Céline Rojon ✉
(ITB Consulting GmbH and University of Edinburgh)
(University of London)
This study investigated individuals’ preference structures for workplace coaching providers. Guided by questions about relative weightings of seven important coach(ing) characteristics (i.e., coach work experience/background/gender; coaching training; personal recommendations; client feedback; coaching cost), we carried out a conjoint analysis, using a mixed occupational sample (N = 383). In addition, we conducted linear regression analyses to determine the extent to which coaches’ perceived competence, likeability and trustworthiness might impact on individuals’ decision-making processes. Potential coachees favoured professionally trained coaches with four to ten years’ experience and a similar background to themselves, were female, and charged below average fees. Personal recommendations and existing client feedback further influenced potential consumers’ decision-making. Moreover, perceived competence was highly predictive of potential coach selection.
workplace coaching, conjoint analysis, attributes, relationships,
Accepted for publication: 10 December 2019
31 January 2020
© the Author(s)
Published by Oxford Brookes University
Workplace coaching is a very popular developmental activity used by organisations worldwide (Chartered Institute of Personnel and Development [CIDP], 2015; International Coach Federation [ICF], 2016). The global coaching market is worth approximately 2.4 billion USD annually, with an estimated 53,300 individuals working as professional coaches globally (ICF, 2016). Though coaching practice is still ahead of scientific examination (e.g., McDowall & O’Broin, 2014; Segers, Vloeberghs, Henderickx, & Inceoglu, 2011), evidence gleaned from research is growing. Coaching has been shown to be effective for both personnel development (e.g., CIPD, 2015) and talent management more widely (e.g., ICF, 2016).
Coaching is frequently defined as a collaborative relationship between coach and coachee for the purpose of attaining professional or personal development outcomes which are valued by the coachee (e.g., Spence & Grant, 2007) and which should be demanding and develop an individual’s current capacity or performance (Grant et al., 2010). Our focus here is on workplace coaching, understood as any form of coaching taking place in organisations to bring about individuals’ development (Grant et al., 2010). Coaches can be employed by the organisation itself or self-employed and paid for delivering specific coaching services.
Workplace coaching is seen as a continuously growing field both from academic and practitioner perspectives (Segers et al., 2011). Though first mentioned as early as the 1930s, most relevant academic literature has been published in the last ten years only, suggesting there is scope for much further scholarly enquiry (Grant et al., 2010). In practice, workplace coaching has been gaining popularity approximately since the 1970s. Currently, coaching takes place predominantly in Western economies, particularly in the United States, Australia and the United Kingdom (UK), although its popularity is growing in other European countries including Austria, France, Germany and Sweden (Hamlin, Ellinger, & Beattie, 2008). Meta-analytical research suggests that coaching can have positive effects on individual-level results including enhanced skills and affective characteristics (Jones, Woods, & Guillaume, 2015) as well as a range of other variables including goal-directed self-regulation (Theeboom, Beersma, & van Vianen, 2014).
Despite such encouraging evidence and continued use in practice, many important questions have yet to be addressed in a sufficiently comprehensive fashion, such as why, how and in what ways coaching leads to positive change or in relation to criteria used to evaluate coaching effectiveness (Jones et al., 2015). A further underdeveloped research avenue, which is directly linked to effective coaching practice, concerns individuals’ or organisations’ selection of coaches (Gray, Ekinci, & Goregaokar, 2011): Owing to an abundance of coaching providers/coaches, the potential coaching consumer (individual or organisation) is overwhelmed with choice. Given that coaching can be ineffective (Buljac-Samardzic & van Woerkom, 2015) or even have negative consequences such as uncovering deeper issues that cannot adequately be addressed through coaching (Schermuly, Schermuly-Haupt, Schölmerich, & Rauterberg, 2014), the following questions merit investigation: Based on which criteria or attributes do individuals and organisations select coaches? Are some attributes more important than others? Addressing such questions about selection and alignment is important for advancing coaching research. An ICF survey (2013) provided some insight into what personnel practitioners consider important criteria in choosing coaches, suggesting that recommendations and referrals from practitioners’ own networks are important and coach experience highlighted as key to ‘choosing the right coach’. Whilst other studies (e.g., Gray et al., 2011; Wycherley & Cox, 2008) have also looked into characteristics that may drive (potential) coaching consumers’ decision-making processes, these researchers have focused on executive coaching only, rather than considering the broader realm of workplace coaching in general. Moreover, prior research has examined individual criteria in isolation rather than in conjunction, which is what a person interested in choosing a coach would naturally be more inclined to do (Backhaus, Erichson, Plinke, & Weiber, 2008).
