Intending to work part-time: the influence on
employability
Taru van der Zijde 11159588
2018
Master Thesis
2
Statement of Originality
This document is written by Taru Anneli van der Zijde who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
3
Intending to work part-time: the influence on
employability
ABSTRACT
Existing research has shown that part-timers are faced with disadvantages in their career outcomes. This paper complements the literature by investigating the influence of expressing the willingness to work part-time on the employability of an applicant, by sending fictitious applications to 149 real job ads in the Netherlands. The differences found in callback rates turn out insignificant, suggesting that the different part-time requests do not influence employability. Employers do not treat applicants who show the intention to work part-time differently than applicants who do not show this intention and do not treat applicants who provide a reason related to high opportunity costs of work different compared to applicants who do not provide a reason for part-time work. Contradicting the expectations, part-timers appear not to be faced with penalties during the first step of the selection process.
1. Introduction
Compared to the rest of Europe, the Netherlands has the highest percentage of part-time workers with its 46.60% (Eurostat, 2018). Multiple reasons can be addressed for why such a big proportion works part-time. Working part-time for example enables individuals to combine a career with educational training, caring tasks and leisure (Román, 2006). Although working part-time can have its benefits for an employee, as for example less turnover intention and the possibility of combining care and work (Almer & Kaplan, 2002), there is also a downside associated with part-time work. Part-part-timers are faced with a lower social-economic status, receive lower pay, face fewer career opportunities, receive less firm-sponsored training and are promoted less often compared to full-timers (Picchio & Van Ours, 2016; Román, 2006).
These described penalties for working time, can occur throughout the career of a part-timer. Based on the presence of these penalties, the question arises whether individuals that express the willingness to work part-time are also disadvantaged during the process of searching for a job
4
and starting a career. Do applicants who voice the intention to work part-time get selected less often compared to applicants who do not express this intention? So far, little to no attention has been paid in the literature to this process of selecting. Therefore, this thesis contributes to the literature by examining possible penalties for applicants who express the willingness to work part-time during the first step of the selection process, namely getting a callback after applying for a job.
It is important to analyze whether applicants who voice the willingness to work part-time receive less callbacks than applicants who do not express this intention but possess completely identical skills, as it can result in organizations missing out on capable workers. When part-time applicants receive less callbacks than identical applicants who do not state anything about the number of working hours, a specific group of applicants is excluded from being considered a candidate for the job regardless of their capabilities. The employer will ignore good candidates who can contribute and be valuable to the organization, resulting in missing out on new talents and possible successes for the organization.
Besides missing out on good employees, the possible attitude of selecting part-timers less often can also have an important influence on the individuals who want to work part-time. When part-time applicants are less likely to be invited for a job interview, they will be faced with a more difficult job search and will start their career with a backlog on top of the penalties they are likely to face during their career as part-timers. Part-time workers will be disadvantaged in even more ways, which can lead to more difficulties in their careers and personal lives. Being selected less often, might also decrease the feeling of having a feasible option to engage in part-time work as finding a job becomes harder. This may result in a frustrated workforce and workers deciding to opt out, who otherwise would have participated in the labor market (White, 2016). Williams (2001) finds support for this reasoning in her article, as she finds that American parents decide not to work part-time, although they have a preference for it, because of the fear of being marginalized by a culture that emphasizes full-timers.
Moreover, it is important to examine possible penalties during the process of hiring since it can affect the female-male balance on the work floor. Because women form the majority of part-timeworkers, 75.8% in the Netherlands (Eurostat, 2018), the possible attitude of employers to select part-timers less often can result in decreasing the chances of hiring women specifically. These reduced chances will result in less females participating on the work floor, on top of women being
5
already underrepresented in business life (Centraal Bureau voor de Statistiek, 2017). The diversity on the work floor and the culture within the firm might be affected, which can result in an employer missing out on certain benefits that come with a diverse workforce, as for example increasing sales and profitability (White, 2016). The decreased chances of women getting selected, will also reduce the likelihood of women making it to the top of the corporate hierarchy, making women even more underrepresented in top positions (Elliott & Stead, 2018). These processes entail that women in particular will have more troubles with achieving a successful career and will be more likely to fall behind, as they work part-time much more often than men (White, 2016).
These described negative effects of the possible attitude of selecting part-timers less often on the personal career advancement, can have a major impact at the aggregate level as almost half of the Dutch working population works part-time. A social problem might arise according to Román (2006), which calls for more in depth attention. Therefore, it is relevant to study whether applicants who voice the intention to work part-time are faced with a penalty during the selection process. This thesis aims to provide these insights by examining the relationship between expressing the intention to work part-time and the employability of applicants. The research question is as follows:
“Does the intention to work part-time affect employability?”
The research question is answered by means of a field experiment, where fictitious job applications are send to job adds. Within the applications send, the content of the cover letter differs with respect to requesting part-time work or not. In order to get more detailed information about the effect of the desire to work part-time on employability, a distinction is made between expressing the intention to work part-time with providing a reason for this preference and without providing a reason. As a result, there are three different treatments accompanied by three different cover letters. In the control cover letter, the applicant does not voice the willingness to work part-time, in the second cover letter the applicant does but without stating a reason, and in the last cover letter the applicant expresses the intention to work part-time and states a reason for this. The abilities of the applicant are held constant throughout the treatments, to be able to see whether employers treat applicants who want to work part-time differently than applicants who do not show this intention during the application procedure. The reactions of employers are measured by tracking callback rates within the different treatments.
6
Conducting this field experiment and creating new data, enables investigating the effect of expressing the willingness to work part-time on the applicant’s employability and possible penalties for wanting to work part-time during the selection process. In total, 149 applications are send to employment ads in the Netherlands mostly in the fields of finance, consultancy and information technology.
Unexpectedly, the empirical results do not support the idea that expressing the intention to work part-time influences the employability of the applicant as all results appear insignificant. The insignificant differences found in callback rates between the treatments do appear to be in the expected direction: applicants who show the intention to work part-time have a lower probability of receiving a callback compared to applicants who do not show this intention and applicants who show the intention to work part-time without providing a reason have a lower probability of receiving a callback compared to applicants who provide a reason related to high opportunity costs of work for the willingness to work part-time. As the results are not significant, this paper suggests that employers do not treat part-time applicants differently compared to applicants who do not express the willingness to work part-time during the selection process. Part-time applicants do not seem to face a penalty during their search for a job.
