To test the hypothesized model, a Structural Equation Model (SEM) was used. The Statistical Package for Social Sciences (SPSS 28) and Analysis of Moment Structures (AMOS 28) was used for the study. The research analysis was conducted using two-step approach. Measurement model and Structural models were tested. The measurement model was checked for validity, internal consistency and reliability. To test the scale items Confirmatory Factor Analysis (CFA) was used. Present study reported Comparative Fit index (CFI), Root Mean Square Error of Approximation (RMSEA), Root Mean Residuals (RMR). The six latent constructs of the measurement model are tested to check if all the coefficients indicate FWFH. The coefficient values show that work dependence, work life balance and work environment are significant determinants of FWFH.
Extensive literature review has revealed the existing models developed by various researches. The Model framework proposed by Nordin et al., 2016 is as under. Previous research findings and the model framework set by Nordin et al., 2016 was studied. The change in the circumstances advocate the need for supplementary variables to the existing model. We would like to study the moderating effect of pandemic lockdown on employee preference to WFH post pandemic.
As it is identified by many researchers and evident from the previous literature that job performance is one of the essential components in the study of work from home. The authors Garg and van der Rijst (2015) have studied the relationship between the job performance and professional isolation. Job performance and work from home are related and are inter dependent. When there is clear understanding of job performance and when the job indicators are quantifiable, work from home possibility is more even after pandemic. Therefore, it is hypothesized as there is a positive influence of job performance on work from home in future.
In past research was directed towards the importance of telecommuting and increasing work dependence (Vana et al., 2008). The study made by Garg and van der Rijst (2015) found that work dependence had a weak positive relation to experience with virtual work. The focus of present study is to assess the willingness of employees to work from home post pandemic. The present study is during the peculiar times of Covid 19 which makes the concept of WFH a unique one.
Another important component of factors influencing willingness to work from home in future (FWFH) is Social Interaction. Previous studies (Baumeister & Leary, 1995) have highlighted that work from home with less social interaction in employees will make them aggravated due to isolation. Mintz-Binder & Allen, 2019 observed the factor social contact in terms of virtual meetings and online interactions. Many researchers in the past have focussed on the need to maintain firm and well-built interpersonal social relationships. There exists a negative influence of social interaction on work from home in near future.
Raghuram and Fang (2014) have studied the role of the supervisor in controlling the employees working from home. Previously Lautsch et al. (2009) have studied the general perceptions regarding supportiveness of supervisors. Madlock (2012) has studied the leadership styles and their results suggested that supervisors occupied in work oriented more than relational oriented leadership style in the virtual workplace.
According to Wheatley (2012), work from home eliminates the workplace related distractions and allows to work productively without interruptions. The results of the present study are in agreement with the study conducted by Golden (2007) which pointed out that the virtual technology like e-mail and online-conferences to interact with other employees lack the warmth and social presence of face-to-face interaction.
Study conducted by Venkatraman et al. (1999) emphasised that working overtime informally without any extra payment affects the personal life of the employees. The study conducted by Tietze and Musson (2010) elicits that balance between work and home is essential to understand the relationship between household and professional life. The results of the present study agreed with a balanced work and family life will have greater willingness to work from home. Thus, the proposed hypothesis is that there is a positive influence of work life balance on the employee’s willingness to work from home in future (FWFH).
Demographic Profile of Respondents
The study consisted of 138 participants working from home during the lockdown. 21% of respondents were female whereas 79% were male. The largest group 58% fall in the age group of 18–25 years, 34% of respondents were in 26–35 years of age group and 36–45 years of the age group is represented by 8% in the current study. The largest group 50% are Professionals (None of them are front end medical workers), 24% are IT software employees and others represent 26% (Design engineers, BPO employees and backend support). In terms of the highest educational qualification, 45% of participants were degree/diploma holders, 40% were postgraduates and 16% were holding a professional qualification. None of them were below graduation level, the group is mature.
