A user acceptance model for e-HRM system implemented in IT and Non-IT business organizations

2.1 Research gap

The research objectives were set after an extensive review of the literature which suggested that though much has been highlighted, the inter-segmental (IT & non-IT) differences need to be addressed. The best practices to accommodate these differences have yet to be highlighted. Technology firms are better able to accept external technology knowledge than non-tech firms. E-HRM being a web technology-based system, requires technical know-how, which IT firms have in their very inherent nature. At the organizational and individual levels, e-HRM acceptance and usage depend on perceived usefulness, ease of use, attitude, and behavioral intentions associated with organizational systems. Since IT work culture differs from work culture in the non-IT sector, the acceptance level for e-HRM systems may differ in these two sectors, and finding out the actual scenario is the objective behind this study. A firm with businesses in IT adopts HR management practices very different from those of firms dealing in other services. Technology firms have a better capability to absorb external technology knowledge [18] successfully. There needs to be more initiative to understand e-HRM practices under the umbrella of IT/non-IT sectors, though many organizations have implemented various initiatives within the last few decades to deal with this web-based technology. This is a big gap as, going by the current trends, the work culture in these two segments will converge slowly to a common platform. The practices and culture in IT are gradually creeping into the non-IT regarding technology introduction, implementation, and acceptance. This exchange of practices is needed in the current scenario as non-IT companies structured mainly around the old homogeneous model are diminishing their potential to grow and compete in a global marketplace. At the same time, IT companies also envisage a stable and mature organizational culture to sustain their business models. Hence for both sectors to survive and thrive, there is an inherent value in incorporating each other’s best e-human resource management practices.

3.1 Questionnaire description

In this section, questionnaires used for the study incorporate a set of questions with a choice of answers for the statistical and empirical study of the effectiveness and usefulness of the e-HRM system in an organization. The layout of this questionnaire rests on the research questions that address the issue of ‘how e-HRM can be best and effective for the organization.’ Generally, the questionnaire can be prepared based on two survey methods: (a) self-administered questionnaires and (b) interviewed-administered questionnaires [18]. The Self-administered survey method [19] is conducted by sending the questionnaires the hard copy or soft copy electronically (i.e., via mail or email through the internet). The interview survey method is performed through telephonic interview recordings on the phone or in person. Here we have employed the self-administrated questionnaire (SAQ) survey method. Here the SAQ method is executed without any feedback process. We minimize the data collection errors, double negatives, confusing words, and ambiguous words in the questionnaire-based e-Survey method.

The questions in the questionnaire are well-formatted and made available online for the web-based survey purpose to maximize the information gathered while incorporating the cross-cultural data collection designs. The aim is to measure the user perception of factors of the e-HRM acceptance model that affect an organization’s human resource management performance. So, using this e-study based on a questionnaire, we try to find the fruitfulness, effectiveness, and success of the usability of the e-HRM system within the organization. Here, we have introduced twenty-two questions (variables) for the questionnaire (shown in Fig. 1 in the Annexure section). The questionnaire variables have been fixed based on some literature reviews [20] discussed above in the context of e-HRM technologies and their applications. Each variable has been framed as a sentence. The answer type of each variable is categorized into five scales such as Strongly Agree (scaled 5), Agree (scaled 4), Neither Agree Nor Disagree (scaled 3), Disagree (scaled 2), and Strongly Disagree (scaled 1) respectively corresponds to each variable. Here, each variable has positive sentences, while some negative sentences are introduced to the 14, 15, 17, 18, 19, and 20 variables. To maintain the unbiasedness between the answers of the variables, these positive and negative sentences have been introduced in the proposed questionnaire. Here, the five scales of answers to negative sentences are set as Strongly Agree (scaled 1), Agree (scaled 2), Neither Agree Nor disagree (scaled 3), Disagree (scaled 4), and Strongly Disagree (scaled 5), respectively.