The rest of this paper is set-up as follows. In section 1 related literature about working part-time and the selection process is discussed, as well as the contribution of this thesis to the literature. Section 2 gives an overview of the experimental design, whereas section 3 presents the results of the field experiment conducted. Lastly, in section 4 the main findings, conclusions and possibilities for future research are discussed.
2. Literature
As described in the introduction, almost half of the Dutch working population works part-time, even though working part-time seems to have a downside. This downside of part-time work has received a lot of attention in the literature: multiple studies have investigated the different treatments part-timers face in comparison to full-timers. The overall conclusion that flows from these studies, is that working part-time has a negative effect on one’s income and labor market position (Román, 2006). These negative effects come into play via different ways, with part-time work being undervalued structurally (Thornley, 2007). Before examining these downsides of working part-time in detail, a definition of part-time work is established.
7
Part-time work entails an arrangement of employment characterized by structurally working fewer hours per week compared to what is established as full-time (Fine-Davis et al., 2004; Román, 2006). What is constituted as full-time hours however, differs per organization, sector and country. This thesis classifies working part-time as working 4 days a week.
Now that the definition of part-time working is formulated, the negative effects associated with part-time working compared to full-time working can be analyzed. Firstly, a disadvantage of working part-time can be found in the salary. Part-time work is characterized by a negative wage effect, whereby part-timers are disproportionately situated at the bottom of the wage hierarchy and in low wage sectors (Conolly & Gregory, 2008; Fernández-Kranz et al., 2015; Hirsch, 2005; Matteazzi et al., 2014). Bardasi and Gornick (2000) for example show that part-timers are subjected to an 8-22% wage penalty compared to full-timers, using cross-nationally comparable data of five European countries. The study of Aaronson and French (2004) argues that a decline in working hours from 40 to 20 hours, decreases the wage with 25% for American men. This wage difference between full-timers and part-timers also appears in the United Kingdom and Spain, in the UK switching from full-time to part-time work results in a 32% wage drop for women and in Spain in a 20 log percent loss of hourly earnings (Conolly & Gregory, 2009; Fernández-Kranz & Rodríguez-Planas, 2011). These studies indicate that the decision to work part-time is accompanied by a wage penalty.
The study of Paul (2016) based on German data also confirms this wage penalty, as she finds that working part-time has a negative long-term effect on wages. The specific effect on current wages of part-timers however, is dependent on the number of hours worked. The results of Paul (2016) suggest that working fewer hours comes with a large immediate effect on current wages, although more extensive part-time work does not reduce current wages. Garnero et al. (2014) also find that the wage penalty inflicted on women becomes higher as the working week becomes shorter. Together these studies conclude that there is an unequal pay difference between part-timers and full-timers, which highlights the undervaluation of part-timers (Thornley, 2007).
Part-timers are also much more likely to work in lower function jobs, they seem to be concentrated in jobs in the lowest part of the job hierarchy. According to Thornley (2007) occupations with the highest proportions of part-timeworkers correspond strongly to jobs with the highest proportions of working-class workers. Román (2006) indeed finds that workers have a lower occupational status when they work part-time or have worked part-time in the past, which
8
negatively affects their socio-economic status. She suggests that part-time work is not compatible with higher level functions. Connolly and Gregory (2008;2009) support this idea based on British longitudinal survey data from the years 1991-2001, as they find evidence for occupational downgrading when women switch from working full-time to part-time. Between 14-25% of the women that make the switch, move to an occupation where the average qualification level is lower than that of their previous full-time job and the salary is lower. Returning to full-time work does not provide full recovery from this loss in earnings, even when the occupational downgrading is reversed.
Besides the lower occupational status, female part-timers are also at disadvantage when it comes to security in the form of contractual conditions and irregularity of working hours. Thornley (2007) finds that in the UK over one in five part-timers is employed on a fixed term or temporary contract, in contrast to the 93% of full-timers who are employed on a permanent contract. These less advantageous agreements for timers are also prevalent during worktime. Female part-timers are more likely than full-part-timers to work on weekends and holidays and have more split/mixed shifts, meaning that they are more likely to work irregular and unsocial hours.
But part-timers are not only faced with different contractual conditions, they also appear to be provided with different career opportunities compared to full-timers. Part-timers seem to be offered less opportunities on the work floor, Goldin (2014) for example finds that part-time lawyers in a large law firm have less access to rich clients compared to full-timers. Even lawyers that switch from being a part-timer to a full-timer, only start gaining access to rich clients slowly with the process being spread out over several years. These findings indicate that there is a penalty for past part-time work, even though employees have started working full-time again. In the report of the Employment Issues Committee (2000) 43% of the respondents of 100 law firms also address that their substantive work assignments were affected by their decision to work part-time, indicating that they are given different opportunities.
These different career opportunities for part-timers can also be found in the likelihood of receiving training. Picchio and Van Ours (2016) and Backes-Gellner et al. (2014) show that men are less likely to receive firm-sponsored training with any reduction in working hours compared to full-time hours. For women they found a training disadvantage with a reduction in working hours between 20-49% compared to working full-time.
9
Part-timers are also at a disadvantage when it comes to promotions. Multiple studies conclude that part-timers are less likely to be hired for important positions and are less often selected for promotions (Cohen and Single, 2001). Kornberger et al. (2010) for example conclude that the representation of flex-work program employees in senior management roles decreased after this flex-work program was implemented in a Big 4 firm. Besides receiving less promotion, part-timers also face a disadvantage in their performance evaluations as they are subjected to lower evaluations (Johnson et al., 2008; White, 2016).
Finally, part-time work seems to have negative consequences on a part-timer’s long-term relationship with his or her manager. Johnson et al. (2008) find that supervisors were less inclined to engage part-timers in future assignments. This idea of part-time work having a negative effect on relationships, is supported by the study of the Employment Issues Committee (2000). Of the respondents, 30 to 40% reported that their relationship with partners and associates weakened after starting to work part-time. Cohen and Single (2001) also observe that supervisors show less interest in engaging in a future commitment with part-timers.
All these described downsides imply that part-timers are penalized during their career for working part-time and are faced with disadvantages. The question rises whether part-timers also face a penalty earlier in the career process, namely when applying for a job. Especially since part-timers are less often chosen for promotions and future engagements, it is interesting to investigate whether they are also less often considered for selection. As this has not been researched so far, this thesis contributes to the existing literature.
