Evaluation of success criteria in job search A process perspective

57 Journal of Occupational and Organizational Psychology(1999),72,57–70Printed in Great BritainÓ1999The British Psychological SocietyE valuation of success criteria in job search:A process perspectiveE ric E.Brasher*and Peter Y.ChenDepartment of Psychology,Ohio University,USAJob-search studies possess a disturbing lack of consistency in the operationaliz-ation of search success.This prevents researchers from attaining their goal ofexplaining and predicting the use of job information sources and search activities.In the current study,we examined two theoretical models that depicted structuralrelations among nine success criteria often employed in the job-search literature.Overall,the criteria are weakly related to each other.The ndings suggest theimportance of careful consideration of the meaning and process of job-searchsuccess criteria,and that these issues should not be further overlooked.At any given time,millions of individuals are searching for job opportunities.It is apparent that knowledge of the job-search process may help job seekers plan e Y cient search strategies,and cope with uncertainty,as well as alleviate psycho-logical or physical strains during the course of searching(Stumpf,Colarelli& Hartman,1983;Wine eld,Wine eld,Tiggemann&Goldney,1991).Additionally, it can facilitate organizations to choose cost-e V ective recruitment methods and identify prospective applicants.Survey results estimated that Fortune500 rms allocated16%of their human resource budget to recruitment(Rynes&Boudreau, 1986).Recognizing the prevalence of job search in society and its momentous implications for both individuals and organizations,researchers have increasingly investigated central aspects of the search process over the past few years(e.g. Barber,Daly,Gianntonio&Phillips,1994;Saks&Ashforth,1997;Vecchio,1995; Wanberg,Watt&Rumsey,1996).This research can roughly be classi ed into two approaches:organizational and individual.The former approach examines how organizations’recruitment and selection programmes relate to job seekers’search decisions,performance and attitude(e.g.recruiting sources,Breaugh,1981;recruiter behaviour,Rynes&Miller, 1983).In contrast,researchers focusing on the individual approach attempt to explain and predict individual success in job search.In the current study we took the latter approach,which generally focuses on three major research topics(i.e. sources of search information,search activities and individual characteristics).For instance,e V ectiveness of job information sources is evaluated on the basis of success criteria(e.g.Allen&Keaveny,1980;Rosenfeld,1975).Other researchers *Requests for reprints should be addressed to Eric E.Brasher,National Computer Systems Inc.,9701W.Higgins Road,Rosemont,IL60018,USA.58E ric E.Brasher and Peter Y.Chenstudied how search behaviour or intention(e.g.persistence or intensity)related to desirable search outcomes(e.g.Dyer,1973;Kanfer&Hulin,1985).Furthermore, relations of individual characteristics(e.g.personality or demographic variables) with search activities and outcomes(e.g.Cook,Vance&Spector,1995;Ellis& Taylor,1983;Ste V y,Shaw&Noe,1989)were explored.Among the noted research perspectives,researchers encounter a similar important question.That is,what search outcomes are appropriate criteria for evaluating job search?In other words, what criteria best represent success in the search?Weitz(1961)correctly pointed out that the choice of the criterion has a major e V ect on the results attributed to the independent variables or predictors.Without adequate criteria,de nitive answers to researchers’inquiries(e.g.e Y ciency of search activities or e V ectiveness of search information sources)would be di Y cult to substantiate(Bingham,1926). Although a wide variety of behavioural and attitudinal criteria,operationalized as alternative indicators of job-search success,have been developed,evidence of convergent validity among them did not seem available in prior research.We reviewed various operational de nitions of job-search success and their limitations. Because our interest focuses on search success from job seekers’perspectives, success criteria that primarily concern organizations(e.g.performance or productivity)were excluded in the following review.1Review of success criteriaThe rst intuitive criterion of search success is based simply on who has found a job.A variation of this measure is to determine whether job seekers have found a job by a set time(Rosenfeld,1975).Typically,a somewhat arbitrary cut-o V time for nding a job is chosen,such as one month after being laid o V another job(e.g. Kanfer&Hulin,1985)or whether students have found a job by graduation. Whatever cut-o V is selected,this criterion implicitly assumes that all employed persons are equally successful in their search by virtue of having found a job within a speci ed time.This assertion does not seem tenable because job quality is not taken into account.Facing a variety of pressures,job seekers may complete their job search within a relatively short period of time,and accept a less than desirable job.A related measure,search duration,assesses how long it takes job seekers to nd a job(e.g.Dyer,1973).Similar to the previous dichotomous criterion,search duration does not take job quality into consideration.However,the latter measure allows for much more variability among job seekers.Several researchers have evaluated job search success based on various measures of job quality,such as