The questionnaire (Fig. 1 in the Annexure section) has 22 variables, each with five scales to be answered. The items of e-HRM acceptance measure the experiences, knowledge, comfortability, level of difficulties, and understandability of the use of the e-HRM model by the employees. Each item of the questionnaire is responded to in five Likert scales: ‘Strongly Agree’ (scaled 5), ‘Agree’ (scaled 4), ‘Neither Agree Nor Disagree’ (scaled 3), ‘Disagree’ (scaled 2) and ‘Strongly Disagree’ (scaled 1)). The objective of using these variables is to study the acceptance of e-HRM technology for IT as well as Non-IT organizations and derive the prediction model that will bring the following criteria:

A cause-effect relationship is established between IT and other organizations by statistically analyzing the combined effects of IT and HR capabilities and validating the e-HRM model.

So, to establish the above criteria, some rigorous data analysis requirements have been employed to prove these criteria and test the hypotheses. Working flow diagram of the employed data analysis is demonstrated in Fig. 2.

OPEN IN VIEWER

Fig. 2

Working flow diagram of the employed data analysis.

Apart from these criteria, using the proposed data analysis techniques, a well-known technology acceptance model for an e-HRM system [21, 22] has also been tested with the questionnaire variables. The employed Technology Acceptance Model (TAM) was developed by Davis (1986). It comprises two beliefs, the perceived utilities and the perceived ease of application, which determine attitudes to adopt new technologies. The attitude toward adoption will decide the adopter’s positive or negative behavior concerning new technology. External factors influencing the perceived ease of use and attitude are social (Peers and Superiors) influences, self-efficacy, and technological facility. Adapted from the study Extended TAM model, conducted by Nath, Bhal and Kapoor (2013) (Factors influencing IT adoption by bank employees: An extended TAM approach. Vikalpa, 38(4), 83-96), the e-HRM’s acceptance will be determined.

A block diagram of this TAM is shown in Fig. 3. The usefulness of TAM here comes with its capability of explaining user behavior in a wide range of end-user acceptance of computing technologies. The objective for employing this model is to validate the perception model through the parameters in the questionnaire so that the acceptability of e-HRM systems by the end users of IT and Non-IT organizations can be found. To fulfill this objective, some hypotheses are also proposed for testing the TAM for e-HRM in light of parameters employed in the questionnaire for the acceptability of e-HRM systems.

OPEN IN VIEWER

Fig. 3

Technology acceptance model (TAM). Source: Davis et al. [21].

So, to test the perception model of acceptance of the e-HRM system using Davis TAM, here, the grouping of twenty-two variables of the questionnaire is performed into six groups based on the multi-factor analysis (performed in the Results and discussion 2 section) for each mediating variable which are as follows:

OPEN IN VIEWER

Fig. 4

Grouping of principal components with respect to parameters listed in Table 7, keeping 0.0 threshold for factor values.

OPEN IN VIEWER

Fig. 5

Grouping of principal components with respect to parameters listed in Table 9, keeping 0.0 threshold for factor values.

During testing the proposed perception model of acceptance, the following hypotheses are set:

Hypothesis 1: e-HRM applications in IT and Other Select Non-IT organizations are equally accepted among users.

For the above hypothesis testing, the following data analysis techniques are performed.

3.5 Data analysis for prediction model

Several statistical-based data analysis techniques help the system predict the model and test the proposed hypotheses. So, some required statistical-based techniques employed in this work are as follows:

(a) Data reliability: Data reliability is used to evaluate the quality of research. It empirically measures how well techniques and methods are. Here data reliability is used to measure the consistency achieved for the proposed questionnaire. Several employees of inter and intra-organizations answer the same questionnaire, so it is mandatory to check whether their responses are similar. So, to check the proposed questionnaire’s reliability, we have employed Cronbach’s alpha measure in this work. This measure computes the internal consistency, i.e., finding how the features are related to a group of samples. The Cronbach’s alpha measures the dataset is (i) excellent if α≥0.9, (ii) good if 0.8≤α< 0.9, (iii) acceptable if 0.7≤α< 0.8, (iv) questionable if 0.6≤α< 0.7, (v) poor if 0.5≤α< 0.6, and (vi) unacceptable if α< 0.5. Data reliability is used to evaluate the quality of research. It empirically measures how well techniques and methods are. Here data reliability is used to measure the consistency achieved for the proposed questionnaire. Several employees of inter and intra-organizations answer the same questionnaire, so it is mandatory to check whether their responses are similar. So, to check the proposed questionnaire’s reliability, we have employed Cronbach’s alpha measure in this work. This measure computes the internal consistency, i.e., finding how the features are related to a group of samples. The Cronbach’s alpha measures the dataset is (i) excellent if α≥0.9, (ii) good if 0.8≤α< 0.9, (iii) acceptable if 0.7≤α<0.8, (iv) questionable if 0.6≤α< 0.7, (v) poor if 0.5≤α< 0.6, and (vi) unacceptable if α< 0.5.

(b) Cross Tabulation: This is a statistical mainframe model [23] that helps to find the trends and patterns and the correlations between the required study and the set of selected parameters within the data to make informed decisions for the required research. This cross-tabulation needs categorical data to examine the relationship to find readily evident or not within the data.

(c) Normality with Percentiles and z-scores: The normality tests supplement the graphical assessment for normality. Here the test for normality assessment is the Shapiro-Wilk test, which can be conducted in the SPSS. This normality test tells us that the scores in the sample are a normally distributed set of scores with the same mean value and standard deviation; the null hypothesis is “sample distribution is normal.” If the test is insignificant, the distribution is normal, and vice-versa. Here, The Shapiro-Wilk test is based on the correlation between the data and the corresponding normal scores and is advantageous over the Kolmogorov-Smirnov test. This test rejects the hypothesis of normality of scores in the sample if the p-value is less than or equal to 0.05. Failing the normality test, which means a p-value more than 0.05, will indicate that the data does not have a normal distribution. Percentile is one such word used daily without a formal definition. A percentile is considered an estimate utilized in statistics to indicate the value under which a particular percentage of considerations remains within a group of considerations. The existence of percentiles is observed in various reports such as exam scores, census, etc. For estimating relative standing, the value percentile is used as a tool. A z-score [28] is a statistical measurement. The position of a raw score concerning the distance from the measuring standard deviation is represented by a z-score. If the value of the z-score is above the mean, then the z-score is considered positive, and if the z-score value is below the mean, then the z-score is considered harmful. If the value of the z-score is zero, it simply represents that the score of the data point is uniform concerning the mean score. As the z-score can measure the variability of observation, it is widely used by traders.

(d) KMO and Bartlett’s Test: Factor analysis suitability of any data is assessed by a Kaiser-Meyer-Olkin (KMO) [29] test. The competence of sampling concerning every variable within the model and for the entire model has been assessed through the test KMO. It is one assessment regarding the scope of variance between the variables, which can be a common variance. The lower value of the scope indicates more suitability of the data for factor analysis. The value returned by KMO always remains between 0 to 1. For adequate sampling, the KMO value is between 0.8 to 1. If the KMO value is less than 0.6, it indicates inadequate sampling. Bartlett’s test checks whether k samples have equal variances or not. This test is also known as the ‘Test of homogeneity of variance.’ It examines whether the presumption of equal variances is correct without running any statistical test like ANOVA. It is used when one confirms that the data source is normally distributed. For the data from non-normal distribution, Bartlett’s test is not suitable. In that situation, one can consider another test known as Levene’s.