For an organization, selecting competent employees during the application procedure is critical for enhancing competitive advantages (Ployhart & Holtz, 2008). Recruiters try to select high performing workers, but it is difficult for an employer to know the true performance of a potential worker at the time of hiring. This makes the selection decision an uncertain one, which results in employers trying to predict the future performance of applicants by certain indicators. During the selection, attention is paid to for example work experience and education level of applicants, which can be seen as verifiable indicators of achievements (Aigner & Cain, 1977; Román, 2006; Spence, 1973).
Although employers prefer to collect reliable information, this quest can be costly and not always realizable. Employers therefore, fall back on subjective assessments of the applicant’s skills in order to get a more complete image of the applicant’s productivity level. Recruiters will assess
10
certain characteristics of the applicant, for example age, gender, and race, when they believe these are indicators for predicting productivity. This subjective assessment creates room for statistical discrimination. Statistical discrimination refers to a situation where a recruiter bases its selection decision on proxies, because of a lack of information on an applicant’s productivity. These proxies are not on the level of the individual, but take place at the group level as the decision of the employer is based on ‘previous statistical experiences’ with members of the specific group related to the proxy. The underlying assumption is that the applicant’s productivity is equal to the average productivity of the relevant group he or she belongs to.
A statistical discriminated group is plagued by the assumption that they have a lower productivity on average. The group is seen as an economic risk by the employer, which can result in discriminating against the certain group in different ways. A group that is assumed to contain more varieties of productivity levels than usual, is also seen as a risk for organizations when selecting new employees (Román, 2006).
Part-timers may be seen as such a discriminated group (Román, 2006). A recruiter is uncertain about the motive of the applicant for wanting to work part-time, which makes the productivity level of the applicant more noisy and difficult to establish than normal. The applicant might sort into part-time work because of high opportunity costs of work, meaning he/she has a valuable outside option, or because of high effort costs, meaning that the worker has a low productivity level (Delfgaauw & Dur, 2007; Delfgaauw & Dur, 2008). These different motives result in a distinction of two types of part-time applicants. When the employer cannot distinguish between the different types, he might see the intention to work part-time as an indicator for a lack of commitment and reliability and establish the part-time applicant as untrustworthy in terms of employment (Román, 2006).
Several studies indicate that this negative perception of part-timeworkers is present within firms, regardless of the reason for working part-time. Working reduced hours is perceived as a signal of lower commitment to the organization and profession, on top of that part-timers are considered less serious about their career (Fine-Davis et al., 2004; Himelstein & Forest, 1997; Johnson et al., 2008; Picchio & Van Ours, 2016). Part-timers are assumed to have less dedication and drive than full-time workers (White, 2016). The Employment Issues Committee (2000) also identified that part-time lawyers are subjected to skepticism regarding their professional commitment, regardless of their seniority level. Thornley (2007) concludes in her study that
part-11
time workers are subjected to widespread prejudices around undervaluation and lack of commitment, although part-timers in local government services do not actually confirm to this stereotype of being an uncommitted worker. She finds that these workers exhibit similar levels of commitment as full-timers and have a high level of responsibility.
Besides this perception of a lack of commitment, employers might assume that part-timers have a different disutility from work compared to full-timers since part-time workers face more external constraints. Having an outside option to engage in next to work, might draw on an individual’s concentration and energy during work. This might have a negative impact on the quality of the work of part-timers, making them less attractive compared to full-timers in the eyes of the employer (Delfgaauw & Dur, 2008; White, 2016). Because of these possible effects of having more external constraints, employers might assume that the productivity level of part-timers is lower compared to full-timers.
Finally, aspects of organizational culture might also be a possible reason for the negative perception of part-time workers. Organizational culture is seen as the combination of shared beliefs, values and assumptions by employees about appropriate behavior, that provides guidelines for organizational decisions. Gender theory argues that the dominant organizational culture of traditional firms, especially in the finance sector, reflects a “male-typed” world view, as this was the prevalent culture during the founding of most organizations in the beginning of the 20th century. This culture results in value and reward systems being subjected to a male-typed cultural bias.
Within the male-typed culture an ideal worker is a worker who has few responsibilities for family and caring tasks, meaning he/she can engage in working full-time. Traditional workplace standards, like working routinely long hours and fixed working times, are idealized as essentials for organizational success. Part-timers as a result, may be viewed as not compatible with the prevailing ethic of organizational success. These traditional standards are also seen as essentials for positive organizational citizenship behavior (OCB), which reflects employee performance that exceeds formal expectations, with for example working additional hours. As part-timers might be assumed to lack the essential norms of OCB, their commitment to the organization might be perceived as lower (Johnson et al., 2008).
Part-timers might also be characterized as more “feminine” rather than “masculine”, with “feminine” reflecting supportive and nurturing values and “masculine” reflecting assertive values (Cohen & Single, 2001). These assertive values are appraised more than compassionate values in
12
the male oriented organizational culture. As a result, part-time working might conflict with the traditional view of organizational culture. Johnson et al. (2008) indeed conclude that flex working is viewed as inconsistent with the traditional full-time workplace culture and is destructive to one’s career successes. This implies that culturally-based biases can possible be an underlying reason for a negative image of part-timers.
So, as described above, there seems to be a negative perception of part-time workers. Because of the more questionable commitment level and energy level, employers might categorize part-time workers as a risky group to select from (Román, 2006). As recruiters are faced with incomplete information about the productivity level of an applicant during the application process, they might use the request for part-time working as a proxy for the productivity of the applicant. Using this as a proxy entails that the part-time applicant will be subjected to the negative associations adhered to part-timers, since part-timers are assumed to be less committed, engage in less OCB, and have less energy for work. As employers want to avoid risks as much as possible, they might decide to hire a part-time applicant or invite them for a job interview less often as a result. Based on this line of reasoning, in combination with the specific findings that part-timers are less often selected for promotions and future engagements, the expectation is that applicants who express the willingness to work part-time are less often invited for a job interview and thus receive less callbacks compared to applicants who do not express this willingness.
With this expectation the question arises whether the applicant can influence the possible negative reasoning of the employer and the uncertainty perceived by the employer. As described, the recruiter is uncertain about the reason of the applicant for wanting to work part-time and does not know which type the applicant reflects. Does the applicant have a low productivity, or does he/she simply have high opportunity costs of work? Because of this additional uncertainty, there is more room for subjective assessments from employers and the formation of certain believes about the part-time applicant.
