Understanding flexibility – dimensions of employee behavior flexibility

Recently, amongst theoreticians and practitioners of management, one can notice a growing interest in the subject of the flexibility of the work environment (among others: Galinsky et al., 2008; Kossek et al., 2015). In a changing, turbulent, unpredictable environment, an organisation must be able to recognise and adapt to new situations. The necessity of a quick response to change, forces the use of an array of solutions, including those aimed at making human capital more flexible. Some of these responses are direct and refer to personnel policy, like flexible working time solutions (Campbell et al., 2012; Koekemoer & Downes, 2011) or changes in the model of career management (Blustein, 2013; Baruch, 2006). Other solutions, while aimed primarily at making the organisation more flexible (eg changes in structure, technology), only affect employees indirectly, though without a doubt change the nature of the employee-organisation relationship (Januszkiewicz, 2018). Solutions that make the work environment more flexible by creating space for flexible organisational behaviour of employees (FOBE) are being increasingly implemented by employers both in relation to the expected productivity growth and the positive impact on work-life relationships (among organisational behaviour, flexibility, and others: Kossek et al., 2015; Pitt-Catsouphes & Matz-Costa, 2008; White et al., 2003; Hill et al., 2008).

Despite the wealth of literature, it seems quite symptomatic that there is no homogeneous terminology in the area discussed. To describe the flexibility of the work environment, authors most often use such categories as workplace flexibility (Hill et al., 2008), flexibility work arrangements (de Sivatte & Gaudamillas, 2013; Kossek et al., 2016; Rau & Hyland, 2002), flexibility policies (Eaton, 2003; Noonan et al., 2007) and flexibility practices (van der Meer & Ringdal, 2009; Lambert, 2008). In addition, the use of a certain type of label has been noted, such as a new work environment, agile work, flexible, wise flexibility (Lake, 2015), and smart work (Lee & Lee, 2012), suggesting a departure from the old, traditional, non-agile, inflexible ways of working characteristic of industrial production towards other, more flexible, solutions.

Analysing the differences in the presented approaches of flexibility, it should be pointed out that this category in the subject literature is treated on the one hand as an organisational feature (see: also: Kalleberg, 2003; Michie & Sheehan-Quinn, 2001), and on the other hand, as a fragmentarily feature of certain organisational solutions. In both perspectives, however, it should be noticed that the lack of attention to relations or connections between particular dimensions of flexibility makes it difficult to conduct deliberations within the discipline. Therefore, on the assumption that flexibility is manifested in various dimensions, in varying degrees, and with different dynamics, it seems justified to build a model that, on the one hand, takes into account these differences and on the other hand, allows for a complete and consistent description of flexibility.

In addition, as Lonti and Verna (2003) note, flexible solutions do not occur by accident or singly, but in clusters that allow for some flexibility in their application, which seems particularly important from the point of view of the theoretical analysis of the phenomenon. The study of organisational solutions in clusters, in contrast to the study of isolated incidents, allows for the emergence of certain patterns and the possibility to capture regularities (Cutcher-Gershenfeld, 1991). According to McDuffie (1995) it also enables a more complete picture of the relationship between organisational solutions and the behaviour of employees. Therefore, it should be assumed that the flexibility of organisational behaviour of employees is related to the flexibility of organisational solutions that occur in qualitatively separated clusters.

Under the above assumptions, the purpose of this article is to present a model of flexible organisational behaviour of empolyees [FOBE], which is a proposal for a consistent description of employee flexibility and its empirical verification.

The flexibility of the organisational behaviour of employees should be understood in this approach as a manifested change in behaviour that fosters either a development, or maintains the position of, an employee with regard to his/her potential and the potential of the organisation in which he/she is employed. The frequency of change determines the level of flexibility, while its scope determines the type of flexibility of organisational behaviour of employees (Januszkiewicz, 2018).

Such a conceptual framework for flexibility was the basis for identifying the indicators of flexible behaviour described in the literature and assigning them to individual dimensions (Table 1).