(e) Factor analysis or Principal Component Analysis: Principal Component Analysis (PCA) [30] is an important standard tool in the contemporary data analysis era for visualization and dimensionality reduction in diversified fields, from pure science to computer illustrations. Principal component analysis (PCA) transforms high-dimensions data into lower dimensions while retaining as much information as possible. The principle component analysis (PCA) is used for feature reduction. It is a simple, non-parametric method for extracting relevant information from confusing data sets and an unsupervised machine learning technique. The traditional PCA operates directly on a whole pattern represented as a vector and acquires a set of projection vectors to extract global features from given training patterns. Pattern recognition is the traditional PCA [30] application, an efficient approach to extracting features [31]. Nonlinear relationships between features are not captured by visualization as the standard PCA is a linear algorithm [32] which is the major limitation of this approach.

(f) Correlation Coefficient: In the Statistical measurement technique, a correlation coefficient is used to measure the strong relationship between the relative movements of two variables. It reveals a perfect negative correlation of –1.0 and a perfect positive correlation of 1.0, and a result of zero indicates no relationship at all. Correlation coefficients are widely used to identify sequences in data that may be of particular interest [24]. One very commonly used correlation coefficient, the cluster match correlation coefficient (ccc) is widely used in biological measurements like human gene expression pattern recognition [24]. CCC captures general patterns from human gene expression data and identifies robust linear relationships. One More is the Pearson’s correlation coefficient, which can check the correlative strength between two variables. It is adopted for analysing the correlation between the frequency and functions of Chinese visual verbs [25]. Pearson’s correlation is limited because it cannot tell the difference between dependent and independent variables. Correlation coefficients are also used to check the interdependency between two variables [26]. The correlation coefficient is used for clustering analysis instead of the Euclidean distance. Correlation coefficients are also used to measure feature dependency when redundant or repeated features generate for some object detection and recognition problems. Some deep learning models use the correlation coefficient to find redundant features by testing pairwise feature dependency. It is also used to measure maximal dependency that causes during final feature selection for some classification tasks.

(g) Regression: The regression method [27] overcomes the limitation of the correlation coefficient. The regression method is a powerful statistical tool. The statistical technique, regression method, estimates and determines a relationship between an independent and dependent variable by providing a function. It helps predict the independent variable’s effects on the dependent one. There are three regression models, i.e., linear, non-linear, and multiple regression. Regression analysis helps an enterprise understand their data representation among various business analysis techniques and helps them make better decisions. Business Analysts and other data professionals use powerful statistical tools to remove unwanted variables and choose only the important ones. Regression predictions can be interpreted using root mean squared error.

(h) Hypothesis Assumptions: A hypothesis [33] is a predictive statement, which can be tested by scientific methods, that shows the relation between an independent variable with some dependent variables. A statistical hypothesis is a prediction done by assuming about a population that may or may not be true. Testing of these assumptions (Hypothesis Testing) [33] is to accept or reject them using a set of formal procedures used by statisticians called statistical hypotheses testing. Statistical hypotheses are of two types: null hypothesis (H0) referring to a chance-based hypothesis, and alternative hypothesis (Ha) referring to an observation-based hypothesis. Consider that we want to determine the fairness and balance of a coin. A null hypothesis might assume that coin flips will be split evenly between heads and tails, whereas an alternative hypothesis might say that heads and tails flips could be very different such as H0: P = 0.5, Ha: P ≠ 0.5. Steps of Hypothesis Testing:

The hypothesis testing consists of four steps such as:

State the hypotheses - This step involves stating null and alternative hypotheses. The hypotheses must be stated so that both null and alternative hypotheses are mutually exclusive. If one is true, then the other must be false.

(i) Pearson Chi-Square Test: The Chi-square test [34] is a non-parametric test for the significant differences between group frequencies. When working with data, we frequently only sometimes receive it in continuous data format. Data may be presented as frequencies. As an illustration, how many men and women work in an office? Do they differ considerably, indicating that there are noticeably more men working in offices or jobs than women? For instance, let’s say you work for the government and want to spread the word that there might be more males than women employed in dangerous industries like mining. Therefore, in such a situation, we may perform a Pearson chi-square test and convey the finding that mining employs much more men than women due to the hazardous nature of the profession. Let’s say we work in a field where an equal number of men and women should be employed, such as information technology. If so, it would be interesting to demonstrate that there is no appreciable difference in the proportion of men and women who work in the office. So, we can compute the Pearson chi-square test using our data and present our findings in such a situation. We cannot do parametric tests like ANOVA or t-test because the number of men and women is presented in frequency format. Therefore, we can only employ the chi-square test when our data are presented as frequencies.