When employers review the application of the part-time applicant, they form an initial belief about the applicant’s productivity after reading the request to work part-time. Based on the literature reviewed, it is expected that this belief is less positive compared to the beliefs about applicants who do not show the intention to work part-time. This belief however, can be updated when the employer receives additional information about the applicant. The applicant can provide a reason for wanting to work part-time during the application process, in order to try and help the
13
employer with distinguishing the different types of part-time applicants and the specific type of the applicant. When the applicant expresses that he/she wants to work because of high opportunity costs of work, the recruiters receive additional information about the productivity of the applicant. The applicant shows that he/she has a high valuable outside option as a motive for working part-time and tries to influence the belief of the employer that he/she wants to work part-part-time because of a low productivity level.
This provision of extra information does not mean that the employer can now determine the exact level of productivity, the employer still cannot observe the actual productivity level of the applicant and the available information about the productivity stays incomplete. The information does, however, reduce the range of possible productivity levels of the applicant or at least gives some direction about the productivity level (Delfgaauw & Dur, 2007; Poinas et al., 2012). With this information, the employer can update his initial belief about the productivity of the part-time applicant. The additional information can reduce the uncertainty perceived by employers, which can lead to less perceived risks of selecting the part-time applicant compared to the situation where a part-time applicant does not provide a reason for wanting to work part-time. As a result, the expectation is that applicants who show the intention to work part-time and provide a valuable outside option as the reason for working part-time receive more callbacks compared to applicants who voice the willingness to work part-time without providing a reason, as they provide information that enables updating the initial belief of employers.
3. Experimental design
In this section the experimental design adhered to the field experiment is presented and predictions for the effects of the different treatments are derived. The statistical tests used to test the hypotheses are also covered.
A. Treatments
In order to examine whether the intention to work part-time affects employability, a field experiment is conducted where fictitious applications are send to job adds with a variation in the content of the cover letter. For the content of the cover letter, there are three different treatments that vary with respect to part-time working. Firstly, in the control treatment nothing is mentioned about part-time working in the cover letter. In the part-time treatment the applicant expresses a willingness to work part-time but does not state a reason why this is the case. In the last treatment,
14
the part-time + reason treatment, the applicant expresses a willingness to work part-time and provides a reason for this. Table 1 shows an overview of the different treatments. With these different treatments the effect of mentioning the intention to work part-time on the employability of an applicant is investigated.
Table 1 – Different treatments
Treatment name Request to work part-time Reason for part-time work
Control No -
Part-time Yes None
Part-time + reason Yes Yoga teacher
Note: The table reports the different characteristics of the treatments with respect to part-time work.
As stated earlier, an employer does not know whether a worker expresses the willingness to work part-time because of a high valuable outside option or because of high effort costs and this might make the employer reluctant to invite a part-time applicant. This thesis wants to examine whether providing extra information about the reason for the desire to work part-time influences the number of callbacks an applicant receives. To investigate this, a reason that reflects high opportunity costs of work is used for the willingness to work part-time.
More specifically, teaching yoga classes on a voluntary basis is chosen. Teaching yoga classes means that the applicant has high opportunity costs of work, as he/she has a valuable outside option to engage in during his/her free time. The employer does not have to worry about the applicant spending working hours on the outside option with this reason, since the activity cannot be executed during worktime. This reduces the chances of employers having an additional argument against hiring the part-time applicant. Besides that, the choice for a voluntary basis is made because this makes it possible to avoid a connection to an existing yoga school and it makes the applicant long-term dependent on the finances of the job applied for.
B. Fictitious resume
To examine the difference in employability, a fictitious applicant is created to apply to job vacancies. As uploading a resume is required during the application process, a fictitious resume is created with the content remaining constant regardless of the treatment variation. The resume does not include information about part-time working, but does include information about the engagement of the applicant in teaching yoga classes on a voluntary basis. To increase the number
15
of possible applications, the resume is written in both Dutch and English. The resumes are available upon request.
In order to avoid suspicion from recruiters and to gather data that resembles the real world as closely as possible, the resume is made as realistic as possible by examining 10 real-life resumes of master students and recent graduates. From these examples the structure and level of experience are used as input, to ensure that the chosen education and career profile for the fictitious applicant is comparable to those of actual applicants.
The structure of a resume includes a phone number, e-mail address and postal address, creating three different channels for receiving reactions. To deal with reactions via phone, a personal mobile number is acquired with a voice mailbox attached to it, voiced by the thesis writer. Phone calls of recruiters are not answered and send to the voice mailbox in order to keep the applications as identical as possible. Recruiters are likely to ask additional questions on the phone and answering these questions might influence whether the fictitious applicant gets invited for a job interview or not.
To handle reactions via e-mail, a personal Gmail-account is made for the fictitious applicant. Finally, although it is not common to receive an invitation for a job interview via postal mail, an existing postal address is added to the resume to make sure that the number of reactions via postal mail can be measured. The address chosen is close to the center of the Randstad and a station, to reduce the potential travel time to possible employers and to increase the attractiveness of the resume.
The last part of the fictitious applicant’s identity includes a name. To make it less suspicious that the fictitious applicant is untraceable on the internet and does not have personal internet pages as LinkedIn, a very common Dutch name is chosen: Laura Jansen. This name is chosen because it is a combination of one of the most common Dutch female names and surnames in the years 1994-1996 and the Google search results for Laura Jansen are dominated by a famous Dutch-American singer (Brouwer, 2010; Meertens Instituut, 2015).
C. Fictitious cover letter
As the experiment contains three treatments which are expressed in the cover letter, three different cover letters are created in both Dutch and English. The cover letters are available upon request. The general content of the cover letter is based on the skills and experiences included in
16
the resume of the fictitious applicant and is identical for all treatments. In all three cover letters a reference is made to the applicant being a yoga teacher, in order to keep the cover letters identical with respect to the skills of the applicant. Only in the part-time + reason treatment being a yoga teacher is mentioned as the reason for the desire to work part-time. As a result, the cover letters solely differ with respect to the part-time working request and provided reason. In the part-time treatment the following sentence is added to the control cover letter: ‘My ideal contract would involve working 4 days per week.’ and in the part-time + reason treatment: ‘My ideal contract would involve working 4 days per week so I can continue providing yoga lessons on my day off.’ is added. These sentences are placed in a separate paragraph, in order to make the part-time request as visible as possible for recruiters.
D. Responding to vacancies
The field experiment is carried out in the period May-July 2018. Over that period, applications are send via internet to Dutch and English job ads for entry-level professional positions in fields related to economics and business. These includes the fields finance, information technology, consultancy, sales, marketing, commerce, management, research, compliance, human resource management and strategy. To get an overview of available employment ads, ad collecting websites are used. However, applications are send directly to the specific company website to avoid the possibility of general recruiters seeing multiple applications with different cover letters of the fictitious applicant. In total 149 applications are send to real job ads in the Netherlands.