The qualitative diversity of indicators assigned to particular dimensions of behaviour flexibility raises three basic questions:

are the indicators of flexible behaviour related to each other?

The answer to such questions requires an empirical study that will allow verification of the FOBE theoretical model.

The validation study was conducted among 322 randomly selected employees, as a part of a research project carried out in 2018–2020. The sample was diversified in terms of basic social and demographic features: gender (61.2% women, 38.2% men, 0.6% no data) and age (at least 19 years, max = 67 years, M = 31. 22 years, Me = 27 years, Q1 = 25 years, Q3 = 38 years, SD = 8.58 years, skew rate = 1.070, Kr = 0.796) Also taken into account was the situation at the workplace: the period of employment in the current workplace (less than 1 year: 24.8%,; 1–3 years: 30.1%; 3–5 years: 11.5%; 5 10 years: 15.8%; over 10 years: 17.7%); position held (managerial: 23.9%; non-managerial: 76.1%); employment sector (services: 43.8%; sales: 20.5%; production / industry: 9.6%; other: 26.1 %). The respondents considered each of the 32 indicators of behavioural flexibility (Table 1), on a four-level scale, indicating how often the given situation occurred in their case. The variables analysed were measured on an ordinal scale. Each of their variants has been coded numerically: 1 corresponds to the variant less than once; 2: from two to three times; 3: from four to five; and 4: more than five.

Assessment of the homogeneity of positions – in total and across four separate dimensions – was made using Cronbach’s alpha. Cronbach’s alpha for the general scale of elasticity reaches 0.891, and each of the items is valid in it (none of the variables interferes with the assessment of flexibility, that is, removal does not improve the value of the reliability coefficient). With a full, prior proposed set of variables, the Cronbach’s alpha coefficient reaches 0.819 for task flexibility, 0.831 for functional flexibility, 0.734 for time flexibility, and 0.757 for spatial flexibility.

The results of the research also showed that the individual indicators are correlated, and it is mostly a statistically significant and positive correlation (Table 2). It should be noted that within theoretically isolated blocks, correlations are important for each pair of variables (the only exception is block 3, in which the P3.7 variable is not significantly related to P3.1 and P3.2, while it is significantly correlated with all items in blocks 1 and 2). The strongest relations are observed within block 2 (functional flexibility). The individual functional flexibility indicators are also quite strongly related to the task flexibility indicators; similar relations are observed between the time and spatial flexibility indicators. Significantly weaker relations are observed between the time and spatial flexibility indicators and task flexibility. For a few pairs of variables (apart from the blocks indicated in the theory) correlation coefficients are even negative, although they were not statistically significant.

With respect to detailed analyses, it should be pointed out that in the dimension of task flexibility, P1.1 and P1.2 are most closely related to each other; the greater the flexibility in changing the type of tasks performed, the more flexibility regarding their scope (and vice versa): rho = 0.575. In addition, the greater the flexibility in parallel multitasking, the greater, on average, the openness to the reactive change of tasks (and vice versa) – rho = 0.569. In turn, flexibility in the scope of changing technical and technological solutions related to the performance of tasks goes, to a small extent, hand in hand with flexibility regarding the type of tasks performed (rho = 0.161).

In the case of the functional flexibility dimension, particularly striking is the relationship between work flexibility in different types of teams as well as in teams performing various functions (rho = 0.612) and between the flexibility of performing different team roles in the same team and performing different team roles in parallel (formally unspecified functions in the team, eg Initiator, Team Soul) (rho = 0.606) and performing different team roles in different teams (rho = 0.531).

In the time flexibility dimension, the strongest relationship is observed between the separation of individual working time from the overall working time of the organisation versus asynchronous presence at work due to different working hours (rho = 0.540) and self-organisation of working time (rho = 0.502) and also between the change in working time and its distribution (rho = 0.501).

