You are on page 1of 12

Available online at www.sciencedirect.

com

Journal of Vocational Behavior 73 (2008) 112 www.elsevier.com/locate/jvb

Self-perceived employability: Construction and initial validation of a scale for university students
Andrew Rothwell
a

a,*

, Ian Herbert b, Frances Rothwell

Centre for Studies in Higher Education, Coventry University, Priory Street, Coventry, Warwickshire CV1 5FB, UK b The Business School, Loughborough University, Loughborough, Leicestershire LE11 3TU, UK c School of Education, Nottingham Trent University, Clifton Campus, Nottingham NG11 8NS, UK Received 19 July 2007 Available online 15 December 2007

Abstract This paper reports the development of a research instrument to examine the expectations and self-perceptions of employability of business undergraduates (bachelor degree students) in three UK universities, and examines relationships with other measures. The scale was found to have a satisfactory internal reliability coecient for an untested measure (a = .75) and had discriminant validity from other new measures such as university commitment. Internal and external dimensions were identied, the latter relating to the perceptions of the strength of the university brand, the state of the external labour market, and the demand for the degree subject. Perceptions of future employability were surprisingly modest, notably for those in arguably the strongest position, suggesting that these self-perceptions may be a product of relative societal expectations, consistent with positional conict theory. 2007 Elsevier Inc. All rights reserved.
Keywords: Employability; University commitment; Graduate employment; Positional conict theory

1. Introduction In this study employability was examined from the perspective of individuals. That is, what individuals seeking a particular type of work believe their chances of success are, and what factors inuence their perceptions. This sets the study apart from much of the extant work on employability, which primarily focuses on the impact of government policy, organisational HR strategies, mid-career workers, wider society and educationalists (Bowers-Brown & Harvey, 2004; Forrier & Sels, 2003; Hillage & Pollard, 1998; Yorke, 2006). There are a number of denitions of employability, but few of these were appropriate for the present study. Hillage and Pollard (1998, p. 24) suggested employability is the ability to realise potential through sustainable employment. Undergraduate job-seekers might be expected to aspire to some level of continuous employment to
Corresponding author. Fax: +44 02476887599. E-mail addresses: andrew.rothwell@coventry.ac.uk (A. Rothwell), i.p.herbert@lboro.ac.uk (I. Herbert), frances.rothwell@ntu.ac.uk (F. Rothwell). 0001-8791/$ - see front matter 2007 Elsevier Inc. All rights reserved. doi:10.1016/j.jvb.2007.12.001
*