With each application the fictitious resume and one randomly selected cover letter are uploaded. The selection of this cover letter takes place via randomization with categories, where the vacancies are sorted into categories based on the language of the ad and the number of working hours mentioned in the ad. This randomization with categories is conducted since the two variables might influence the probability of receiving a callback for the different treatments. According to Lawler and Bae (1998) the home-country culture of a company can have an influence on the internal culture of a company and can therefore affect the behavior and actions of organizational participants. When a vacancy is recorded in English the adhered organization is most likely not from Dutch origin and operates internationally, implying that the culture of this organization might vary from the culture of organizations that post Dutch vacancies which might lead to different behaviors of employers. The variable number of working hours mentioned in the ad might also
17
have an influence, as the number of working hours mentioned can reflect a solid preference of the employer for a certain availability of the applicant. Using this randomization process with categories allows for creating enough variation for the subcategories and being able to control for the two variables in the analysis, resulting in a reduction of possible sources of variability and a more precise analysis.
Next to the variables used in the randomization process, additional information is logged when a vacancy is selected to be able to correct for possible confounding variables later in the analysis. This additional information regards the assessment of fit with the job requirements and the required experience for the job. Hypotheses with respect to the described variables are included in section F.
E. Measuring responses
The dependent variable in this research entails the responses of recruiters on the applications send and reflects a dummy variable of receiving a callback. An invite for a job interview or shown interest via phone, e-mail and postal mail are measured as a callback. For each ad, it is recorded whether the fictitious applicant receives a response and what kind of response it entails, an invite or interest. When a recruiter contacts the fictitious applicant with additional questions, this is recorded as interest since the recruiter is willing to put in effort in contacting the applicant. The distinction in callbacks is recorded, to be able to check whether the effects of the treatments for the combined callback rates stay constant when the callbacks are split into the two different callback types.
F. Hypotheses
Based on the literature review, hypotheses with respect to the research question are drafted. Firstly, the expectation is that expressing the intention to work part-time will lead to less callbacks compared to not expressing this intention. Part-timers seem to be confronted with negative subjective assessments and cultural conflicts, resulting in the perception of employers that they are a risky group to select from. Employers want to avoid risks as much as possible, therefore they are expected to select applicants who voice the intention to work part-time less often. This leads to the hypotheses:
18
Hypothesis 1a. Expressing the intention to work part-time without providing a reason for part-time
work, leads to a lower probability of receiving a callback compared to not expressing the intention to work part-time.
Hypothesis 1b. Expressing the intention to work part-time providing a reason, i.e. being a yoga
teacher, for part-time work, leads to a lower probability of receiving a callback compared to not expressing the intention to work part-time.
Within this intention to work part-time, a difference in the number of callbacks is expected between the applicant providing extra information about her decision to work part-time compared to not providing this information. When the applicant provides information about the reason for the willingness to work part-time, the employer can update his initial belief as he gets more information about the possible productivity level of the applicant. The additional information can reduce the uncertainty perceived by the employer, which can lead to less perceived risks of selecting the applicant. This reasoning leads to the expectation that recruiters will invite applicants who show the intention to work part-time and provide high opportunity costs of work as the reason for working part-time, more often than applicants who show the intention to work part-time but do not provide a reason for the willingness to work part-time. As a result, the second hypothesis is:
Hypothesis 2. Expressing the intention to work part-time without providing a reason for part-time
work, leads to a lower probability of receiving a callback compared to expressing the intention to work part-time providing a reason, i.e. being a yoga teacher, for part-time work.
Both hypotheses are statistically tested with a proportion test via Stata, which tests whether two variables have the same proportion or whether the proportions are statistically different from each other. With this test, the callback rates for the three treatments are determined reflecting the number of callbacks in proportion to the number of vacancies applied to for a particular treatment. The proportion test tests whether these callback rates adhered to the different treatments are statistically different from each other. If this appears to be the case, the direction of the difference between callback rates is interpreted in order to see which treatment has a higher probability of receiving a callback.
To check whether differences in the probability of receiving a callback can be ascribed to the different treatments and are not influenced by confounding factors, a probit regression with additional control variables is conducted. When the subcategories of the additional variables are balanced over the three treatments, the assignment of the treatments can be determined as random
19
which means that the results of the analysis can be ascribed to the different treatments. However, when the different subcategories are not balanced over the three treatments, they might influence the results found and there needs to be corrected for these variables via a probit regression. A probit regression is chosen in order to avoid possible problems with the standard errors that might arise when using a linear probability model.
The first additional control variable considered, is the assessment of fit between the qualifications of the fictitious applicant and the requirements stated in the job ad. Four subgroups are distinguished, namely perfect fit, good fit, okay fit, and poor fit. The assessment of fit is judged by the thesis writer and is being scaled into a lower subgroup for every requirement that the fictitious applicant does not meet. The expectation is that applications to vacancies that are not a perfect fit will lead to less callbacks compared to applications to vacancies with a perfect fit, as the applicant meets less criteria and might become less attractive in the eyes of the employer. This leads to the following hypothesis:
Hypothesis 3. Applications to job ads with a perfect fit lead to a higher probability of receiving a
callback compared to applications to job ads without a perfect fit.
The required experience for the job applied to is considered as the second and last additional control variable. A distinction is made between vacancies that do not state anything about experience, vacancies that explicitly require experience and vacancies that explicitly require no experience. As the fictitious applicant is a starter without experience, the expectation is that applications to job ads that explicitly require experience will lead to less callbacks compared to applications to job ads that no dot explicitly require experience since the applicant meets less requirements within these applications. The following hypothesis is based on this reasoning:
Hypothesis 4. Applications to job ads that explicitly require experience lead to a lower probability
of receiving a callback compared to applications that do not require experience explicitly.