The responses were complete in all aspects. There is no missing data in the columns. Also, observed quite normally distributed data of our latent factors and other variables like job performance, work dependence, social interaction, supervisor’s role, work environment and work life balance. To measure the multivariate normality, kurtosis and skewness measures were used which was generated using AMOS 26. The data exhibited normal distribution which ranged from −1.3 to 2.04. The threshold value for Kurtosis and Skewness is −2 to +2 (Byrne, 2010). However, the value of 2.04 does not violate the normality. The threshold is 3.3 according to Skarpness, 1983. This number indicates a good fit. Multivariate Analysis was suggested by Hu & Bentler, 1998 as an indication of goodness of fit. The multivariate measure in the study is 15.472 at critical ratio 1.298. The data is perfectly well behaved.
The present study has attempted to explore the structural relationship between the multiple factors relating to Work from Home. Questions were measuring the variables on five point Likert scale. This was run in SPSS 28 using Varimax with Normalization method for rotation. The rotation and iteration were run until the ultimate clear pattern matrix arrived. The factor patterns arrived under each column were thoroughly diagnosed to understand the plausible cross-loadings of factors and elimination of redundant variables (Brown & Moore, 2012). Six factors were identified under different heads like job performance (JP), work dependence (WD), work life balance (WLB), social interaction (SI), supervisor’s role (SR) and work environment (WE). These six factors explained were calculated from the sum of squared loadings from the structure matrix. The total accumulated variance explained is 71.709% for work from home during pandemic. The total variance explained by first factor job performance is 13.65%, the second factor work dependence is 13.656%, work life balance is 12.965, social interaction is 12.450, supervisor’s role is 10.392 and work environment is 8.998. Absolute values below 0.5 were eliminated. During the principal axis factoring, few items cross loaded on another component and few items in scale were deleted due to low factor loadings. An item in the job performance scale “There are objective criteria by which my performance can be evaluated” was cross loaded on supervisor’s role component during factor analysis. Third item in work life balance was deleted due to poor loading. The rotation converged in 7 iterations. Bartlett’s Test of Sphericity was significant at 000 indicating the result was acceptably valid. In addition to this, the model fit indices were verified for the proposed factor structure. The CFA result yielded an adequate fit. The CMIN = 164.268, CMIN/df = 1.711, CFI = 0.922, RMSEA = 0.08, RMR = 1.55 (See Appendix Table 3). The overall model exhibited a good fit. The Harman single factor test was used for examining if the problem of common method variance (CMV) exists or not. All the factors have not significantly loaded on a single factor. This test confirms that CMV is not a significant problem in this study.
The job performance scaled on three measures. It is easy to measure and quantify employee performance (with path coefficients = 0.932), the measures of employee job performance are clear (with path coefficients = 0.829), the feeling that employee engagement is more during the lockdown (with path coefficients = 0.704). The hypotheses that there exists a positive influence of job performance on employee’s willingness to WFH in future is refuted with estimate of 0.003 at p value greater than 0.05. There is a negative influence of work dependence on employee’s willingness to WFH in future. In this factor three aspects of work dependence are measured, the extent to which the employee performance depends on working with others (with path coefficients 0.892), the need to work independently for performing the best (with path coefficients 0.872), the nature of work in terms of independent task or projects (with path coefficients 0.675). All three are significant with p value less than 0.05. However, the study has revealed the negative influence of Work Dependence on employee’s willingness to work from home in future post pandemic situation. It may be inferred that the higher degree of WFH is associated with weakened work dependence. This is due to the inter-dependence of departments for work completion. Like for example, the dependence on IT department for setting up remote access to all the employees for completion of work during the sudden lockdown. Next, social interaction was measured. The first item, social interactions are more in the current lock down situation (deleted due to low loadings), The work-related meetings in my office are adequate to build good working relationships (with path coefficient 0.915), the social events in virtual office are adequate to build a sense of community (with path coefficient 0.725). The research hypotheses relating to negative influence of social interaction on employee’s willingness to WFH in future is refuted in the current study. The relationship between social interaction and willingness to WFH in future is −0.193 at p value greater than 0.05. Thus, we refute the hypothesis.