However, in the dimension of spatial flexibility, the relationship between the frequency of short- and medium-term trips is clearly distinguishable (rho = 0.647). The relationship between flexibility in working in another town/country and short- and medium-term trips is quite strong (rho 0.455 and 0.479, respectively).

It is also worth noting that the performance of tasks outside the workplace (P4.9) is strongly related also to the items from block 3, especially with self-organisation of working time (P3.7), asynchronous presence at work (P3.5), separation of individual working time from the general work time of the organisation (P3.6), and change of the working time distribution (P3.4). Self-organisation is quite strongly related to parallel multitasking (P1.6). The performance of different organisational roles in parallel in several organisations (P2.5) is in turn quite strongly related to the work in two or more organisations at the same time (P3.2). In the course of further work, the compliance of the obtained data with the value of the latent variable to be described was subjected to empirical verification. The assessment of theoretical accuracy by means of factor analysis requires that the sample size should not be less than 50 elements (Sapnas & Zeller, 2002). Tabachnick (2007) suggests that the sample should have at least 300 elements. Another determinant in this respect is STV, ie the relationship between the number of variables and the sample size was at a level not lower than 5:1 (Costello, Osborne, 2005). In this case both conditions are met: n = 322 (thus even the strictest Tabachnick criterion is met), while STV = 32/322 = 9.9. Factor analysis also requires that the variables analyzed, measuring the latent variable in a synthetic way (here, employee flexibility), are strongly correlated with each other, which is assessed on the basis of the correlation coefficients between variables and in particular on the basis of the KMO measure (Kaiser-Meyer-Olkin) and Bartlett’s sphericity test. It is recommended that the KMO should have a value not lower than 0.5, while the H₀ hypotheses are tested in the sphericity test: the correlation matrix is the unit matrix, for H₁ : H₀ (and therefore it is expected that p < α) (Field, 2000). In this study, KMO is high (0.856), and the sphericity test is significant (χ² = 3893.3, df = 496, p < 0.001 *). In addition, for each variable the coefficient on the diagonal of the matrix of inverse image correlation significantly exceeds the threshold value of 0.5, and the correlation matrix determinant is close to 0. As mentioned earlier, in the adopted set of 32 variables there is not one that would be irrelevant to most other items (Table 2). This confirms the legitimacy of conducting the factor analysis based on the extracted list of flexibility indicators.

The resources of shared variability can be estimated using various methods, the most popular being the maximum likelihood factor analysis (MLFA) and the principal axis factoring (PAF) method, as well as – most often used – the adaptation of Hotelling’s classic method of main components for factor analysis (Walesiak & Gatnar, 2009; de Winter & Dodou, 2012). The main component method was used in this study. The communalities (ZZW) – and therefore a part of the common variance presented by a given index – are quite high for most variables, the exceptions being P1.8, P3.1, P3.3, P4.1, P4.4, and P4. 7 (ZZW does not even exceed 0.3; see Table 3). This may mean the need to eliminate the abovementioned variables from the final catalogue of flexibility indicators. In the case of other variables, the results confirm their significant discriminant power.

The next step in verifying the validity of the model is to determine whether within the 32 theorems examined the four areas, in accordance with the assumptions, should be distinguished. For this purpose, mainly, the Kaiser criterion (factors for which the eigenvalue is greater than 1 were selected) and the Cattell criterion (in the scree plot, the number of factors is determined on the basis of the “scree” slope) were used (Wiktorowicz, 2016). The Kaiser Criterion points to as many as eight factors (components) that explain a total of 61.9% of the variance of the latent variable, while the last four factors explain it in only for 3%–5%, and additionally, generate factors consisting of single variables. The scree plot confirms the legitimacy of distinguishing four dimensions of the flexibility of organisational behaviour of employees. These factors explain in total 46.2% of the variance of the latent variable, of which the first factor, corresponding to the task flexibility (EZ) explains 23.7%; the second, spatial flexibility (EP), 9.9%; the third, functional flexibility (EF), 6.5%; and the last one, temporal flexibility (EC), in 6.1% (Table 4).