(j) Mann-Whitney U: Mann-Whitney U test [43] is the non-parametric alternative test to the independent sample t-test. The Mann-Whitney U test checks whether there is a difference between the two independent groups. It is a non-parametric counterpart to the t-test for the independent samples. In the t-test, we find, “is there a mean difference”? In the Mann-Whitney U test, we find, “is there a difference in the rank sum”? Since group rank differences rather than the group mean differences are sometimes used to illustrate the results, it might be challenging to interpret the Mann-Whitney U. In this test, the data must be normally distributed. The analysis is explained in simple terms via the Intellectus Statistics tool below: As a non-parametric test, the Mann-Whitney U test makes no assumptions about the distribution of results. However, some presumptions are made: 1) The population sample taken is randomly chosen; 2) Only two independent random samples with at least ordinally scaled characteristics must be available. The variables do not have to fulfill a distribution curve.

(k) Wilcoxon W: The Wilcoxon test [34] is of two types Wilcoxon signed-rank test and Wilcoxon rank-sum test. Wilcoxon signed-rank test is used to compare two related samples, matched samples, or to conduct a paired difference test of repeated measurements on a single sample to assess whether their population means ranks differ. In comparison, the Wilcoxon rank-sum test is used to compare two independent samples. They are nonparametric alternatives to the paired and unpaired Student’s t-tests (also known as “t-test for dependent samples” or “t-test for matched pairs”), respectively. The Wilcoxon test is a nonparametric test and is therefore subject to significantly lower requirements than its parametric counterpart, the t-test for dependent samples. Thus, the Wilcoxon test is used as soon as the boundary conditions for the t-test for dependent samples are no longer fulfilled. Since the Wilcoxon test is a nonparametric test, the data need not be normally distributed. However, to calculate a Wilcoxon test, the samples must be dependent. Dependent samples are present, for example, when data are obtained from measurement repetition or when so-called natural pairs are involved. A person’s characteristic, e.g., weight, was measured at two points. The values do not come from the same person but from persons who belong together, for example, lawyer/client, wife/husband, and psychologist/patient. If the data are unavailable in pairs, the Mann-Whitney U test is used instead of the Wilcoxon test.

By using above these statistical-based data analysis techniques, the following list of tests are performed to derive the prediction model and testing of hypotheses.

Implementing effective HR systems is difficult, but firms frequently need to invest in their people and invest correctly. Many firms are unaware of the consequences of their investment decisions involving people. Introducing e-HRM systems in companies cannot solely justify the considerable investment into installing this newer technology-based HRM, the e-HRM. Proper implementation, goal realization, and its impacts on HR in particular and the organization as a whole will finally decide the success of e-HRM systems being widely introduced and implemented in various companies. Different companies may have different e-HRM goals, different levels of implementation, and varied impacts of e-HRM applications. Also, the acceptance of technology varies from organization to organization. The conceptual framework for this study was developed through a review of the literature and is a depiction of the potential track that results in the appropriate planning and components necessary to successfully e-HRM implementation and acceptance.