A. Rothwell et al. / Journal of Vocational Behavior 73 (2008) 112

establish a career following the acquisition of a degree. Thus for a bachelor degree students, self-perceived employability is dened as the perceived ability to attain sustainable employment appropriate to ones qualication level. Previous studies have suggested that employability is a multi-faceted construct (Forrier & Sels, 2003) with both internal and external dimensions (Kirschenbaum & Mano-Negrin, 1999; Rajan, 1997; Rothwell & Arnold, 2007). Internal factors for the job hunter include vocational or job-related knowledge and skills, and mastery of job search (Hillage & Pollard, 1998), together with the potential to learn (Lane, Puri, Cleverly, Wylie, & Rajan, 2000). As this is a study of self-perceptions, these were encapsulated under the heading of self-belief. The prevailing state of the external labour market may also be signicant in terms of perceptions of employability (Kirschenbaum & Mano-Negrin, 1999; Lane et al., 2000; Rajan, 1997). This has been included as a second major dimension of the employability scale constructed in the present study. Fugate, Kinicki, and Ashforth (2004, p. 15) considered employability as a psycho-social construct which subsumes a host of person-centred constructs as a synergistic combination of career identity, personal adaptability, and social and human capital. They suggested that employability was (p. 28) a more variegated construct than for example career motivation, personal initiative or proactive personality, and in fact subsumed each of these variables. Fugate et al. suggested that employability helped individuals cope with work transitions in a turbulent employment market. In this respect their conception of employability is similar to that of Rothwell and Arnold (2007), in that it is a future-oriented perspective to do with individuals and their ability to proactively address the challenges of the labour market. Both of these studies used employed individuals. Van der Heijde and Van der Heijden (2006) also developed an employee-based individual operationalization of employability, with dimensions including occupational expertise, anticipation and optimization (in short, creatively preparing for future job changes), personal exibility, and balance. In the latter case the researchers refer to the balance between individual and employer needs, drawing on exchange theory and thus developing conceptual similarities to the psychological contract (Rousseau, 1995). The notion that individuals might proactively prepare for job changes is arguably most relevant to those already in employment. However since the emergence of the employability curriculum in UK universities, an individual evaluation of the impact of this could usefully be incorporated into future studies. Unfortunately Van der Heijde and Van der Heijden (2006) had not been available at the time the present study was being prepared. In developing a measure of subjective employability for undergraduates, the present study incorporated a further dimension; that is the impact that the reputation (brand image) of the university attended might have lent on a students perception. Bainbridge (2000) found that the brand of an existing employer on a resume credibility to individuals job seeking potential. It was believed that there might be a similar eect in line with the popular ranking of universities following Murray and Robinsons suggestion (2001, p. 140) that: There is strong evidence that large-scale recruiters of graduates target a limited range of universities. This component of employability was included as my university in the new instrument. The chosen eld of study was also included as potentially impacting on self-perceptions of employability. The demand for individuals with degrees in particular subjects varies, not just according to market conditions but also to perceived status differences between vocational areas. To some extent perceptions of future demand by employers manifest themselves during the annual application process, with students trying to anticipate future trends in the vocational attractiveness of individual subjects (BBC, 2006). Because the students in the present study were not yet in employment a measure of ambition was incorporated as a proxy for perceptions of future success to test discriminant validity. Fugate et al. (2004) argued that employability is a psycho-socially constructed concept and thus a measure of commitment that would relate the present study to the lived experiences of our respondents as university students was developed. This was based on the measure of organisational commitment identied by Mowday, Steers, and Porter (1979, p. 225). A positive correlation between university commitment and self-perceived employability was expected, to the extent that both constructs reect an orientation towards high performance, the rst as an input, the second as an outcome. Positional conict theory (Strathdee, 2003) suggests that there might also be relationships between various biographic measures and employability. While there was some degree of homogeneity in the sample to the extent that subjects of the present study were all full-time, second year, business students, studying in the same region of the UK (and the data were collected in the same semester) they were from three quite dierent uni-

A. Rothwell et al. / Journal of Vocational Behavior 73 (2008) 112

versities selected across the ranked league table produced by the Times Higher Education Supplement (THES). At the time of the data collection (2006) the three universities in the sample were ranked in the rst, third and fourth quartiles of the THES table, respectively, and thus could be expected to have diering levels of reputational capital in the eyes of students and employers. The two lower ranked institutions were new (1992) universities having formerly been a polytechnic and college of higher education, respectively. The third institution is highly regarded within the ranks of older (pre-1992) universities and, whilst achieving university status later than some of its peer group, is now most associated with the premier universities in the UK. Names and precise rankings have not been used to preserve anonymity. For analysis purposes below, the highest ranked university is referred to as University A, the former polytechnic as University B, and the former college of higher education as University C. Given the suggestion that employers favour more highly rated universities (Murray & Robinson, 2001), one might reasonably expect that students in University A will believe themselves to be more employable, all other things being equal, than students at the two lower ranking universities; consistent with positional conict theory. The competitive edge of the higher ranking university is likely to be reinforced by the higher credentials required for undergraduate entry as one might expect a more highly rated university to recruit students with higher grade entry points. Thus grade scores on entry should also correlate with self-perceptions of employability. Finally, because Universities B and C promote what is known in the UK as a widening participation agenda (HEFCE, 2007a), it was believed that there may be dierences by social class between the three universities, possibly impacting on employability scores. In the absence of any other social class measure, respondents were asked to identify their parents occupation. Thus the specic aims of this study of self-perceptions of this study was to construct and initially validate a measure of self-perceived employability for bachelor degree students. To do this, we identied the constructs that make up employability for undergraduates and operationally dened them with psychometric items. Next, we examined whether scale scores were distinguishable from ambition and university commitment. Then we identied the extent to which a range of biographical variables correlated with self-perceived employability, including university attended, level of prior educational attainment, age, gender and social class. 2. Method To produce an overall measure of employability, four components were represented in a matrix (Fig. 1). Based on the literature review, it was suspected that these components would not exist in isolation, so the corners of the matrix were intended to represent their interactions. Each cell in the matrix in turn represented the interaction of two of the four components of employability, and each cell generated two questions. Thus for

Fig. 1. Student self-perceived employability.