The determined values of factor loadings (using Promax’s slanted rotation) indicate the degree of connection of a given item with a specific factor. Rotation is used to simplify and refine the data structure (Malarska, 2005). There are two groups of rotation methods: orthogonal (assuming lack of correlation of factors: Varimax, Quartimax, Equamax) and diagonal, recommended when factors are correlated (Promax, Oblimin, Geomin). In the absence of a correlation of factors, both methods lead to analogous results. This was also the case with this study. As emphasised earlier, items within particular blocks were poorly correlated with items in other areas. This could indicate the orthogonal rotation method (Varimax was used). Simultaneously, the diagonal method Promax (with kappa = 4) was used, which showed a relatively low degree of linkage of the isolated factors (the highest correlation coefficient applied to blocks 1 and 4, task and spatial flexibility, and the correlation coefficient of these factors was the highest at 0.455). However, the factor structure, determined using both methods, was analogous; hence, the results were finally presented using Varimax rotation (Tab. 4).

The results presented confirm the importance of task-based flexibility, decisive for the overall flexibility of employees, as this factor is the main component, explaining 23% of its variance. This group also includes those theorems treated in the theoretical model as elements of functional flexibility (P2.1, P2.2 and P2.3; their factor loadings for the EF component are also quite high). Nevertheless, three of the items included here (P2.2 and P2.3, as well as P1.8) obtained factor loadings lower than 0.5. The most important factor for the flexibility of employees is the change in the scope of competencies used during work, the type and scope of tasks performed, and the manner and methods of performing work. These are followed by work in various types of teams, parallel multitasking, and reactive change of tasks (Table 4).

The second factor, explaining about 10% of the total variance of flexibility, includes a total of nine variables from block 4 (spatial flexibility). This area is, therefore, the second most important for the overall flexibility of employees (with short- and medium-term trips, diagonal displacement, work in another organisation in another town or country being at the top of this particular factor). This group also included work in two or more organisations at the same time and work in the same organisation but in another town. Amongst the variables treated in the model as indicators of spatial flexibility, the low values of factor loadings should be noted for P4.1 and P4.7, as well as the ambiguous identification (associated with several factors) of P.3.1, work based on non-standard forms of employment (Table 4).

The next most important area of behaviour flexibility is functional flexibility. Amongst the eight indicators of flexibility taken into account in this dimension (in the light of factor analysis), it seems appropriate to include five: performing in the same team or parallel different team roles (these two elements are the most important), as well as performing in parallel different team/organisational roles in different teams or organisations (P2.4, P2.5, P2.6, P2.7 and P2.8).

In the last group, including theorems regarding temporal flexibility, in the light of factor analysis, five indicators seem to be justified: asynchronous presence at work due to differentiated working time, separation of individual working time from the general working time of the organisation, self-organisation of working time, and change of working time distribution. P4.9 was also included in this group, performing tasks outside the workplace, which, in accordance with assumptions, is an element of spatial flexibility. Factor loadings are also quite high for other dimensions of flexibility, although on a statistical basis this variable is most strongly associated with time flexibility. Guided by theoretical premises, however, we decided to include this item in the dimension of spatial flexibility.

At the next stage of research, confirmatory factor analysis (CFA), belonging to the class of structural equation models that allows for verification of a specific theoretical model, was used. The assumed measurement model was characterised by satisfying adjustment measures in the light of the most frequently accepted criteria: χ² / df = 2.73; RMSEA = 0.073; p (RMSEA < 0.05); CFI = 0.837. The obtained standardised regression coefficients were interpreted as factor loadings of the indicator for each of the separate dimensions. The test results confirmed that all indicators reach values above 0.4. Thus, it can be concluded that the four separate dimensions are well reflected by their indicators; moreover, the results confirm the factorial accuracy and internal coherence of the distinguished dimensions of behaviour flexibility.