The current study addresses the above questions by examining the preferences of (potential) coaching consumers, in terms of personal, knowledge/experience and general (e.g., coaching cost) attributes impacting on their decision-making in the coach selection process. Conjoint analyses of coach(ing) attributes (e.g., coach gender, coach background, clients’ feedback) were conducted, these providing an insight into the preference structures of coaching consumers, with the aim to sensitize both providers and consumers of coaching and derive recommendations. The following questions guided our approach: (i) What is the relative importance that (potential) coaching consumers ascribe to different coach attributes? (ii) Do (potential) coaching consumers ‘trade off’ between different coaching attributes when evaluating a potential coach?
At this point, it might be useful to note that our research focuses on the initial decision a coaching consumer takes for or against a potential coach, prior to even meeting with them. It was out with the realm of our study to examine what happens once a decision has been made, for instance in the very first meeting a coach and coachee have to become acquainted with one another, the so-called ‘chemistry meeting’.
We addressed our research questions by drawing on participants’ views in relation to a range of stimuli, which were hypothetical coach profiles, to develop broader themes. We utilized conjoint analysis, a technique developed in the 1960s by mathematical psychologists and statisticians R. Duncan Luce and John W. Tukey (Orme, 2010); this enabled an assessment of how potential consumers of coaching might evaluate respective coaches Generally used to examine individuals’ preferences in decision-making, frequently in marketing (Green, Krieger, & Wind, 2001), but also in education (Soutar & Turner, 2002), leadership (Soutar & Ridley, 2008) or entrepreneurship (Lohrke, Holloway, & Woolley, 2010), amongst others, conjoint analysis presents a novel use within the wider field of personnel psychology (Höft, 2005) and specifically within coaching research. We now outline conjoint analysis’ core tenets; relevant methods literature provides further elaboration (e.g., Green et al., 2001).
Conjoint analysis assumes that individuals do not consider characteristics of different objects, offerings or products separately, one-by-one, but rather holistically, and thus individuals’ decision-making processes ought to be examined holistically, too (Backhaus et al., 2008; Beall & Perttula, 1991). Conjoint analysis is a multivariate research method and allows researchers to gain insight into individuals’ preferences based on a ranked order of the objects, offerings or products being examined. It facilitates investigating the influence or impact of a product (or object/offering)’s single characteristics, attributes or parameters on its total utility. Participants are presented with product profiles each representing different combinations of characteristics and are asked to indicate their preferences for these profiles. To provide an illustration of how conjoint analysis might be used in practice, a market research company might wish to determine what type of coffee beverage the public generally prefers. In order to find out, they could run conjoint analyses using data gathered on individuals’ preferences for coffee beverages with a variety of different attributes, such as varying levels of caffeine and milk respectively, the types of beans or sweetening agents used, the size of the beverage (small, medium, large), the price charged and so forth.
Compared to other methods, conjoint analysis’ advantage lies in providing participants with real decision-making scenarios, requiring them to evaluate a combination of characteristics, rather than merely sequentially assessing these, which is traditionally the go-to approach in psychologists’ quantitative research. Thus, the unit of analysis is the overall preference, and within this a potential rank order of important attributes. On the basis of participants’ overall evaluations of multi-attribute alternatives, conjoint analysis enables researchers to calculate the relative importance (score) for each attribute used in the study. As a result, it is possible to discern which attributes influence individuals’ preference evaluations the most. In the case of our study, we sought to investigate which characteristics, and more specifically which facets of these characteristics (e.g., experience as a coach: Do potential coaching consumers prefer coaches that have a little, some or a lot of experience working as a coach?) potential consumers of coaching draw on most in their decision-making regarding who to possibly approach for a coaching session.