A. Rothwell et al. / Journal of Vocational Behavior 73 (2008) 112

example, cell 2 related to students perception of their universitys reputation (brand strength), contextualised in employment terms (Bainbridge, 2000). Two questions per cell was regarded as a limit as this in turn produced a 16-item scale that was long enough to have every chance of good internal reliability, without overloading the respondents or making the overall survey instrument to long. It was decided at the research planning stage to present ambition at the centre of the matrix, more as a reection of its perceived conceptual proximity to self-belief than an attempt to represent it as any sort of central construct. Although the sample in this study comprised UK business undergraduates, the scale did not include any subject or culturally specic items and could thus be used in a range of contexts. The 16-items scale (Appendix A) were presented to respondents in the order shown. Responses were on a ve-point Lickert-style scale with anchors strongly disagree (1) to strongly agree (5), this also being used for the ambition and university commitment scales. The alpha internal reliability coecient was .75. Means and standard deviations are shown in Table 3. In the absence of a scale for ambition, one was developed, with six items (Appendix B). Two items were adapted from scales of subjective career success. The rst, I want to be in a position to do mostly work which I really like, was adapted from Nabi (2001, p. 464) the second from Greenhaus, Parasuraman, and Wormley (1990, p. 86), I am satised with the progress I have made meeting my goals for the development of new skills. Four other items were generated by the authors, examples being I have clear goals for what I want to achieve in life and the reverse-scored what I do in the future isnt really important. The alpha internal reliability coecient for this scale was .60. The scale mean score for the six items was 12.00 (low) considering it could have been as high as 30, suggesting that either our sample were not very ambitious, or the measure was weak (or both). The standard deviation for the extended scale was 2.80, and the internal reliability coecient .76. University commitment was measured using eight items selected and adapted from the nine-item organizational commitment scale identied by Tsui, Pearce, and Tripoli (1997) as relating to aective commitment. The word university was substituted for organization. One item was excluded (in the original: I am willing to put in eort beyond the norm for the success of this organization) as it was perceived as inappropriate. Examples of items used included I am proud to tell others I am at this university and I am extremely glad I chose this university over others I was considering at the time I joined. One item, I would have accepted almost any type of course oer in order to come to this University, was removed at the analysis stage, as data screening revealed that this item skewed the results and had an adverse eect on the internal reliability. For the remaining seven items, the alpha coecient was .87. For comparison purposes, some demographic variables were considered. These included the university attended (coded after collection), gender, age (in ve categories), and qualication level on entry (by points tari). All respondents were in their second year of full-time study, as part of a three or four year undergraduate (bachelor) business or business-related honors degree. Students undertaking a four year program would normally spend the third year on work experience (internship). 3. Procedure and participants Four hundred questionnaires were distributed, in class time, in three separate universities in April 2006. All respondents were given the same powerpoint presentation explaining the research, time in the class to complete the questionnaires (typically 1215 min), and were assured of data condentiality. Three hundred and fty one responses were collected. Of these, 142 (40.5%) were from University A, 124 (35.3%) from University B, and 85 (24.2%) from University C. Only 4.3% of respondents were 25 or over and could be categorized as mature students. Following data screening seven responses were rejected as containing signicant outliers, thus the analyses were undertaken on 344 cases. This was regarded as an adequate sample size, after Henson and Roberts (2006). 4. Results A series of principal components analyses (PCA) were conducted to explore the dimensions of employability and the (potentially) related measures. This was conrmatory factor analysis driven, as Henson and Rob-

A. Rothwell et al. / Journal of Vocational Behavior 73 (2008) 112

erts had suggested (2006, p. 395) by theoretical expectations regarding the structure of the data. Taking a broad perspective in the rst instance, the three scales were (all together) subjected to a PCA (Pallant, 2007, p.179). Data were tested for suitability for PCA. The correlation matrix was inspected for correlation coecients of .3 or above, and these were found in a moderate number of instances. Other indicators were more reassuring. The ratio of items to respondents was 11.9:1. The KaiserMeyerOlkin measure of sampling adequacy was .83, above the recommended value of .6 (Pallant, 2007, p.182), and Bartletts test of sphericity was signicant at p < .001. Kaisers criterion (eigenvalues of one or more) extracted eight components. Inspection of the scree plot revealed there was a clear break after the fourth component. The decision as to the number of factors to extract was seen as crucial. According to Hayton, Allen, and Scarpello (2004, p. 192), extracting too few can result in distorted loadings due to factors that are not included in the model, and falsely loading on the factors that have been included. Specifying too many factors leads to a further range of problems including factors that are dicult to replicate. To further inform our decision making, and following Pallant (2007, p. 183), parallel analysis was undertaken using the Monte Carlo PCA. Henson and Roberts (2006) suggested that this was the most accurate predictor of the number of factors to extract. Comparing the randomly generated eigenvalues with those from the PCA revealed that in four cases the actual eigenvalue exceeded the criterion eigenvalue, thus the maximum number of factors that should be extracted in the full 29-item PCA was four. Having specied a four-factor solution, and using varimax rotation, the four components accounted, respectively, for 20.18, 9.64, 6.18 and 5.90 percent of the variance. The solution is presented as Table 1. The rst component consisted of the seven items from the university commitment scale. There was just one convergent item above the .3 cut-o criterion. This indicated a relatively weak relationship to component