Finally, specific expectations regarding the variable working hours mentioned are tested via a separate probit regression as this variable is expected to influence the differences in the probability of receiving a callback between the control and two part-time treatments. Within this variable the following subgroups are distinguished, full-time, flex, part-time, and not applicable meaning that the number of working hours is not explicitly mentioned. Job vacancies mentioning that the function regards a full-time job show an explicit preference for a worker who is available for a certain number of hours. The employer adhered to this job vacancy might not be willing to
20
invite and hire an applicant who deviates from this criterion, as they explicitly chose to mention the desired full-time working hours. As a result, the expectation is that the difference in callback rates of applicants who do not show the intention to work part-time and applicants who show the intention to work part-time is bigger for job ads that explicitly mention the working hours being full-time compared to vacancies that do not mention this. This leads to the following hypotheses:
Hypothesis 5a. The difference in probability of receiving a callback between applicants who do
not express the willingness to work part-time and applicants who express the willingness to work part-time without providing a reason for part-time work, is bigger for vacancies that mention the job being full-time compared to vacancies that do not mention the job being full-time.
Hypothesis 5b. The difference in probability of receiving a callback between applicants who do
not express the willingness to work part-time and applicants who express the willingness to work part-time providing a reason, i.e. being a yoga teacher, for part-time work, is bigger for vacancies that mention the job being time compared to vacancies that do not mention the job being full-time.
4. Results
In this section the empirical results are presented. Table 2 shows the number of vacancies applied to and the different responses per treatment. The different treatments are equally distributed across the applications. The overall average callback rate is 53.02%, meaning that the fictitious applicant has a chance of 53.02% to receive a callback after applying for a job.
Table 2 – Treatment specifications
Treatment Number of
applications
% proportion Number of callbacks Total Interview Questions
Control 48 32.21 27 9 18
Part-time 50 33.56 25 14 11
Part-time + reason 51 34.23 27 13 14
Total 149 100 79 36 43
Note: The table shows the number of observations per treatment (column 2). Column 3 reports the proportion of vacancies adhered to the specific treatment. Column 4, 5 and 6 show the number of callbacks per treatment and callback type.
In the remaining section attention will be paid to the effects of the different treatments in more detail, by first looking at the callback rate for the combined callback types and then for the
21
different callback types separately. Finally, the possible control variables and limitations are highlighted.
A. Callback rates
Table 3 reports the callback rates per treatment, with the accompanied number of applications in brackets underneath. As expected, the results show that the control treatment has the highest callback rate with 56.25%. This callback rate is higher than the callback rates for applications where the intention to work part-time is expressed, regardless of whether a reason is provided. Within expressing the willingness to work part-time, the part-time + reason treatment shows a higher callback rate than the part-time treatment, as predicted.
Table 3 – General callback rates per treatment
Treatment % callback Ratio % point
difference (p-value) Control 56.25 [48] Part-time 50.00 [50] Part-time + reason 52.94 [51] Control vs part-time 1.13 6.25 (0.5354)
Control vs part-time + reason 1.06 3.31
(0.7411)
Part-time + reason vs part-time 1.06 2.94
(0.7675)
Notes: The table reports, for the entire sample, the callback rates in percentage (column 1) for applicants who do not show the intention to work part-time, who do express the intention to work part-time without providing a reason and who show the intention to work part-time with providing a reason. In each cell the number of apllications sent in that cell is reported in brackets. Column 2 reports the ratio and column 3 the differences between the callback rates with the p-value for a test of proportion, testing the null hypothesis that the callback rates are equal across the treatment groups.
The difference between receiving a callback when the applicant does not voice the intention to work part-time and when the applicant does show this intention but does not provide a reason, is 6.25 percentage points. This means that an applicant who does not show the intention to work part-time has a 13% higher chance of receiving a callback compared to an applicant who shows
22
the intention to work part-time without providing a reason. The difference in callback rates between the control treatment and part-time + reason treatment is 3.31 percentage points or 6.25%. For the part-time + reason treatment and part-time treatment the difference in callback rate is 2.94 percentage points, meaning that an applicant who expresses the willingness to work part-time while providing a reason, i.e. being a yoga teacher, has a 5.88% higher chance of receiving a callback compared to an applicant who does not provide a reason for part-time work. These described differences, however, are not statistically significant. As a result, the null hypotheses cannot be rejected meaning that the callback rates are equal across the treatment groups compared. No support is found for hypothesis 1, 2a and 2b.
As described in the experimental design, the callbacks recorded can be divided into two different categories. In order to see whether the results found for the combined callback rates stay constant across both callback types, the callback rates per callback type are examined separately.
A.1 Interview callback rates
Table A.1 in the appendix presents the interview callback rates per treatment and the differences between the callback rates. Surprisingly, the results show a contradicting tendency compared to the described results for the combined callback rates. Applications that do mention the willingness to work part-time have a lower callback rate than applications where this willingness is expressed, as both part-time treatments show a higher callback rate than the control treatment. Within voicing the intention to work part-time, the part-time treatment has a higher callback rate than the part-time + reason treatment. These differences between the treatments, however, remain insignificant. The null hypotheses cannot be rejected, meaning that the callback rates are equal across the treatment groups compared for the interview callback rates.
A.2 Interest callback rates
Table A.2 in the appendix shows the interest callback rates per treatment, reflecting the callbacks where recruiters do not explicitly invite the applicant for a job interview but do show interest in the candidate. For this callback type the same trend can be found in the results as for the combined callback rates, since the control treatment has a higher callback rate than both part-time treatments and the part-time + reason treatment has a higher callback rate than the part-time treatment. The differences between the treatments, however, remain statistically insignificant. The null hypotheses cannot be rejected, meaning that the callback rates are also equal across the treatment groups compared for the interest callback rates.
23
In summary, the differences found between callback rates for the three treatments appear to be insignificant for all types of callback. As a result, no support is found for hypothesis 1, 2a and 2b. Expressing the willingness to work part-time does not lead to a lower probability of receiving a callback compared to not expressing this willingness and providing a reason, i.e. being a yoga teacher, for part-time work does not lead to a higher probability of receiving a callback compared to not providing a reason for part-time work.
Result 1. The callback rates for an applicant who expresses the intention to work part-time and an
applicant who does not express this intention, do not statistically differ from each other.
Result 2. The callback rates for an applicant who expresses the intention to work part-time
providing a reason, i.e. being a yoga teacher, for this and an applicant who expresses the intention to work part-time but does not provide a reason, do not statistically differ from each other.
B. Confounding factors
In order to check whether differences in the probability of receiving a callback can be ascribed to the different treatments and are not influenced by possible other factors, a probit regression is conducted. Table A.3 and A.4 in the appendix show the proportions of the subcategories of the variable assessment of fit and required experience per treatment. The subcategories of both variables do not appear to be balanced across the treatments. These variables are therefore taken into account as control variables.
