). Competency-based Recruitment and Selection, Chichester:Wiley.

HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 16 NO 2, 2006 213

© 2006 The Authors.Journal compilation © 2006 Blackwell Publishing Ltd.

top related