Table 1 Four-factor solution, varimax rotation, combined scales for employability, ambition and university commitment Component 1 Unicom6 Unicom8 Unicom7 Unicom1 Unicom4 Unicom5 Unicom3 Emp 4a Emp 3a Emp 3b Emp 2b Emp 5b Emp 2a Emp 5a Emp 4b Emp 6a Ambition Ambition Emp 1b Ambition Ambition Ambition Emp 1a Emp 8a Emp 8b Emp 7a Emp 6b Emp 7b Ambition .800 .797 .752 .715 .712 .627 .615 2 3 4

.331

.714 .712 .652 .598 .553 .480 .469 .419 .410 4 3 5 6-ve 2 .699 .639 .630 .570 .517 .494 .308 .748 .656 .500 .488 .480 1

A. Rothwell et al. / Journal of Vocational Behavior 73 (2008) 112

2, which drew items from cells 2, 3, 4, 5 and part of 6 from the self-perceived employability matrix. These items related to the universitys brand strength, the reputation the university had in the eld of study, the status and credibility of the eld of study, the external labour markets demand for the subject, with a modest inuence from the state of the external labour market generally. Component 3 related principally to ambition, and included the items from cell 1 concerned with engagement with study and academic performance. Component 4 related to items from cells 7 and 8 of the matrix (self-condence and awareness of opportunities) with a further hint of the external labour market (cell 6). The results revealed quite substantial, but not total support, for the discriminant validity of the self-perceived employability scale. To a large extent, component 1 represents university commitment, component 2 self-perceived external employability, component 3 ambition, and component 4 self-perceived internal employability. The literature review had indicated that employability itself may have internal and external components: (Hillage & Pollard, 1998; Rajan, 1997; Van der Heijden, 2002). Previous empirical work, albeit with a professional sample (Rothwell & Arnold, 2007), had conrmed this. Because this sample was not in employment, this internalexternal split was believed to relate more to their individual self-belief, and their perceptions of the external labour market, respectively, and seemed likely to incorporate ambition in the former element. Internal employability in this case appears to relate to the participants perception of their own (internal) attributes, skills, abilities, engagement with study, and ambition.The distinctiveness of the employability scale, with ambition in particular, now appeared somewhat questionable, so the relationship of the employability scale with the two other scales was examined in more detail. Principal components analysis was undertaken on all 22 items of the combined employability and ambition scale. As one might expect, the loadings were very similar to those for the 4-component, 29-variable solution shown in Table 1. Further information is available from the corresponding author.

5. Internal and external employability To explore the internalexternal distinction further, and to explore the potential for a rened measure that might have practical applications such as in career guidance, subscales were created for internal and external employability, to force a two-factor solution, based on items previously identied as having weightings of over .3. Because of the strong relationship suggested between internal employability and ambition, the internal employability subscale now included most of the ambition scale items except A1, I want to be in a position
Table 2 Two-factor solution, employability and ambition combined scale, varimax rotation Component 1 Emp 3a Emp 3b Emp 4a Emp 2b Emp 5b Emp 2a Emp 5a Emp 6a Emp 4b Emp 6b Ambition Ambition Ambition Ambition Emp 8a Emp 1b Ambition Emp 8b Emp 1a .726 .699 .691 .661 .560 .549 .485 .437 .426 .328 4 3 5 2 .763 .657 .548 .512 .507 .447 .432 .405 .357 2