Table A.5 and A.6 in the appendix show the number of observations and proportions per subcategory of the two control variables. With respect to the assessment of fit, 33.56% of the vacancies are assessed as a perfect fit. Most vacancies, 43.62%, are however categorized as a good fit and 21.48% reflect an okay fit. Vacancies with a poor fit are not considered during the control analysis, as these only reflect 1.34% of the job ads applied to. Of all vacancies applied to, 25.50% explicitly required experience for the job in comparison to 26.17% explicitly requiring no experience. Finally, 48.32% of the job ads applied to did not mention anything about the level of experience and are categorized as not applicable.
To investigate the effect of the different treatments on the independent variable callback (y) and control for possible confounding factors, two separate probit regressions are set up:
24
𝑦 = 𝛽0+ 𝛽1𝑡𝑟𝑒𝑎𝑡𝑚𝑒𝑛𝑡𝑃𝑇+ 𝛽2𝑡𝑟𝑒𝑎𝑡𝑚𝑒𝑛𝑡𝑃𝑇𝑟𝑒𝑎𝑠𝑜𝑛+ 𝛽3𝑓𝑖𝑡𝑔𝑜𝑜𝑑+ 𝛽4𝑓𝑖𝑡𝑜𝑘𝑎𝑦+
𝛽5𝑒𝑥𝑝𝑒𝑟𝑖𝑒𝑛𝑐𝑒𝑛𝑜+ 𝛽6𝑒𝑥𝑝𝑒𝑟𝑖𝑒𝑛𝑐𝑒𝑛𝑎 (2) The regression output is shown in Table 4 and investigates the estimated discrete changes in the probability of receiving a callback from the base level for the specific dummy variables considered. Column 1 reports the results of a callback dummy on two treatment dummies (Equation 1). The dummy variable callback equals 1 when a callback is received and equals 0 when no callback is received. For the variables treatmentPT and treatmentPTreason the base level is reflected by the control treatment. Column 2 of Table 4 reports the results of a callback dummy on two treatment dummies and two control variables, represented by two assessment of fit dummies and two required experience dummies (Equation 2). For the dummy variables fitgood and fitokay the base level is represented by vacancies with a perfect fit and for the variables experienceno and experiencena vacancies that explicitly require experience are considered as the baseline.
Table 4 – Probit regressions of callback on treatments and control variables
(1) (2)
(1) Treatment – part-time -0.063
(0.101)
-0.070 (0.101) Treatment – part-time plus reason -0.033
(0.100) -0.046 (0.102) (4) Fit – good -0.021 (0.097) Fit – okay 0.035 (0.122) (5) Experience – no 0.175 (0.122)
Experience – not applicable 0.011
(0.108)
Note: Each column presents the results of a probit regression where the dependent variable is the callback dummy. Reported in the table are estimated discrete changes for the variables listed in the rows. Column 1 includes two treatment dummies, with treatment=0 reflecting the control treatment and treatment=1 reflecting the described treatment in the table. Column 2 includes a treatment dummy, with treatment=0 reflecting the control treatment and treatment=1 reflecting the described treatment in the table, and two control variables. The control variables include two assessment of fit dummies with fit=0 reflecting vacancies with a perfect fit and fit=1 reflecting the described fit in the table and two required experience dummies with experience=0 reflecting vacancies that explicitly require experience and experience=1 reflecting the described experience in the table. Standard errors are reported in parentheses; * p < .05.
25
Table 4 shows that the changes in probability of receiving a callback caused by moving between the subcategories of the four control dummies are not significant, meaning that no support is found for hypotheses 3 and 4. Applications to job ads with a perfect fit do not lead to a higher probability of receiving a callback compared to applications to job ads without a perfect fit and applications to job ads that explicitly require experience do not lead to a lower probability of receiving a callback compared to applications that do not require experience explicitly.
The results presented in Table 4 confirm the earlier found pattern that both part-time treatments have a lower probability of receiving a callback compared to the control treatment and that these effects are not statistically significant. After controlling for the assessment of fit and required experience adhered to the job ads, the differences in the probability of receiving a callback between the treatments remain in the same direction and insignificant. The control variables do not influence the results found.
Result 3. The change in probability of receiving a callback caused by moving from an application
to a job ad with a perfect fit to a job ad without a perfect fit is not significant.
Result 4. The change in probability of receiving a callback caused by moving from an application
to a job ad that explicitly requires experience to a job ad that does not explicitly requires experience is not significant.
To investigate whether the working hours mentioned in the ad influences the differences in the probability of receiving a callback between the control and two part-time treatments, an additional probit regression is conducted. Table A.7 in the appendix shows the number of observations and proportions per subcategory of variable working hours mentioned. Most of the vacancies applied to, 62.42%, did not mention information about the number of working hours applicable to the job, followed by 34.90% of the vacancies mentioning that the job regards a full-time job. The two other subcategories do not occur often in this experiment, 2.01% of the vacancies mentioned the working hours being flexible and 0.67% mentioned the job being a part-time job. As a result, these two subcategories are not considered during the analysis and the following probit regression is set up:
𝑦 = 𝛽0+ 𝛽1𝑡𝑟𝑒𝑎𝑡𝑚𝑒𝑛𝑡𝑃𝑇+ 𝛽2𝑡𝑟𝑒𝑎𝑡𝑚𝑒𝑛𝑡𝑃𝑇𝑟𝑒𝑎𝑠𝑜𝑛 + 𝛽3ℎ𝑜𝑢𝑟𝑠𝑛𝑎+
26
Table 5 reports the results of a callback dummy on two treatment dummies, the hours mentioned dummy and two separate interactions of the hours mentioned dummy with the two treatment dummies (Equation 3). The reported coefficient reflects the estimated change in probability for the interaction term. Vacancies that mention the job being full-time are considered as the baseline for the variable hoursna.
Table 5 – Probit regression of callback on treatments and hours mentioned variable Treatment – part-time Treatment – part-time
plus reason (1) Hours – not applicable -0.740
(0.545)
0.399 (0.556)
Note: The table presents the results of a probit regression where the dependent variable is the callback dummy. Column 1 includes the results of the regression for treatment=1 reflecting the part-time treatment and treatment=0 reflecting the control treatment. Column 2 includes the results of the same regression for treatment=1 reflecting the part-time + reason treatment and treatment=0 reflecting the control treatment. Reported in each cell is the estimated change in probability for the interaction term. Standard errors are reported in parentheses; * p < .05.
The results show that both interactions are not significant, meaning that the number of working hours mentioned in the ad does not influence the gap in the probability of receiving a callback between the control treatment and the two part-time treatments. No support is found for hypotheses 5a and 5b.