3.6 Data collection

The proposed questionnaire contains questions, and the employees have given responses to those questions. Here, the answers to queries from each segment have their importance and are used to obtain a sample corresponding to each user. These answers undergo the data collection and arrangement tasks, where the questionnaire responses are arranged sequentially to their objectives. The questionnaire proposed is collected through a Google forms framework that effectively separates the questionnaire representation contents. Here, the Sampling technique is that the organizations must be large enough to have a minimum of 10 HR professionals, and the organization must use the e-HRM systems to perform HR activities. The service organizations included in the study comprise Banking, Electric-distribution companies, Hospitals, Insurance Companies, and others. The data are collected from 20 IT organizations and 20 service organizations, where ten employees of each organization have participated in giving the response to queries of the proposed questionnaire. Hence, 200 individual sets of responses are obtained from IT groups of organizations, and 200 individual sets of responses are obtained from Service organizations. In this e-HRM acceptance, the answers are rated (Likert Scale) here as ‘Strongly Agree’, ‘Agree’, ‘Neither Agree Nor disagree’, ‘Disagree’, and ‘Strongly Disagree’ for the closed-ended type questions. The acceptance of e-HRM measures the experiences, knowledge, comfortability, level of difficulties, and understandability of the use of the e-HRM model by the employees.

The below “Results and discussion 1” section is employed for the data analysis being performed for the acceptance of the e-HRM system, while the“Results and discussion 2” section is for the data analysis for e-HRM acceptance through the Technology Acceptance Model (TAM).

4.1 Comparison of IT and Non-IT using variable-wise analysis of response frequency (Test 1)

Here this test is needed (i) to find the percentages of the matchup of categories of responses to the questionnaire variables and (ii) to check the independence of categorical variables in the parameters. The proposed questionnaire contains questions about accepting the e-HRM system used for IT or Non-IT based organizations. In organizations with the e-HRM model, the employees of those organizations can respond to the questionnaire questions to make a more convenient and effective decision system. Here, in the questionnaire, there are five categories of responses (Strongly Agree, Agree, Neither Agree nor Disagree, Disagree, and Strongly Disagree) shown in Table 2. So, from Table 2 it is shown that for IT organizations, the percentage (%) within the Strongly Agree group, Agree, Neither Agree Nor Disagree, Disagree, and Strongly Disagree groups scores 56.39%, 10.13%, 13.23%, 18.50%, 1.75% whereas the percentage (%) within Strongly Agree, Agree, Neither Agree nor Disagree, Disagree and Strongly Disagree scores 35.25%, 39.05%, 16.20%, 7.25%, and 2.25% respectively for Non-IT organizations. Hence, it is concluded that the score for the Strongly Agree + Agree groups is much higher in IT organizations than in Non-IT organizations. To measure the significant difference in these observed frequencies of five categories (Strongly Agree, Agree, Neither Agree nor Disagree, Disagree and Strongly Disagree) in IT and Non-IT, Pearson Chi-Square has been performed that measures the test statistic of 18.655 with 0.001 p-values which is less than 0.05 corresponds to each category. Hence, from the cross-tabulation and the Pearson Chi-Square test, it is concluded that there are significant differences in the observed frequencies of five categories of parameters, and the values of these observed frequencies of five categories are much higher for IT than Non-IT organizations.

4.2 Comparison of IT and Non-IT using normality Tests for Statistical Analysis (Test-2)

Here, the parameter responses of the questionnaire undergo a normality test where the probability p is obtained corresponds to the observations of IT and Non-IT groups to find the normality in the distribution of parameters corresponding to each group of the organization. So, after the normality test (shown in Table 3), it has been observed that the distribution is not normal in each group. For this, the non-parametric tests such as Wilcoxon signed rank test, Mann-Whitney U Test, and Kruskal-Wallis are applied. Hence, these test statistics reveal that the distribution of values in terms of responses to the questionnaire in both IT and Non-IT organizations is probably not normally distributed in the population. Again, the mean score of the collected samples is higher for IT organizations than for Non-IT organizations. The reason behind this is that the higher scaled categories like ‘Strongly Agree’ with 5 scaled are much answered in the corresponding parameters of the questionnaire respectively.

Hence, the distribution is not normal, so some nonparametric tests Wilcoxon signed rank test, the Mann-Whitney U Test, and the Kruskal-Wallis are applied. The results of these tests are shown in Table 4. From this table, it has been observed that the descriptive statistics for the acceptance of e-HRM systems are different concerning IT and Non-IT organizations, and it is higher in IT than Non-IT organizations.