6-ve

A. Rothwell et al. / Journal of Vocational Behavior 73 (2008) 112

to do mostly work which I really like, which had previously failed to make the .3 cut-o criterion. The question items derived from cell 7 were omitted for the same reason. The results are displayed in Table 2. The alpha coecient for the new external employability subscale was identical to that for the 22-item whole-scales of employability plus ambition, at .76, whilst for the internal employability subscale it was .66. While the employability subscale scores are acceptable for an untested measure, in future studies the relatively weaker ambition measure will be rened or replaced, thus potentially improving the reliability of the internal employability scale as well. Overall, the internalexternal distinction was present. 6. Descriptive statistics and scale inter-correlations Descriptive statistics and scale inter-correlations are presented in Table 3. For the original 16-item employability scale, the mean was only just on the mid-point at 2.51, and the standard deviation was .84. There were variations between universities, which are presented in Table 4. The modest results, and the dispersion, appear to be characteristics of this sample. The commitment measure also had a similarly modest mean, and a similar standard deviation, yet this was developed from a well established and widely tested measure, albeit in other contexts such as organizational commitment or professional commitment. It was in respect of internal employability and ambition that the most modest scores were found, and the mean (2.15) of that subscale was lower than the mean of 2.63 for the external employability subscale at a statistically signicant level [t(342) = 38.54, p < .001]. The outliers were notably A1, I want to be in a position to do mostly work which I really like (mean = 1.72), and the reverse-scored item A6 what I do in the future isnt really important (mean = 1.71). However deleting both of these items from the 22-item scale would have made only a marginal dierence to the alpha coecient in one instance (.76 and .77, respectively), while deleting A6 from the internal employability/ambition combined subscale made no dierence to the alpha coecient whatsoever. With the exception of the internal employability/ambition combined subscale (a .66), the internal reliability of all scales used in this study was good. For the 16-item employability scale, the alpha coecient could be improved marginally by deleting two of the items (to three decimal places, the scale a = .747, and by deleting item E1B it could be improved to a = .751, or by deleting E8b, improved to a = .750). For an untested scale the results were considered acceptable. For the overall employability, external employability and university commitment measures, scores appeared to be reliable, to show the sort of variation between individuals one might reasonably expect, and had no particular oor or ceiling eects. The internal and external employability subscales correlated at only .192. There is however the caveat that internal employability, for this sample at least is qualitatively dierent to what it might be for individuals more established in their careers.
Table 3 Scale descriptive statistics and inter-correlations (number of items in brackets), alpha reliability coecients are on the diagonal Mean 1. 2. 3. 4. 5.
*

SD .84 .83 .87 .84 .97

1 .075 .955* .920* .404* .447*

2 .76 .845* .565* .428*

Self-perceived employability (16) Employability/ambition all items (22) External employability (10) Internal employability/ambition (7) University commitment (7) Statistically signicant at p < .001.

2.51 2.37 2.63 2.15 2.59

.76 .192* .517*

.66 .036

.87

Table 4 Descriptive statistics for total self-perceived employability split by university Statistic Mean 95% condence interval Lower bound Upper bound Median Standard deviation Skewness University A 36.47 35.50 37.45 37.00 5.82 .212 University B 42.91 41.98 43.84 43.00 5.11 .318 University C 42.45 41.37 43.54 43.00 5.01 .175