Conjoint analysis is conducted following a four-stage procedure (cf. Backhaus et al., 2008), which we outline in the following sections, where we provide specific details of how this method was implemented in our study. Findings obtained from analysing participants’ evaluations of presented profiles provide researchers with the utility of each level within each attribute, as well as each attribute’s importance in individuals’ decision-making processes. Attribute importance is concerned with the relative importance (the importance score) of each attribute. Based on the range of utilities (part-worth scores) for each attribute, which represent their ‘desirability’ (Green et al., 2001), the sum of the ranges of all attributes can be obtained, this enables calculation of each attribute’s importance as a percentage of the sum of ranges across attributes (American Marketing Association, 1992).
Aside from the conjoint analysis, which represents our main approach to addressing our research questions, we also conducted linear regression analyses to determine the extent to which three variables, namely competence, likeability and trustworthiness of each coach might impact on participants’ decision-making processes.
We gathered data from 383 participants, of which approximately half were British (n = 189) and half German (n = 194). Recruitment of these participants was undertaken via Clickworker (https://www.clickworker.com), an online research panel provider. Much like Amazon’s widely used Mechanical Turk (MTurk) which, as previously demonstrated, enables the collection of high-quality data (e.g., Buhrmester, Kwang, & Gosling, 2011; Goodman, Cryder, & Cheema, 2013; Paolacci, Chandler, & Ipeirotis, 2010; Peer, Vosgerau, & Acquisti, 2014), Clickworker facilitated recruiting suitable participants from their pool of over 800,000 individuals; the service is frequently used by organisations and researchers alike. Participant selection criteria were relatively broad – mixed occupational samples of individuals 18 years and older were sought to be representative of the target population. A comparison of sample characteristics in the current study with demographics of actual coaching consumers (coachees) in Western Europe (ICF, 2016) shows that, regarding gender, age and management experience (further information enabling comparison was not available), samples gathered are indeed broadly in line with the target population (Table 1). Comparing demographic characteristics of the German and British participants it was evident that these were very similar; thus, all subsequent analyses were carried out on the complete sample of 383 participants.
a In a recent survey (ICF, 2016), 52% of coaching clients in Western Europe were female. b In a recent survey (ICF, 2016), 59% of coaching clients in Western Europe were below 45 years of age.c In a recent survey (ICF, 2016), coach practitioners indicated that 29% of their clients were managers.
A series of stimulus cards were developed to evaluate participants’ preferences for coach attributes. To reduce the risk of information overload for study participants, conjoint measurements are generally confined to approximately five or six attributes (Green and Srinivasan, 1978). Here, seven attributes with two or three levels each were used (Table 2), developed on the basis of a literature review, which suggested that these were frequent variables in prior coaching and training/development studies and thus likely to be of interest to potential coaching consumers (e.g., Brotman, Libery, & Wasylyshyn, 2000; Gray, 2010; Gray et al., 2011; ICF, 2013; Joo, 2005; Sherman & Freas, 2004; Stern, 2004; Wycherley & Cox, 2007).
Asking participants to review all possible combinations of the seven attributes (TrPr x WEx x BaSi x Ge x PeRe x CFe x CoCo) with their varying levels would have exposed them to a total of 648 stimulus cards (2 x 2 x 3 x 2 x 3 x 3 x 3 = 648 hypothetical combinations), which was clearly too much information to consider. Indeed, Green and Srinivasan (1978) recommend no more than 30 stimuli. Using an orthogonal experimental design enabled us to reduce the number of potential combinations to a manageable set (Green, 1974) with 20 hypothetical profiles as generated by IBM SPSS Statistics (version 22). Based on this orthogonal plan, we developed 20 hypothetical coach profiles featuring information on the seven attributes and their varying levels; for example (please note that for this specific example profile, no information is provided in regards to client feedback (CFe), representing the ‘non-existent’ level for this particular attribute):
Ms. Taylor [Ge] [‘Frau Schneider’ for German participants] has successfully completed an accredited, certified training program in coaching [TrPr] and has been working as a coach since 2012 [WeX]. Before becoming a coach, she worked in a similar function and sector to yourself [BaSi]. One of your colleagues previously talked negatively about Ms. Taylor [PeRe]. The fees for a 60-minute coaching session with Ms. Taylor amount to £200, which is higher than the average coaching fee [CoCo].