salary(e.g.Granovetter,1974).However,the results may be misleading if the respondents did not start their jobs at approximately the same time.In this case,salaries reported may di V er from starting salaries for reasons unrelated to job search(bour market,promotions,transfer).This measure also su V ers from range restriction within a homogeneous labour market such as social workers(Rynes,Schwab&Heneman,1983).On the other hand,variations in the labour market a V ect ranges of starting salary across di V erent jobs.It would 1A summary of criteria employed in job-search studies can be obtained from the rst author upon request.Success59 be less meaningful to conduct a similar evaluation in a heterogeneous samplewithout considering labour markets,because such di V erences in salary should be attributed to the choice of profession rather than to success in the search process. The second measure of job quality is the degree of match between academic major and job.An example of a match measure is to assess perceived similarity between the current job and educational training(Allen&Keaveny,1980).Poor match between person(e.g.interest,skills)and environment(e.g.job demand) might have a functional impact on job involvement(Rabinowitz&Hall,1977)or strain reactions(Beehr,1995).However,it may be argued that achieving this similarity is not a primary objective of all job seekers.Job satisfaction is the third measure of job quality.This measure has been used to evaluate recruiting sources(Breaugh,1981)and activity of career exploration (Ste V y et al.,1989).It is presumed that,in general,employees would be more satis ed with a better job(Granovetter,1974).However,it should be noted that the causes of job satisfaction are generally complex and intertwined.Certain people might be more satis ed than their co-workers due to past experience or genetic or personality factors(e.g.Arvey,Bouchard,Segal&Abraham,1989).Empirical evidence also suggested that job satisfaction tends to be associated with age and tenure(White&Spector,1987).Self-report of job satisfaction therefore may vary, depending upon the time at which employees respond.Finally,turnover(e.g.Decker&Cornelius,1979)and tenure(e.g.Caldwell& Spivey,1983)have been used to evaluate the e V ectiveness of job information sources.It has been shown that satis ed employees are likely to stay longer at a job (Mobley,1977),although dissatis ed employees may not necessarily quit their jobs during high unemployment seasons(Carsten&Spector,1987).It is also reasonable to employ intent to quit as an alternative measure to complement measures of turnover or tenure.Intent to quit,an attitudinal measure that may not be a V ected by unemployment rates,is generally viewed as an immediate antecedent of turnover (e.g.Bannister&Gri V eth,1986).As with salary,these variables may be dependent on when employees started their jobs.A few other criteria have been used to represent job search success,such as number of interviews(Stumpf,Austin&Hartman,1984)and number of job o V ers (Ste V y et al.,1989).Selection of these criteria is based on the assumption that all job seekers aim to maximize interviews and job o V ers.Ste V y et al.suspected that this assumption is not plausible because some job seekers may focus on a narrow range of jobs in a particular locale and seek relatively few interviews in a concentrated job search.Ste V y et al.also included a measure of search stress to evaluate career exploration.Job search is likely to be stressful for job seekers because of the importance of work, nancial pressures and rejection(Barber et al.,1994;Caplan, Vinokur,Price&Van Ryn,1989).Presumably,successful job seekers are better able to cope with the strains associated with search than those who are less successful.The current studyThe above review presented a challenge to researchers when choosing among disparate criteria.Conclusions regarding the e V ectiveness of job informationsources or search activities could be drastically a V ected by what success criteria are used.Most of the aforementioned criteria appeared quite di V erent on the surface,and could be weakly related with one another or even negatively associated.For instance,Schwab,Rynes &Aldag (1987)suggest the possibility of an inverse relationship between the speed and quality of employment obtained.It is our contention that the aforementioned criteria are de cient as a single measure of search success.We argued that it is critical to investigate the relations among the criteria,and to study their theoretical structures.The rst purpose of the present study was an attempt to clarify the meaning and measurement of search success.A rather comprehensive group of nine criteria was simultaneously evaluated.These criteria included duration of search,job o V ers,job interviews,search stress,starting salary,job/school match,job satisfaction,pay satisfaction and intent to quit.These nine criteria were rst evaluated on the basis of their interrelations hips.Given their conceptual diversity,it was expected that some of the relations between them would be fairly weak (as described below).The second purpose of the present study was to evaluate the relative e Y cacy of two search success models based on theorized relations among the above criteria.The measurement model among the structural models (see Fig.1)consisted of two latent variables:primary success and satisfaction.Primary success was indicated by three search criteria:search duration,number of interviews per month,and number of o V ers per month.All three are generally