The results of the present study hypothesize that there is a positive influence of supervisor’s role on employee’s willingness to WFH in future has been refuted. In the present study focused on three aspects of supervisory role. The first being close supervision of work during the lockdown (with path coefficients 0.902). Secondly, employees understanding on the criteria for evaluating the performance was studied (with path coefficients 0.760). Lastly, the support extended by the superior in addressing problems during the lockdown (with path coefficients 0.673) was studied. The supervisor’s role estimated −0.002 at p value more than 0.05. Thus, hypothesis is rejected under study that there is a positive influence of supervisor’s role on employee willingness to WFH in future.
Hypothesis results have revealed that there is a significant negative influence of work environment on employee’s willingness to WFH in future (with path coefficients −0.245). In this factor, three aspects of work environment were measured, the interruption caused when colleagues talk in virtual meetings (with path coefficients 0.746) and the distraction caused by other things going on in the work environment, such as background noise (with path coefficients 0.802) and feeling of pressure because meetings take away from work (with path coefficients 0.632) are measured under this head. Moreover, it consumes lot of productive time to effecting work particularly for the complex type of tasks. It may be inferred that the higher degree of willingness to WFH is associated with weakened work environment.
Work life balance is measured using three items. Overall comfort working from home (with path coefficient 0.630), employee’s ability to balance both work and household during the lock down (with path coefficient 0.909) and feeling of difficulty in maintaining work life balance due to the pressure of remaining available all the time (deleted due to low loadings). There is a positive influence of work life balance on employees willingness to WFH in future with regression estimate of 0.546 at p value less than 0.05. It may be inferred that higher degree of work life balance has an incremental effect on willingness to WFH.
Assessment of Reflective Model
Cronbach Alpha was used to assess the inter item consistency between measurement variables. Cronbach’s Alpha for all the factors put together was 0.708. Post factorization, the Cronbach’s Alpha for job performance was 0.750, work dependence was 0.844, work life balance was 0.75, social interaction was 0.64, superior’s role was 0.781 and work environment was 0.66. All these values are above 0.6 indicating acceptable internal consistency (Nunnally, 1978). Next, Composite Reliability (CR) was assessed. CR values ranged from 0.753 to 0.865 higher than minimum requirement of 0.7 (see Appendix Table 4).
Convergent validity was assessed using Average Variance Explained (AVE). The AVE values ranged from 0.533 to 0.684 higher than 0.5 threshold. The factor loadings exceeded 0.5 minimum requirement (Fornell & Larcker, 1981). Thus, Convergent Validity was assured.
Discriminant validity is assured by comparing the square root of AVE and inter-correlations between other constructs as exhibited in Appendix Table 5. The diagonal bold numbers in the table indicate square root of AVE and the non-diagonal numbers are the correlations between constructs signifying discriminant validity.
It is very important to take utmost care while designing the questionnaire. The questionnaire was simple in its structure and the language used was easy to understand. This was principally designed to get better content validity.
Structural Model Testing
In the structural model analysis, multi-dimensional model was hypothesised and tested for significance. While testing the objectives under the study, it was encountered that three out of six path coefficients were considered statistically significant. Work dependence (with path coefficients −0.345), work environment (with path coefficients −0.245), work life balance (with path coefficients 0.546) are significantly related to employee willingness to WFH in future post pandemic. While job performance, social interaction and supervisor’s role are not statistically significant (See Appendix Table 6).
As predicted in Hypothesis 2, work dependence is negatively associated with FWFH (β = −0.345, p < 0.05). Hypothesis 5, work environment is negatively associated with FWFH (β = −0.245, p < 0.05). Hypothesis 6, work life balance is negatively associated with FWFH (β = 0.546, p < 0.05). Hypothesis 2, 5 and 6 are supported.
Unexpectedly, Hypothesis 1 that states that there exists a positive influence of job performance on FWFH was not supported. Hypothesis 3, that there is a negative influence of social interaction on FWFH was also not supported. Finally, Hypothesis 4, that there is a positive influence of supervisor’s role on FWFH was also not statistically significant (See Fig. 1)