The face validity of all 20 coach profiles was examined through a pilot study (N = 6) of materials in both national contexts. Findings were used to improve the wording and user-friendliness of the coach profiles and the online survey per se.
N.B. 1The coach profiles did not specify the particular type of coaching training that had been undertaken by the coaches, nor the training provider; this was left open, suggesting that coaches had (or had not) completed some form of general coaching training. 2 The coach profiles did not provide any further information as to the nature (e.g., coaching outcome or relationship with coach) of the positive or negative client feedback that was (or was not) available to participants, suggesting that feedback could have been available in relation to a number of relevant aspects.
We created a bespoke online questionnaire to gather data. The introduction provided participants with the study’s coaching definition (cf. Grant et al., 2010), to establish a common frame of reference. Participants were then asked to indicate if they had any prior experience with coaching and, if so, to provide information pertaining to their last coach (i.e., coach gender and background): Of those participants who indicated having had prior experience with coaching, coaches had been mostly male (54%), had completed coaching training (62%), typically had between four and ten years’ work experience as a coach (39%) and had previously mostly worked in a similar function and sector to the participant themselves (46%).
Next, participants were instructed to imagine that they were currently seeking to find a coach to facilitate their addressing a variety of personal and/or professional issues. They were further told that they had asked an assistant to perform some background research on potential coaches, the result of which, namely several coach profiles, would be presented to them on the following survey pages. When presented with each of the 20 hypothetical coach profiles, participants then indicated the likelihood of them arranging an initial (no-obligation) coaching session on a scale ranging from zero (very unlikely) to one hundred (very likely).
Participants were further asked to evaluate the competence, likeability and trustworthiness of each coach on five-point Likert scales (e.g., incompetent; rather incompetent; neither incompetent nor competent; rather competent; competent), using one question per variable per coach (e.g., In your view, how trustworthy is this coach?). These three variables were included based on previous research (Hodgetts, 2002) showing them to be important factors in coaching and coaching relationships. Their inclusion might therefore enable us to estimate, using regression analyses, the extent to which participants’ notions of a coach’s competence, likeability and trustworthiness might explain their likelihood for choosing to arrange initial contact with a particular coach. Following the online evaluation of all twenty hypothetical coach profiles, participants provided demographic background information including gender, age, national identity, highest qualification and work experience. Completion of the questionnaire took between 15 and 20 minutes.
This section presents findings of the conjoint analysis, specifically its part-worth and relative importance scores. We further present here results of linear regression analyses, undertaken to examine the degree to which the variables competence, likeability and trustworthiness are related to, and can predict, participants’ coach preferences (see Table 3 for descriptive statistics of study variables).
Part-worth scores (PWS) were obtained for each level of the seven attributes (Table 4). They suggested the following in relation to our research questions: Female coaches were favoured over male ones (coach gender (Ge)). A coach who had successfully completed a coaching training program was seen as more desirable than not having completed one (coaching training (TrPr)). Four to ten years’ work experience as a coach (WEx) were preferred over more (i.e., more than ten years) or less experience (i.e., less than four years). Participants favoured coaches who worked in a similar function and sector to themselves (coach background (BaSi)). Regarding personal recommendations (PeRe) and client feedback (CFe), coaches in the positive condition ranked highest, those for which no information on recommendations and feedback was available came next, coaches with negative recommendations and negative client feedback ranking lowest. Participants preferred coaches who charged average fees coaching cost (CoCo) for their services.
Next, drawing on the PWS for the overall sample, we obtained a rank ordering of the 20 coach profiles – the higher a coach profile’s score, the higher its ranking. Ordering the profiles from best to worst, profile 14 (Ms. Robinson) ranked highest across the two subgroups, whilst profile 2 (Ms. Jones) ranked lowest. The three highest and lowest ranked profiles are presented in Table 5.
N.B. For some of the above profiles, fewer than seven attributes seem present – this is the case where no client feedback (CFe) or personal recommendations (PeRe) have been made available, which represents one of the three levels (positive/negative/none) for those particular attributes.