assessed during or immediately after search and do not involve the actual job ultimately attained.Regarding the latent construct of satisfaction,general satisfaction and pay satisfaction were employed as its e V ect indicators.The rst proposed model served as a basic structural model that consisted of primary success,satisfaction,starting salary,job/school match and intent to quit.Relationships among these variables have been fairly well established in the literature.As stated earlier,primary success generally does not involve the job nally attained because it occurs during or immediately after search.It seemed logical to expect null (or weak)causal relations of primary success with satisfaction and intent to quit.Starting salary and match are expected to positively in uence satisfaction.Salary is a highly valued job attribute (Rynes et al .,1983)and survey data have consistently demonstrated a positive relationship between salary and job satisfaction (e.g.Rice,Figure 1.Structural models of job search criteria.60E ric E.Brasher and Peter Y.ChenSuccess61 Phillips&McFarlin,1990).Several studies have shown that employees’perceptionsof person–environment t were also positively related to job satisfaction(e.g.Cable &Judge,1996;Saks&Ashforth,1997).Higher job satisfaction would in turn reduce intention of quitting a job(Michaels&Spector,1982).Match between education and job may also in uence starting salary because individuals who nd jobs close to their training and achieve good t should expect equitable compen-sation according to hiring practices in the labour market(Adkins,Russell&Werbel, 1994;Rynes&Gerhart,1990).Extending from the rst model,the second model proposed additional paths that were derived from the literature,but had not been demonstrated empirically.These paths depict possible causal relationships among primary success,search stress, starting salary and job/school match.It is assumed that primary success would a V ect search stress,starting salary and job/school match.In addition,it hypothe-sized that search stress may lead to the decrease of starting salary and match.Job search tends to be a stressful experience for most individuals due to uncertainty and rejection(Barber et al.,1994;Stumpf et al.,1983).Job seekers who experienced an extended search and/or received few interviews and o V ers would likely view the process as particularly stressful(Plumly&Oliver,1987).Presumably,a much shorter job search or relatively more job interviews/o V ers received also suggested that job seekers had a much better opportunity nding jobs that reasonably met their nancial(i.e.salary)and personal needs(i.e.match between training and job). Search stress is hypothesized as an antecedent of starting salary and match because job seekers are likely to accept less than optimal job o V ers when faced with stressors such as diminishing nancial resources that intensify as the search progresses(Barber et al.,1994;Soelberg,1967).The more strains experienced by job seekers during their search,the more likely they are to accept a job that did not meet nancial and personal needs.In other words,an observed negative in uence of search stress on starting salary and match is expected.MethodParticipants and procedureParticipants in this study were1993and1994graduates of a medium-sized midwestern American university.A survey package was mailed to1425participants representing all nine colleges,who were randomly selected from these two groups of graduates.They were assured that their responses would be kept con dential.Reminder notices were sent out about10days after the original survey was mailed.Of these contacted,360participants who held a wide variety of jobs returned completed surveys.Excluding70surveys that could not be delivered because of incorrect addresses,the response rate was conservatively estimated to be30%.The participants had an average age of25.4years and 67%were pared to the population,a higher proportion of female respondents completed the questionnaires in the sample.Overall there was no signi cant di V erence between the sample and the university population in their distribution of colleges.Among the completed surveys,65were dropped from the following analysis because the respondents either were holding part-time jobs2(38),were enrolling in graduate schools(5),had2Individuals with part-time jobs were not included because it could not be determined from the responses whether they had actually been searching for full-time employment.It is our contention that e V ectiveness criteria of job search may not be comparable between the above excluded respondents and the remaining sample.62E ric E.Brasher and Peter Y.Chenworked at their job for over three years before graduation(9),or were judged to be outliers(13)on one of the nine measures based on the criterion of3.67standard deviations above or below the mean of a particular variable(Tabachnick&Fidell,1989).Among the remaining295respondents,only the 191who remained in their rst job after graduation were included in the nal analysis.MeasuresThe survey included job search records and search success criteria as parts of a large-scale study of di V erent aspects of the search process.Job search records.Participants were rst asked to indicate the month and year in which they began their search when they accepted a job o V er,and their starting salary.Then they reported whether they held this job before graduation and if so,for how long.Search success criteria.The nine success criteria assessed in the current study included starting