Result 5. The difference in probability of receiving a callback between applicants who do not
express the willingness to work time and applicants who express the willingness to work part-time is not influenced by the number of working hours mentioned in the job ad.
D. Limitations
The results found have to be interpreted with some caution with respect to the external validity. First of all, the callback rates established in this paper appear to be substantially higher than the callback rates of the study of Bertrand and Mullainathan (2004), where differential treatment by race during the application process is studied. The callback rates for applications in their study do not exceed 13.60%. However, the auteurs only record a callback when an applicant receives an invite for a job interview and do not include callbacks where the recruiter solely expresses interest in the applicant. When looking at the interview callback rates of this paper (Table A.1), the callback rates come closer to the callback rates of Bertrand and Mullainathan (2004) study
27
but still appear higher with the highest callback rate being 28.00%. This high callback rate can imply that the profile of the fictitious applicant is appealing in general in the eyes of the employers. The high callback rate might also be explained by the Dutch labor market conditions prevalent during the period of applying to vacancies. According to Centraal Bureau voor de Statistiek (2018) the demand for labor is increasing compared to the past years and above average, while the supply of labor is relatively low. This creates a favorable position for job applicants and might lead to a higher callback rate. When considering the callback rates, this labor market condition has to be taken into account.
Besides considering the labor market conditions, the results might not be generalizable to male applicants as the fictitious applicant in this field experiment is a female. In the Netherlands 75.80% of the working women engage in part-time working compared to 27.00% of working men (Eurostat, 2018). This uneven distribution of part-timers might result in a biased view towards experiences with females (Cohen & Single, 2011). Certain social norms arise because it is more common for women to work part-time, which may lead to a different interpretation of the signal send by males who express the willingness to work part-time compared to the signal of females (Picchio & Van Ours, 2016). Multiple studies have indeed shown that male part-timers are treated differently than female timers. Wood et al. (1993) for example show that the penalty for part-time work is much larger for men than for women. Picchio and Van Ours (2016) find a gender-specific effect of working hours on training, where men are less likely to receive training when they work part-time compared to full-time while women have the same likelihood of receiving training in both cases. These found differences between male and female part-timers imply that employers might also react differently on part-time requests of men compared to women during the application process. Consequently, the results of this thesis might not be representative for men but they are representative for women, the majority of part-timers.
5. Discussion and conclusion
Discussion
Surprisingly, the empirical results do not support the idea that expressing the willingness to work part-time influences the employability of an applicant as all results appear insignificant. Although insignificant, the found differences in callback rates between treatments do appear to be in the expected direction. Showing the intention to work part-time reduces the callback rate compared to not showing this intention and showing the willingness to work part-time without
28
providing a reason for this leads to less callbacks compared to expressing the intention to work part-time with providing a reason related to high opportunity costs of work. If these results would have been significant, they would have implied that part-timers are faced with a disadvantage during the first step of the selection process, possibly caused by statistical discrimination. It would have implied that employers use the request to work part-time as a criterion during the application process, suggesting that the request is used as a proxy for the applicant’s productivity and applicants are subjected to negative associations with respect to this productivity.
An alternative explanation for the, although insignificant, found pattern in callback rates between an applicant who does not show the intention to work part-time and an applicant who does show this intention, could have been employers basing their decision on practicality with respect to hiring part-time applicants. Hiring a part-time worker can be less practical as the recruiter might have to hire multiple workers to fill the desired number of working hours instead of just one worker. This could lead to extra costs since the employer has to provide certain facilities and premiums twice. These extra costs might serve as a reason for employers to not invite applicants who show the intention to work part-time (Nobis, 2000), instead of the expected productivity level being the reason.
When splitting the callbacks into two types, the results remain insignificant. The results adhered to the callback type where recruiters solely express interest also appear to be in the expected direction, but the results of the callback type where the recruiter invites the applicant for a job interview contradict the expectations. With this interview callback type, expressing the intention to work part-time increases the callback rate compared to not expressing this intention and voicing the willingness to work part-time without a reason leads to more callbacks compared to voicing the willingness to work part-time with providing a reason related to high opportunity costs of work. Had the difference in results of the two callback types been significant, it would have been difficult to provide a logical explanation for this difference.
As the callback rates of two identical individuals engaging in an identical job search with the only difference being a request to work part-time are not statistically different, it appears that expressing the willingness to work part-time does not influence the employability of an applicant. Employers do not seem to use the request to work part-time as a proxy for the applicant’s productivity and assume that part-timers have a lower productivity level on average. As a result, it
29
is suggested that statistical discrimination with respect to part-timers does not take place during job applications.
Employers also do not seem to react differently on part-time applicants who provide a reason related to high opportunity costs of work for the willingness to work part-time compared to part-time applicants who do not provide a reason, as the callback rates for these two treatments do not differ statistically. This implies that recruiters are not influenced by the extra information provided during the reviews of applications. This could mean that recruiters do not use the extra information about the part-time request for updating their initial belief or that recruiters do not have a negative association with part-timers. Given the finding that part-time applicants are not treated differently compared to applicants that do not show the intention to work part-time, the suggestion is that recruiters do not have an initial negative belief about part-timers throughout the first step of the application process and therefore do not respond to the additional information about wanting to work part-time.
One possible alternative explanation for not finding a significant difference between the treatments, could be lexicographic search by employers. Recruiters receive a lot of applications, which might make them fall back on using quick scanning techniques during the reviews of resumes and cover letters. The study of TheLadders (2012) shows that recruiters spent almost 80% of their resume review time on 6 data points, namely the name of the applicant, the acquired education, the current job title plus starting date and previous job titles with the starting/ending dates. Next to these data points, recruiters mostly scan for keywords that match the open vacancy. This scanning of keywords might also take place during the reviewing of the cover letter, possibly resulting in recruiters not having read the different treatments and making their decision without knowing that the applicant desires to work part-time and for what reason.
Besides this quick scanning, it is also possible that recruiters do not actually read the cover letter and base their selection decision solely on the resume. The real-life recruiters that were consulted before the start of this field experiment, stated that they do read cover letters and use them when making their selection decision. Other sources imply that cover letters indeed remain important but that they are not read thoroughly and resumes receive more weight during the selection decision (Van der Leij, 2015). This possibility of recruiters basing their decision on the applicant’s resume, might be an alternative explanation for part-time applicants not having a different callback rate compared to applicants who do not express the intention to work part-time.