Hereafter normality with percentile and z score statistics, the KMO and Bartlett’s Test has been performed on all the collected datasets, i.e., 400 samples with 22 parameters. The results of these tests are reported in Table 5, and the following observations are found: (i) Here, the KMO test reports the sampling adequacy of factors is 0.886, which is between 0.8 and 1. Hence, the dataset of the proposed questionnaire is eligible for factor analysis; (ii) Bartlett’s Test of Sphericity reports that the Chi-Square test statistic is 2832.549 and the corresponding p-value is 0.000, which is not higher than our significance level (let’s use 0.05). Thus, the collected dataset is likely suitable for principle component analysis (PCA) or factor analysis.

After KMO and Bartlett’s Test, the principal component analysis (PCA) (discussed at) has been applied to the collected dataset, i.e., on 400 samples with 22 parameters. Hence, during PCA, initially using the whole dataset, 22 Eigenvalues are obtained. Out of these 22 Eigenvalues, only four have values greater than 1. So only these four principal components are considered for computing principal components corresponding to the parameters of the dataset. So, after computing the principal components using four Eigenvalues, the factors under these principal components result in 51.68% variance, shown in Table 6. Hence, after analyzing the rotated component matrix of these four principal components, it has been observed that 1, 2, 3, 5, 6, 7, 8, 9, 10, 11, 12 factors are under principal component-1, factor 15, 17, 18, 19, 20 under principal component-2, factor 1, 16 under principal component-3 and factor 13 under principal component-4 of the proposed questionnaire. These results are reported in Table 6. Here, for convenience, Component 1 can be grouped by the name’ User perception and attitude’, Component 2 can be grouped by the name ‘User behavioral intention’, Component 3 can be grouped by the name ‘User skills set’, and Component 4 can be grouped by the name ‘User experience. The parameters under these components are listed in Table 7, and Fig. 3 shows the grouping of principal components concerning parameters listed in Table 7, keeping a 0.0 threshold for factor values.

From the above multi-factor analysis based on the combined IT and Non-IT organization datasets, we got the following (Table 8) number of principal components (factors) corresponding to each e-HRM system acceptance study variable.

Here, non-parametric tests like Mann-Whitney U and Wilcoxon W are performed to determine the impact of these four principal components in IT and Non-IT organizations. So, after the test, the following findings are obtained:

Table 8 shows that the mean for Component 1 of IT is +0.14 while the mean for Component 1 of non-IT is –0.14, which means that IT people have a positive disposition or favorable disposition for Component 1 (i.e., the factors under this category). At the same time, Non-IT people have a negative disposition/ unfavorable disposition for Component 1, and the Test-Statistics show that this comparison is significant.

Hypothesis 1: e-HRM applications in IT and other service organizations are equally accepted among their users.

Discussion for Hypothesis 1: Grouping of questionnaire items to study the “acceptance of e-HRM systems” in Organizations-22 questions are grouped into 4 variables. This section contains 22 items (factors) (Q1 to Q22). For this hypothesis, we have performed above (i) Test-1: Comparison of IT and Non-IT using yariable-wise analysis of response frequency for determining the percentage or counts for the combinations of categories Strongly Disagree (scaled 1), Disagree (scaled 2), Neither Agree nor Disagree (scaled 3), Agree (scaled 4), Strongly Agree (scaled 5), (ii) Test-2: Comparison of IT and Non-IT using normality Tests for Statistical Analysis for describing the descriptive statistics for the acceptance of e-HRM systems is different with respect to IT and Non-IT organizations, (iii) Test-3: Comparison of IT and Non-IT using Wilcoxon signed-rank test and Mann-Whitney U tests and (iv) Test-4: Comparison of IT and Non-IT using factor analysis. These above test analyses obtain the following decision for Hypothesis 1 that: (i) these test analyses reject the null hypothesis, and accept the alternative hypothesis, and (ii) the e-HRM applications in IT and Non-IT organizations are not equally accepted among its users.