A. Rothwell et al. / Journal of Vocational Behavior 73 (2008) 112

7. The psychometric properties of the employability scale The nal research question was to report the psychometric qualities of the (employability) scale derived from the samples perceptions. Earlier investigations had indicated that there were two components to employability, however the original matrix included four dimensions. A nal set of exploratory principal components analyses were undertaken, with a four-factor solution requested. Using varimax rotation, four fairly distinct components were identied, which in total explained 48.99% of the variance in the employability scale. The rst, which explained 16.28% of the variance, was subject-related, and drew items from the three cells on the matrix being to do with the universitys reputation in the eld of study, the status and credibility of the eld of study, and the external labor markets demand for it. The second, which explained 12.61% of the variance drew items from two cells on that arguably are to do with outward-facing aspects of employability: the strength of the universitys brand, and the state of the external labor market. The third component, which explained 12.1% of the variance, drew items from cells in the matrix to do with individual attributes, namely condence in ones own skills and abilities, and the awareness of opportunities in the labor market. Finally the fourth component, which explained just 8.1% of the variance, was related to the individuals perception of their engagement with studies and academic performance. The rotation converged in seven iterations, although in each case there was a much stronger loading on one component than the other. This analysis suggests that for this sample as a whole employability, when considered on its own, may be most to do with choice of subject, then the inuence of the university brand and the labor market, then skills and market-awareness, and least of all to do with their perception of the adequacy of their engagement with study. 8. Employability and the relationship with demographic variables Having explored the dataset to the extent that relationships between scales had become clearer, and subscales identied, the relationships between the employability scale (and others) and some of the demographic variables were considered. First, scale responses across the sample were examined. Table 4 shows that the mean score for employability was lower for University A than for Universities B and C, despite A being among the highest rated in the UK league tables. A one-way (between groups) ANOVA was conducted which found that dierence in means between University A and Universities B and C was statistically signicant [F(2, 340) = 55.84, p < .001]. The actual dierence in mean was modest, and the eect size was small (g2 = .25) although as Pallant noted (2007, p. 247), with a large enough sample small dierences can become statistically signicant. In this case the dierence appeared large enough to be worthy of note. There was no statistically signicant dierence between the mean scores for Universities B and C. A further surprising result was in respect of entry qualications reported. It had been expected, due to the prestige of the Institution, that University As students would have a much higher qualication prole on entry than respondents at University B or C. This was only really evident in one category, where around 12% more of them reported that they had between 140 and 200 UCAS (Universities and Colleges Admissions Service) points. However this is a fairly modest score in a UK context and their average prole had been expected to be higher. Only .7% of University A respondents were in the highest UCAS tari category, although 10.7% identied themselves as having other qualications. Because Universities B and C promote what is known in the UK as a widening participation agenda (Higher Education Funding Council for England, 2007b), thus aiming to encourage more working-class students to go to university, it was expected that more students from University A might claim to come from a professional or managerial background. This was the case (86.4% as compared to 74% and 71% in Universities B and C, respectively), however only modest numbers claimed to have parents in the other categories, for example only 5% of the whole sample identied skilled workers as their parents occupation, with only one individual at University A. There were no statistically signicant dierences in mean employability score, within universities, between categories. While the modest numbers inhibited further analysis it is believed that the impact of social class on perceptions of employability, when considered by type of university, may be worthy of further investigation with a larger sample. To better understand where these variances had come from the demographic variables were investigated further, beginning with the mean employability score for each university split by gender. The dierence

A. Rothwell et al. / Journal of Vocational Behavior 73 (2008) 112

between University A and Universities B and C appeared markedly dierent. A one-way between groups analysis of variance with planned comparisons was conducted. Equal variances were assumed as Levenes test (p = .259) was not signicant. The contrast specied between University A, and Universities B and C, was statistically signicantly dierent for males [F(1, 340) = 108.58, p < .001]. Further examination was conducted of the mean employability score for each university split by age group (details available from the authors). It appeared as though while younger respondents at University A reported lower scores than at Universities B and C, it was the oldest University A respondents who report the very lowest scores of all. The same respondents also reported much lower levels of university commitment. In previous studies (eg. Rothwell & Arnold, 2007; Van der Heijden, 2002) it had been the older male respondents who tend to have the lowest self-perceived employability. Comparison was problematic with the present study as previous researchers had focused on individuals in employment with a much more diverse age range, while in the present study there were relatively few mature students. Next, dierences in employability scale scores attributable to university attended, age and gender were considered. Social class was omitted from this analysis due to the similarities in scores within universities between the categories. Specically, if university, age and gender were controlled for, would the set of variables (university commitment, ambition) still be able to predict a signicant amount of the variance in self-perceived employability? A multiple hierarchical regression was conducted, where model 1 included the university attended, age, entry qualications and gender; while model 2 added ambition and university commitment. Model 1 explained 19.4% of the variance in employability, with university attended (p < .001) and age (p = .014) being the only variables in model 1 making a statistically signicant unique contribution. From model 2, ambition and university commitment explained an additional 13.3% of the variance, both being statistically signicant, with similar beta coecients (ambition = .272, university commitment = .255). The unstandardized coecients were ambition (b = .571, SE = .95) and university commitment (b = .255, SE = .55). Finally consideration was given as to whether there might be other confounding variables, albeit ones that had not been anticipated in the original research design. Data on student satisfaction are collected nationally in the UK, and reported on the Higher Education Funding Councils Teaching Quality website (2007b). Examination of the latest results for business degrees for the three universities in question (for which data was reported) revealed that the mean score on the seven measures reported for University A was 4.04, while for Universities B and C the scores were 3.66, and 3.76, respectively. This suggests, tentatively, that it was not likely to be dissatisfaction with the course that was interfering with the results for University A, but that this relationship could be something worth investigating in subsequent studies. 9. Conclusions The 16-item self-perceived employability scale was based on a matrix composed of inuences on employability derived from a literature review. The overall scale showed an acceptable internal reliability. Factor analyses (Table 1) suggested that self-perceived employability is distinguishable from, though quite strongly (considering the sample size) correlated with, ambition and university commitment. These latter two (new) scales met with mixed success. The ambition scale had only adequate internal reliability (a = .60), and was arguably too closely related to internal employability. Future studies should incorporate a more robust measure, which may have potential applications either on its own or combined with the internal employability items. University commitment fared better, and this may have potential to be an important measure of student engagement. Exploration of the factor structure of the employability scale was driven both by theoretical insights derived from the employability literature, and data analysis including PCA parallel analysis. In the employability subscales, external employability was found to be the more robust measure, while what has been labelled internal employability in this study related to skills and self-condence and may therefore be more closely related to beliefs of personal ecacy. This internal aspect is worthy of further investigation. An important aspect of this investigation was to address the issue of how undergraduates perceived their future employability. Given that the mean responses to the scale were only just above the mid-point (2.51),