We further obtained relative importance scores for each of the seven coaching attributes, based on attribute levels’ relative ranges. Findings (Table 4) suggested that participants’ evaluations of coach profiles are mostly impacted by personal recommendations (PeRe) and client feedback (CFe) respectively, indicating that these are the two most important aspects for participants regarding the probability of their booking a session with a particular coach. The least important attributes for participants were the coach’s gender (Ge) and their amount of prior work experience (WEx), suggesting that these attributes only had a small impact on whether or not participants would choose to book a session with a particular coach.
In addition to the conjoint analyses, we conducted linear regression analyses to examine the predictive validity of participants’ perceptions of coach competence, likeability and trustworthiness (independent variables) for their likelihood to arrange a session with a coach (dependent variable). Regression statistics were obtained for the three coach profiles that ranked highest (i.e., coaches Ms. Robinson, Ms. Smith and Ms. White; Table 5). In other words, the extent to which Ms. Robinson’s (and, respectively, Ms. Smith’s and Ms. White’s) competence, likeability and trustworthiness scores could predict scores for the likelihood of choosing to arrange an initial coaching session with her (and the other two highest ranked coaches, respectively) was predicted.
Generally speaking, findings across the three highest ranked coaches suggest that scores of all three independent variables are predictive of probability scores for scheduling coaching sessions, with competence being a highly significant predictor, followed by trustworthiness, likeability having the least impact. Specifically, in the case of coach Robinson, the probability of arranging a coaching session can be accounted for, to 30%, by their perceived competence, trustworthiness and likeability (F (3, 376) = 54.92, p < .001, R2 = .30); the most important predictors are competence (β = .38; p < .001) and trustworthiness (β = .16; p < .05). For coach White, 33% of the variance in the dependent variable (i.e., probability of scheduling a session with this coach) is accounted for by the three independent variables (i.e., perceived competence, trustworthiness and likeability) (F (3, 377) = 63.30, p < .001, R2 = .33), the most important predictors being, again, competence (β = .44; p < .001) and trustworthiness (β = .12; p < .10). Finally, the probability of scheduling a session with coach Smith can be explained (to 30%) by their perceived competence, trustworthiness and likeability (F (3, 378) = 55.55, p < .001, R2 = .30), with the most important predictors, once again, being the former two variables (competence: β = .33; p < .001; trustworthiness: β = .17; p < .05).
The current study evaluated the relative importance of coach(ing) attributes to potential coaching consumers to determine which characteristics may influence future coaching contracting. By means of conjoint analysis, we estimated the extent to which participants value seven coach/coaching characteristics (i.e., coaching training; work experience as coach; coach background; coach gender; personal recommendations; client feedback; coaching cost) identified on the basis of their importance and relevance as demonstrated in existing coaching, training and development and wider social psychology literature.
On the basis of our analyses, it can be concluded that personal recommendations and client feedback were, by a large margin, the two most important factors determining coach selection. More specifically, coaches with positive personal recommendations and/or positive client feedback were evaluated much more favourably than those with negative or no feedback/recommendations. Interestingly, these two attributes were the only ‘soft’ ones amongst the seven study attributes, relating to previous clients’ experiences with a coach’s services; ‘hard’ attributes of coaching, including coaching cost and coaching training, evidently playing a less important role. The finding that (positive) recommendations/referrals and feedback are very important determinants of coach selection is also evident in previous research (ICF, 2013), which came to the same conclusion, albeit judged from the perspective of personnel practitioners, rather than potential consumers as in the current study.
Coach gender and amount of prior work experience as a coach were the two least important attributes for potential coaching clients here, both factors only playing a marginal role regarding their impact on coach attractiveness. However, whilst for potential consumers of coaching neither similarity of coach background (to coachee background) nor experience working as a coach (in years) were particularly important precursors of coach selection, these factors were seen to be key by personnel practitioners (ICF, 2013). This discrepancy in findings may indicate differences between organisations and individuals in the decision-making process, suggesting that personnel practitioners would do well to be aware of what employees may value most – and least – in a potential coach, to facilitate commitment to the coaching process (Gray et al., 2011).