In accordance with the assumptions adopted, the empirical analysis in this work covered the dimensions of flexibility described with the help of clusters of indicators, not separate incidents of behaviour. With respect to the first research question:

the findings confirmed the existence of links between the behavioural indicators assigned to individual dimensions of flexibility. In most cases, this is a statistically significant and positive correlation (virtually for each pair of variables).

However, the strength of relationships between indicators in individual dimensions is diverse, and in addition, trans-dimensional statistical dependencies have also been noted. Such a state may result, among others, from the fact that diverse behaviours may have a common source in a specific organisational solution. For example, employee mobility programs change not only the scope of tasks performed (task flexibility), but also the roles performed (functional flexibility) or the place of work (flexibility of space). Sometimes, however, different solutions may lead to the same behaviours; multitasking may result from the implementation of agile management methodology, but it may also be the result of employment reduction or implementation of work humanisation methods. For this reason, it should be assumed that for the description of behaviour it seems justified to refer not so much to the organisational solutions themselves as to the indicators of flexible behaviour.

With respect to the second question, it should be noted that

the results of the statistical analysis also confirmed the heterogeneity of behavioural flexibility and the legitimacy of separating it into four dimensions connected with the general category. (These dimensions explain in total 46.2% of the variance of the latent variable.)

However, in the theoretical model, it was assumed that each dimension of flexibility distinguished is equivalent, while empirical studies have shown that for flexibility characteristics of employee organisational behaviour, task flexibility is most important (it explains it in 23.7%), followed by spatial flexibility (9.9%), functional flexibility (6.5%), and finally – time flexibility (6.1%). The discrepancies revealed indicate (in the course of further work) the need to give individual dimensions varied variable weights, the size of which will depend on the assessment of information affluence (discriminant ability and information capacity) of diagnostic variables.

In answer to the third question confirmatory factor analysis (CFA) also confirmed the internal consistency of the categories highlighted in the FOBE model:

the four dimensions of behavioural flexibility are well reflected by the indicators describing them, meaning they should be considered qualitatively accurate and coherent.

Therefore, it should be recognised that the FOBE model presented in the article is a proposal for a consistent and terminological description of the flexibility of the organisational behaviour of employees. In the theoretical model, flexibility is treated as a heterogeneous category, thanks to which, in the empirical dimension based on identified indicators, it becomes possible not only to determine the general level of flexibility of behaviour but also the flexibility in these dimensions (task-oriented, functional, temporal and spatial).

The concept presented in the article may constitute the basis for testing the flexibility of organisational behaviour of employees. It can also be used as a basis for research on broadly understood changes in this area, as has been said before. Both directions of research seem to be equally prospective, as it should be expected that in the era of such dynamic changes, the methodological recognition of organisational behaviour will be increasingly desirable. It can also be used for research on the relationship between the implemented organisational changes and behaviours, as well as the assessment of their psychological, economic, social, and professional costs. The methodical description of behaviour, based on the FOBE, also enables the tracking of the direction and pace of the transformations taking place.

On the basis of this concept, it is possible to perform both a qualitative analysis – a comparison of the results obtained for individual categories and the size of the differences between them; an indication of the dominant elasticity and the secondary elasticity; and a qualitative interpretation of the identified behaviours, and quantitative – defining the level of general flexibility and the levels of partial flexibility.

Indicators can be used to assess the level of flexibility of organisational behaviour, to estimate, on the basis of the sum of the identified indicators and their frequency, how often an employee undertook flexible behaviour. However, their proper application in various areas of organisation and management requires respecting certain rules:

when one interprets the results as desired or undesirable, the requirements of the job and the characteristics of the entire organisation should be taken into account each time;

The principles formulated above are not limiting. However, since the FOBE concept is used to study organisational behaviour of employees and thus individual patterns of behaviour of an individual, there is a justified fear of overly instrumental treatment of the tool and a simplified interpretation of the results. That is why it is so important to properly define the rules and scope of using the obtained data.