10

A. Rothwell et al. / Journal of Vocational Behavior 73 (2008) 112

expectations at least for this sample appear to be modest, more so for those individuals at the highest rated university, which was counter-intuitive. The groups appeared to be reasonably similar in prole, but it may be that in this study students in the highest rated university had higher expectations in terms of job level or employer brand, and recognised the challenges they faced. For students at Universities B and C, typically the rst in their families to go to university, that achievement in itself may have been a signicant result for them, to the extent that their desire for a graduate job came second, or they may have had less realistic expectations. Because the university commitment scale showed a more normal distribution, and because the original data were checked thoroughly, we believe these ndings to be accurate. In future, a better denition of precisely what the respondents are aiming for might explain the dierences in scores on the employability scale. The relationships with demographic variables were non-signicant on the whole. The research design could have been improved by using focus groups to discuss the unexpected outcomes, however this was not possible with this sample due to the timing of the survey. The study had a number of further limitations. As self-perceived employability was a self-report measure, how realistically might some individuals report the fact that, privately, they did not expect to do very well? The research appears to be vindicated in this respect in that those individuals with arguably the strongest university brand were the least condent in their employability, and also reported a much lower score on the university commitment scale, albeit without a statistically signicantly dierent result on the ambition scale. The measure of employability developed in this study related to how well the sample expected to do, rather than actual attainment. Future studies might include a longitudinal dimension over a two year (to test initial attainment of appropriate employment) and ve year (to test sustainability of employment) timeframe. It is suggested that this research represents an important rst step in investigating how undergraduates themselves feel about their future employment prospects. Theoretically, it begins to ll the gap in relation to understanding the relationships between internal factors, such as the individuals skill-set and their application to their studies, and external factors such as the general state of the labour market, the strength of the university brand, and the demand for particular subject areas. Practically, the employability scale ts well with contemporary developments in the UK in relation to employability skills in the curriculum and personal development planning. It could be developed into a tool to have potential application in individual undergraduate careers counselling, in vocational guidance, or institutionally for groups of students in relation to evaluating the employability skills curriculum. The scale has potential for development in other cultural contexts as well as a range of subject areas. The university commitment scale also shows good potential for future development. Employability is a dynamic area, high on the contemporary Higher Education agenda in the UK, with signicant potential for further research and development. Appendix A. Employability scale items (scale item numbers correspond to cell numbers in Fig. 1) 1a. 1b. 2a. 2b. 3a. 3b. 4a. 4b. 5a. 5b. 6a. 6b. 7a. 7b. 8a. 8b. I achieve high grades in relation to my studies I regard my academic work as top priority Employers are eager to employ graduates from my university The status of this university is a signicant asset to me in job seeking Employers specically target this university in order to recruit individuals from my subject area(s) My university has an outstanding reputation in my eld(s) of study A lot more people apply for my degree than there are places available My chosen subject(s) rank(s) highly in terms of social status People in the career I am aiming for are in high demand in the external labour market My degree is seen as leading to a specic career that is generally perceived as highly desirable There is generally a strong demand for graduates at the present time There are plenty of job vacancies in the geographical area where I am looking I can easily nd out about opportunities in my chosen eld The skills and abilities that I possess are what employers are looking for I am generally condent of success in job Interviews and selection events I feel I could get any job so long as my skills and experience are reasonably relevant

A. Rothwell et al. / Journal of Vocational Behavior 73 (2008) 112

11

Appendix B. Ambition scale items A1. A2. A3. A4. A5. A6. I want to be in a position to do mostly work which I really like I am satised with the progress I have made meeting my goals for the development of new skills I have clear goals for what I want to achieve in life I regard myself as highly ambitious I feel it is urgent that I get on with my career development What I do in the future is not really important (reverse-scored)