salary, duration of search,job o V ers,job interviews,job/school match,search stress,job satisfaction,pay satisfaction and intent to quit.Duration of search was operationalized by the di V erence in months between the reported start and end of search.In order to correct overestimates of job o V ers and job interviews resulting from duration of search,we used monthly average job o V ers and job interviews received during job search(i.e.total numbers of interviews and o V ers received divided by duration of search)to assess search success.Job/school match and experience of search strain were assessed on the basis of two6-point items. Response categories for job/school match ranged from‘not at all similar’to‘very similar’,and for experience of search strain varied from‘not at all stressful’to‘very stressful’.Intent to quit was assessed by three items developed by Michaels&Spector(1982).Four items extracted from Spector’s(1985)Job Satisfaction Survey were used to measure pay satisfaction. General job satisfaction was evaluated on the basis of three items employed in Camman,Fichman, Jenkins&Klesh(1979).All items were measured on a6-point response category ranging from ‘disagree strongly’to‘agree strongly’.Coe Y cient alphas reported by these prior researchers were.84 for intent to quit,.75for pay satisfaction,and.91for general satisfaction.ResultsThe results of the study are reported in two parts.First,reliability estimates and correlations among various search criteria are described.After that,results and comparisons among the proposed models that depict structures of search criteria are presented.Reliability estimates of intent to quit,general job satisfaction,and pay satisfaction were all satisfactory,ranging from.81to.91.On the basis of the Hotelling T2(Wilks’=.82,p>.05),there were no signi cant mean di V erences on all relevant variables between respondents graduating in1993 and1994.With a few exceptions,magnitudes of the signi cant relations among success criteria were generally small or modest,which suggested apparent multiple dimensions underlying the construct of search success.As seen in Table1,slightly under half of the correlations among the nine search success criteria were signi cant.Among the primary success criteria,search duration was negatively related to monthly job interviews as well as job o V ers.In other words,the longer time was spent in completing a job search,the fewer interviews or o V ers were received per month.T a b l e 1.R e l i a b i l i t y e s t i m a t e s (a l p h a s )a n d c o r r e l a t i o n s a m o n g s e a r c h s u c c e s s c r i t e r i a*p <.05;**p <.01;t w o -t a i l e d .N o t e s .N =191.N A =n o t a p p l i c a b l e .C o e Y c i e n t a l p h a s a r e i n p a r e n t h e s e s .Success6364E ric E.Brasher and Peter Y.ChenGeneral job satisfaction and pay satisfaction were moderately related to each other.As expected from the satisfaction literature,starting salaries more strongly related to the latter than the former.Job/school match was also positively associated with salary and general job satisfaction.None of the primary success criteria were related to the above criteria.Stressful experience during job search negatively correlated with salary and both satisfaction measures,but not job/school match.With respect to intent to quit,it was associated with stressful search experience,mismatch between job and school training,low starting salary and job dissatisfaction.The structural models of the search criteria were evaluated based on the Bentler–Weeks(Bentler&Weeks,1980)model(EQS v.4.02)with maximum likelihood procedures.No multivariate kurtosis(Mardia,1970)was identi ed.We followed Hoyle&Painter’s(1995)recommendation and employed the following t indices,including2,the non-normed t index(NNFI;Bentler&Bonett,1980), the comparative t index(CFI;Bentler,1990),and the average absolute standard-ized residual.The overall2test assesses the magnitude of the discrepancy between the sample and tted covariance matrices(Hu&Bentler,1995).The non-normed t index and comparative t index are used to compare a restricted model with a null model,based on an assumed distribution for variables.The average absolute standardized residual is an indicator of the correspondence between the observed covariance matrix and the covariance matrix produced by the model.We also inspected the magnitude of residuals to determine if any variable was poorly predicted by the models.Table2presents the standardized path coe Y cients and t indices for the search criterion models tested.The chi-square value for Model1(2(27,N=191)=36.46, p=.11)suggested the model t the data quite well.Additionally,the values for the non-normed(NNFI=.97)and comparative(CFI=.98) t indices were well above the standard cut-o V of.90(Bentler&Bonett,1980).The average absolute standardized residual of.04is also acceptably low.All of the parameters estimated in the model were in the expected directions and signi cantly di V erent from zero. Overall,the various indicators suggest that Model1is a plausible model in describing structural relations among the search criteria.Model2tested additional paths from primary success to salary,match and search stress.The paths from search stress to salary and match were also examined.These paths did not improve on the t of the initial model(gain in2(5)=7.87,p>.05). Although the non-normed(NNFI=.98)and comparative(CFI=.99) t indices are somewhat higher for Model2,the more parsimonious Model1is clearly superior.The mean residual value of.04represents very little