5.1 Test-4: Comparison of IT and Non-IT using factor analysis

Here for Technology acceptance model testing, the factor analysis of acceptance variables of the e-HRM system is performed considering the Eigenvalues greater than 0.9. This factor analysis is shown in Table 9.

Six different components are required to test the TAM model, so, from the factor analysis of acceptance of the e-HRM system performed in Table 8, six factors (components) are derived. These six factors (shown in Table 10) are used to test the Davis TAM model.

Hence, from Tables 9 10, the following conclusions are found:

Here, the KMO test reports the sampling adequacy of factors is 0.886 which is in between 0.8 and 1. Hence, the dataset of the acceptance questionnaire is eligible for factor analysis.

The Correlation Coefficient of combined effect of IT and Non-IT along with Pearson correlation test for TAM using External variables (EV), Perceived Usefulness (PU), Perceived Ease of Use (PEOU), Attitude towards using (ATU), Behavioral intention (BI), Actual System use (ASU) parameters obtained by grouping of variables of the employed questionnaire of acceptance of e-HRM have been demonstrated in Table 11. Similarly, the Correlation Coefficient of individual effect of IT and Non-IT along with Pearson correlation test for TAM using using External variables (EV), Perceived Usefulness (PU), Perceived Ease of Use (PEOU), Attitude towards using (ATU), Behavioral intention (BI), Actual System use (ASU) parameters, have been demonstrated in Table 12.

Here, for testing Technology Acceptance Model for e-HRM, the objective is to test the proposed model of acceptance of e-HRM using Davis’s TAM, where the grouping of 22 parameters in the questionnaire is performed into six groups such as External Variables (EV), Perceived Usefulness (PU), Perceived Ease of Use (PEOU), Attitude towards using (ATU), Behavioural Intention to Use (BIU), Actual System Use (ASU). These groups of variables are discussed above. For TAM, here the factor analysis (Test 4), and Correlation Coefficient Based Analysis (Test 5), are performed at Tables 9, and 12 respectively. So, using the results reported in Tables 11, and 12, the following hypotheses are tested.

Hypothesis 2: External variables have equal effect on perceived ease of use of e-HRM systems in select IT and other service organizations.

Hence, from the above hypothesis testing it can be concluded that parameters employed for e-HRM systems validates the TAM model that methods employed for e-HRM systems accord to TAM and the corresponding e-HRM acceptance model is given in Fig. 6 which is based on our analysis.

OPEN IN VIEWER

Fig. 6

User acceptance model for e-HRM system based on our analysis.

Hence, finally, the Regression-based analysis (Test 6) has been performed to evaluate IT and Non-IT e-HRM system based equations through the TAM model. Here, the regression based analysis of combined effect of IT and Non-IT along for TAM using External variables (EV), Perceived Usefulness (PU), Perceived Ease of Use (PEOU), Attitude towards using (ATU), Behavioral intention (BI), Actual System use (ASU) parameters obtained by grouping of variables of the employed questionnaire of acceptance of e-HRM have been demonstrated in Table 13. Similarly, the regression based analysis of individual effect of IT and Non-IT using External variables (EV), Perceived Usefulness (PU), Perceived Ease of Use (PEOU), Attitude towards using (ATU), Behavioral intention (BI), Actual System use (ASU) parameters, have been demonstrated in Table 14.

The final outcome of regression equation from Table 13 are as follows:

IT: Regression Model

Actual System use = 2.89 + 0.02 * External Variables + 0.01 * Perceived Usefulness + 0.06 * Perceived Ease of Use –0.01 * Attitude towards using –0.16 * Behavioral Intention to Use

Non-IT: Regression Model

Actual System use = 5.88 + 0.07 * External Variables + 0.0 * Perceived Usefulness –0.1 * Perceived Ease of Use –0.08 × Attitude towards using –0.09 * Behavioral Intention to Use