Appendix C. University commitment scale items UC1. UC2. UC3. UC4. UC5. UC6. UC7. UC8. I talk up this university to my friends as a great university to be at (I would have accepted almost any type of course oer in order to come to this university1) I nd that my values and this universitys values are very similar I am proud to tell others that I am at this university Being at this university really inspires the best in me in the way of study performance I am extremely glad I chose this university over others I was considering at the time I joined I really care about this university and its future For me this is the best of all universities to be a member of

References
Bainbridge, B. (2000). The futures bright, the futures branded. Recruitment Portfolio, 512. British Broad Casting Company (BBC) Online (2006). Debate over employable degrees. Available from http://news.bbc.co.uk/1/hi/ education/4719222.stm. Bowers-Brown, T., & Harvey, L. (2004). Are there two many graduates in the UK?. Industry and Higher Education 12, 243254. Forrier, A., & Sels, L. (2003). The concept of employability: A complex mosaic. International Journal of Human Resources Development and Management, 3(2), 102124. Fugate, M., Kinicki, A. J., & Ashforth, B. E. (2004). Employability: A psycho-social construct, its dimensions, and applications. Journal of Vocational Behavior, 65, 1438. Greenhaus, J. H., Parasuraman, S., & Wormley, W. M. (1990). Eects of race on organizational experiences, job performance evaluations and career outcomes. Academy of Management Journal, 33(1), 6486. Hayton, J. C., Allen, D. G., & Scarpello, V. (2004). Factor retention decisions in exploratory factor analysis: A tutorial on parallel analysis. Organizational Research Methods, 7, 191205. Henson, R. K., & Roberts, J. K. (2006). Use of exploratory factor analysis in published research. Educational and Psychological Measurement, 66(3), 393416. Higher Education Funding Council (HEFCE) (2007a). Widening Participation Web Page. Available from http://www.hefce.ac.uk/widen/. Higher Education Funding Council (HEFCE) (2007b). Teaching Quality/National Student Survey Website. Available from http:// www.hefce.ac.uk/learning/nss/. Hillage, J., & Pollard, E. (1998). Employability: Developing a framework for policy analysis, EfEE Research Brieng No.85. Institute for Employment Studies. Available from http://www.dfes.gov.uk/research/programmeofresearch/projectinformation.cfm?projectid= 12855&resultspage=1. Kirschenbaum, A., & Mano-Negrin, R. (1999). Underlying labour market dimensions of opportunities: The case of employee turnover. Human Relations, 52(10), 12331255. Lane, D., Puri, A., Cleverly, P., Wylie, R., & Rajan, A. (2000). Employability: Bridging the gap between rhetoric and reality; second report: Employees Perspective. London: Create Consultancy/Professional Development Foundation. Mowday, R., Steers, R., & Porter, L. (1979). The measurement of organizational commitment. Journal of Vocational Behaviour, 14, 224247. Murray, S., & Robinson, H. (2001). Graduates into salesemployer, student and university perspectives. Education and Training, 43(3), 139144. Nabi, G. R. (2001). The relationship between HRM, social support, and subjective career success among men and women. International Journal of Manpower, 22(5), 457474.

Deleted item.

12

A. Rothwell et al. / Journal of Vocational Behavior 73 (2008) 112

Pallant, J. (2007). SPSS survival manual (3rd ed.). Buckingham: Open University Press. Rajan, A. (1997). Employability in the nance sector: Rhetoric vs. Reality. Human Resource Management Journal, 7(1), 6778. Rothwell, A., & Arnold, J. (2007). Self-perceived employability: Development and validation of a scale. Personnel Review, 36(1), 2341. Rousseau, D. M. (1995). Psychological contracts in organizations: Understanding written and unwritten agreements. Thousand Oaks, CA: Sage. Strathdee, R. (2003). The qualications framework in New Zealand: Reproducing existing inequalities or disrupting the positional conict for credentials. Journal of Education and Work, 16(2), 147164. Tsui, A. S., Pearce, J. L., & Tripoli, A. M. (1997). Alternative approaches to the employee-organisation relationship: Does investment in employees pay o? Academy of Management Journal, 40(5), 10891121. Van der Heijden, B. (2002). Prerequisites to guarantee life-long employability. Personnel Review, 31(1), 4461. Van der Heijde, C. M., & Van der Heijden, B. I. J. M. (2006). A competence-based and multidimensional operationalization and measurement of employability. Human Resource Management, 3, 449476. Yorke, M. (2006). Employability in higher education: What it iswhat it is not. ESECT Learning and Employability Series (Vol. 1). York: Higher Education Academy.

You might also like