change from Model 1.Among the ve additional estimated parameters in Model2,only the path,search stress®salary,was signi cantly di V erent from zero.Therefore,the signi cant path was added to Model1in creating the revised model.This single path did signi cantly improve the t of the initial model(gain in 2(1)=5.43,p<.05).The t indices for the revised model are also somewhat higher than for the previous models(NNFI=.99;CFI=.99).The mean residual of .03is slightly lower than the values for the previous two models.Overall,the revised model appears to t the data very well.DiscussionThere are several important ndings observed from this study.First and most importantly,many of the presumably relevant criteria for job search success are not signi cantly related to each other.For those signi cant correlations,most of the magnitudes were generally small or modest.It appears that job search success is a complex and multidimensional construct.The lack of strong relationships between success criteria suggests that the term ‘search success’should not be used lightly in referring to the criteria studied.The structural analysis is fairly supportive of the predictions regarding potential causal relations between search criteria.All the paths proposed in the initial model were signi cant,as was one of the others tested (i.e.between stress and salary).As in previous research (e.g.Cable &Judge,1996;Rice et al .,1990),satisfaction was positiv ely a V ected by starting salary and match.The anticipated inverse relationship between satisfaction and intent to quit (Michaels &Spector,1982)was also observed.The analyses suggest that starting salary is positively a V ected bymatch,Table 2.Standardized parameter estimates and t indices for search criterion modelsNotes .N =191.The parameters for paths are listed with the destination rst and the source second.Parameters without a value for a particular model were xed at zero.An asterisk is used to indicate non-signi cance of a parameter estimate at the =.05level.Success 6566E ric E.Brasher and Peter Y.Chenperhaps because t is so highly valued by employers(Adkins et al.,1994;Rynes& Gerhart,1990).The nal signi cant path included in the revised model indicated that starting salary was inversely a V ected by search stress,which supported the view of Soelberg(1967).Soelberg proposed that job seekers under stress would accept the rst acceptable o V er rather than trying to maximize their outcomes.As noted above,the causal relationships between primary success and other criteria were either null or weak.These ndings cast some doubts on the wisdom of attempting to maximize job o V ers and interviews if these strategies do not actually result in a better job.It has been noted elsewhere that some job seekers would not have this objective while they focus on a narrow range of jobs and engage in a more concentrated search(Ste V y et al.,1989).We suspected that relations among success criteria may depend on a combination of environmental and individual factors,including nancial need for employment (Sheppard&Bellitsky,1966),possible ranges of job quality indicators within a speci c occupation(e.g.di V erent salary ranges or job characteristics),standard recruiting practices for certain occupations,quali cations,personality,and personal goals and values.Weak relations observed in the current study may be attributed to job seekers’quali cations(e.g.skills,academic performance),academic majors or characteristics(e.g.personality).It is possible that job seekers with higher quali cations are able to obtain higher quality jobs with fewer interviews than those with lower quali cations.Other research has suggested that people with high negative a V ectivity received fewer job interviews(Cook et al.,1995),and tended to work on jobs with less autonomy(Spector,Jex&Chen,1995)which likely resulted in less satisfaction(Hackman&Oldham,1976).Certain background or personality characteristics may also be more appealing to particular employers who view them as evidence of t,which is often used in making hiring decisions(Adkins et al., 1994;Rynes&Gerhart,1990).Moreover,job seekers are often in uenced by factors such as nancial constraints that may prevent them from reaching an optimal choice in terms of job quality(Rynes&Lawler,1983;Ste V y et al.,1989; Wanberg et al.,1996).Personal goals and values also play an important role in job search because job seekers may di V erentially value various search outcomes.According to the rationale of goal-setting theory(Locke&Latham,1990), nding a desired job(i.e.a goal) may direct attention to search behaviours or strategies(e.g.utilizing informal sources)that job seekers believe will lead to the goal,increase persistence or e V ort (Dyer,1973;Schwab et al.,1987),and encourage employing other e V ective strategies.It seems reasonable to assume that individual goals may contribute to variation of search activities in pursuing di V erent search outcomes.It is our contention that a close examination of the goals and values of job seekers is warranted in future research.Search stress is associated with several of the other success criteria(four of eight).It is possible that job seekers who were less successful in their search would look back and evaluate the process as more stressful.An alternative reason is that job seekers who have worked harder in searching for a job remember it as being more stressful or experience greater con ict thinking about whether they chose the right position(Ste V y et al.,1989).Another possibility is that some job seekers。