Based on the PWS, a ‘most preferred coach’ profile can be developed: This person has successfully completed an accredited, certified coaching training program and has four to ten years’ experience working as a coach. Having previously worked in a similar function/sector to her potential client, this coach comes recommended by colleagues and has received positive feedback from previous clients. The session fees this coach charges are lower than the average coaching fee and they are female (although gender was less important than other attributes). Conversely, a coach is seen to be least attractive to potential consumers when they are male, are lacking specific coaching training and have been working as a coach for less than four years. Such a coach’s prior background is different from that of his potential client and feedback from previous clients has been negative, as has been talk from colleagues when referring to this coach. The fees this coach charges for one session are above average.
Results from regression analyses highlighted that perceived coach competence, trustworthiness and likeability are predictive of whether or not a coach will be selected by potential coaching consumers. This finding appears to sit with Hodgetts (2002) who suggested that personal coach variables, particularly their perceived competence and trustworthiness, can have an important impact on the establishment of a (positive) coach-coachee relationship, as our findings indicate that such factors may also influence initial contracting.
Firstly, we offer a methodological contribution as the use of conjoint analysis has provided a more holistic and, as such, arguably more realistic outlook on relevant coach attributes compared to previous studies (e.g., Gray et al., 2011), which have considered such attributes separately, in isolation from one another. The current study went beyond lists of coach attributes as produced by previous studies (e.g., ICF, 2013), identifying statistically significant key attributes and taking a more rounded approach as facilitated by conjoint analysis. To the best of our knowledge, the current study is the first conjoint analysis to have been conducted within the coaching domain. Even within the wider training and development area, this multivariate statistical technique has seen very little use only. Given, however, that it has great potential for personnel psychology research, as demonstrated here, fellow researchers are encouraged to consider it further for future studies.
Our second contribution is the sampling strategy and study framing. With one notable exception (Gray et al., 2011), most relevant prior studies identified important coach attributes from the perspective of coaches only rather than focusing on the perspective of potential coaching consumers as in our approach. Arguably, adding this additional viewpoint, gathered from those that are meant to be benefitting from coaching, its customers, is a crucial step in gaining a comprehensive insight into coaching preference structures. Besides, the current findings have implications beyond the realm of coaching for training and development more widely, given that the attributes studied here are also very often relevant for training/development settings, such as for mentoring.
The study findings have several potential practical implications for the different coaching stakeholders, namely coaches and coaching consumers including coachee and organisations using coaching, as well as coaching institutes (e.g., those offering access to coaches and/or coaching training). Individuals involved in the setting up of coaching could be drawing on the findings to pre-select coaches for potential coaching consumers. One clear implication is that relevant coaching training is a ‘must’ – so initial screening should deselect those without relevant qualifications, regardless of their other experience, fee structure or other characteristics. We acknowledge that our qualifications variable was binary only; it would be a rich avenue for future investigation to test out if the type of training coaches have undertaken makes a difference to initial selection and further contracting.
The second implication is for organisations to collect relevant coach characteristic data in comparable formats to facilitate informed choices, but then also offer a framework for further evaluation. Our study only went as far as initial choice and contracting, but it would be a crucial next step to evaluate whether for instance ‘coach-coachee background matching’ has any implications for long-term effectiveness, as research appears equivocal, indicating few effects of matching (Bozer, Joo, & Santora, 2015). Nevertheless, whilst making explicit and gathering such data appears imperative, the importance of chemistry meetings to determine if the coach-coachee relationship will be mutually successful should not be underestimated (Boyce, Jackson, & Neal, 2010).
In addition, the findings might have important implications for self-employed coaches and for coaching institutes, in that they offer an insight into coachees’ decision-making processes regarding those attributes that are most and least valued when considering coach profiles. They suggest, for example, that highlighting positive experiences by previous clients is likely to be more effective for marketing purposes than referring to a coach’s extensive amount of experience. This is an interesting observation, which might also be worth investigating further. If all coach profiles had contained positive recommendations (so held constant), would this attribute cease to matter? It was a consistent finding in our study that ‘soft’ attributes mattered more than ‘hard’ ones, although it is also of interest to note that likeability mattered less than perceived competence or trustworthiness. That a coach came with a positive personal recommendation may not necessarily indicate that they were any more effective than coaches who do not make such information available. Thus, our findings indicate that coaches are unlikely to benefit from a modest experienced-based approach in how they position themselves but would do well to gain specific client feedback on their competence and how they build trust and use this for promotion.
We acknowledge that we did not consider the full range of possible attributes that might act as precursors to coach selection. Whilst an examination of more than seven attributes would have been desirable, the limit was set to this number, which required participants to evaluate 20 coach profiles, to minimize potential boredom, tiredness and cognitive overload – had just two more attributes been included, participants would have been required to evaluate almost twice the number of coach profiles (k = 38). Thus, building on the study findings, future studies could usefully examine further possible attributes, such as psychological ones (e.g., creating rapport; showing empathy), given that such ‘soft’ factors should not be underestimated (Gray et al., 2011), this being corroborated, too, by the ‘soft’ attributes used here (i.e., personal recommendations; client feedback) having been found to be highly relevant for coach selection. A potential study progression might be to use filmed ‘talking heads’, rather than brief profiles to introduce a more realistic element and then ask participants for more differentiated participant ratings.
Besides, data being collected through an online survey and drawing on a convenience sample from two countries and made available by an online panel provider, it was not possible to exert much control over participants; this may pose a threat to the study’s external validity (McBride, 2010) and may have introduced some sampling error. Whilst demographics from our sample were broadly representative of the target population for workplace coaching, it would be advisable for future research to replicate the study with different (types of) samples, thus increasing the potential for generalizing study findings. For example, future research might consider using samples with more management responsibility to determine the extent to which findings are applicable to executive coaching; it may be the case, for instance, that executive level coachees prefer to work with coaches who have more than 10 years’ professional experience (rather than up to 10 years, which was the case here).
We also acknowledge that there may be instances where providers of workplace coaching are chosen by organisational representatives (e.g., personnel practitioners) rather than by coachees themselves; in such cases, even though individual coachees may have personal preferences, their decision-making power may be limited. Moreover, the three variables competence, likeability and trustworthiness may be more prevalent during or after the aforementioned chemistry meeting with a new coach, rather than in the preceding coach screening and selection stage that our study was concerned with. As such, future research could usefully draw out to what extent an initial decision for or against a particular coach might differ from a coachee’s impressions following a chemistry meeting and the consequences of such discrepancies on the subsequent coaching process and its success. That being said, as aforementioned, our research complements existing studies that have examined the preferences of personnel practitioners and coaches, providing additional insight into precursors that matter for those benefitting from coaching.
Moreover, only quantitative data about potential coachees’ preferences for coach attributes was gathered, disregarding qualitative information such as, for instance, participants’ motives and reasons for providing certain probability scores (e.g., Why are female coaches favoured over male ones?). Supplementing the quantitative findings with such qualitative data, for instance by asking coachees what their criteria and rationales were for preferring one coach over another, thus represents a further avenue for future research. Similarly, the three criteria coach likeability, trustworthiness and competence would benefit from additional investigation through qualitative research, to further understand how they may impact on coachees’ decision-making processes.
To conclude, we highlight the importance of taking a holistic, consumer-centred (rather than coach or personnel practitioner-centred) perspective in examining precursors of coach selection. Our study has gone beyond mere lists of coaching attributes as developed in past research, emphasizing instead the complex interplay of attributes that individuals consider in their decision-making processes, with ‘soft’ attributes such as personal recommendations and client feedback playing a particularly important role. Both our study design, which manipulated coach profiles, and the analytic approach through conjoint analysis, offer opportunities for future research to further our understanding of how coaches are selected and coaching is contracted.
Céline Rojon is a HR Senior Consultant at ITB Consulting, Germany, having previously been a Lecturer in HRM at the University of Edinburgh Business School/UK.
Nicole Bode obtained her PhD in Psychology from the University of Erfurt/Germany. Her research interests pertain to leadership and professional development, focusing on coaching.
Almuth McDowall is Professor of Organisational Psychology at Birkbeck University of London/UK. Her expertise spans several topics in organisational/ occupational psychology and personnel psychology.
RADAR: Research Archive and Digital Asset RepositoryAbout RADAR