Comparison of employee branding through discriminant analysis

Chia sẻ: Huỳnh Ngọc Toàn | Ngày: | Loại File: PDF | Số trang:11

Thêm vào BST

Báo xấu

36
lượt xem 2
download

Download Vui lòng tải xuống để xem tài liệu đầy đủ

Through the Analysis it is clear that if the companies are able to maintain employee Branding then they will definitely earn profit through the most important factor i.e. satisfaction of employees which is converted into customers satisfaction.

Chủ đề:

Bình luận(0) Đăng nhập để gửi bình luận!

Lưu

Nội dung Text: Comparison of employee branding through discriminant analysis

International Journal of Management (IJM) Volume 7, Issue 7, November–December 2016, pp.395–405, Article ID: IJM_07_07_044 Available online at http://www.iaeme.com/ijm/issues.asp?JType=IJM&VType=7&IType=7 Journal Impact Factor (2016): 8.1920 (Calculated by GISI) www.jifactor.com ISSN Print: 0976-6502 and ISSN Online: 0976-6510 © IAEME Publication COMPARISON OF EMPLOYEE BRANDING THROUGH DISCRIMINANT ANALYSIS Dr. D.K. Ghosh UGC BSR Faculty Fellow, Retired Professor and Head of Statistics Department, Saurashtra University, Rajkot, India Shweta S. Kulshrestha Assistant Professor, School of Management, RK. University, Rajkot, Gujarat, India ABSTRACT The company needs to maintain the employee brand and due to which the satisfaction level of the employee is maintained and the effectiveness of the organisation can be improved. In my data analysis I have collected the data from the employees of Amul group of companies, Balaji Wafers Prvt. Ltd., Echjay Industries Prvt. Ltd., Falcon Pump Prvt. Ltd., P.M. Diesels Prvt. Ltd., Jyoti CNC Prvt. Ltd., Kadwani Forge Ltd., Patel Brass Work Prvt. Ltd., Saral group of Industries and Thirth Agro Technology Prvt. Ltd. I have collected the data through preparation of questionnaire and have analyzed the data using discriminate analysis for analyzing the employee branding impacting the organizational success through SPSS software. Through the Analysis it is clear that if the companies are able to maintain employee Branding then they will definitely earn profit through the most important factor i.e. satisfaction of employees which is converted into customers satisfaction. Key words: Employee Branding, Customer Satisfaction, Favorable Reputation, Employee Satisfaction Cite this Article: Dr. D.K. Ghosh and Shweta S. Kulshrestha, Comparison of Employee Branding Through Discriminant Analysis. International Journal of Management, 7(7), 2016, pp. 395–405. http://www.iaeme.com/IJM/issues.asp?JType=IJM&VType=7&IType=7 1. INTRODUCTION TO DISCRIMINANT ANALYSIS The term discriminant analysis (Fisher; 1936; Cooley and Lohnes; 1971; Tatsuoka; 1971; Kshirsagar; 1972; Lachenbruch; 1975, 1979; Gnanadesikan; 1977; Klecka; 1980; Hand; 1981, 1982; Silverman; 1986) refers to several different types of analyses. Classificatory discriminant analysis is used to classify observations into two or more known groups on the basis of one or more quantitative variables. Classification can be done by either a parametric method or a nonparametric method in the DISCRIM procedure. A parametric method is appropriate only for approximately normal within-class distributions. The method generates either a linear discriminant function (the within-class covariance matrices are assumed to be equal) or a quadratic discriminant function (the within-class covariance matrices are assumed to be unequal). When the distribution within each group is not assumed to have any specific distribution or is assumed to have a distribution different from the multivariate normal distribution, nonparametric methods can be used http://www.iaeme.com/IJM/index.asp 395 editor@iaeme.com
Dr. D.K. Ghosh and Shweta S. Kulshrestha to derive classification criteria. These methods include the kernel method and nearest-neighbor methods. The kernel method uses uniform, normal, Epanechnikov, biweight, or triweight kernels in estimating the group-specific density at each observation. The within-group covariance matrices or the pooled covariance matrix can be used to scale the data. The performance of a discriminant function can be evaluated by estimating error rates (probabilities of misclassification). Error count estimates and posterior probability error rate estimates can be evaluated with PROC DISCRIM. When the input data set is an ordinary SAS data set, the error rates can also be estimated by cross validation. In multivariate statistical applications, the data collected are largely from distributions different from the normal distribution. Various forms of nonnormality can arise, such as qualitative variables or variables with underlying continuous but nonnormal distributions. If the multivariate normality assumption is violated, the use of parametric discriminant analysis might not be appropriate. When a parametric classification criterion (linear or quadratic discriminant function) is derived from a nonnormal population, the resulting error rate estimates might be biased. If your quantitative variables are not normally distributed, or if you want to classify observations on the basis of categorical variables, you should consider using the CATMOD or LOGISTIC procedure to fit a categorical linear model with the classification variable as the dependent variable. Press and Wilson (1978) compare logistic regression and parametric discriminant analysis and conclude that logistic regression is preferable to parametric discriminant analysis in cases for which the variables do not have multivariate normal distributions within classes. However, if you do have normal within-class distributions, logistic regression is less efficient than parametric discriminant analysis. Efron (1975) shows that with two normal populations having a common covariance matrix, logistic regression is between one-half and two-thirds as effective as the linear discriminant function in achieving asymptotically the same error rate. Do not confuse discriminant analysis with cluster analysis. All varieties of discriminant analysis require prior knowledge of the classes, usually in the form of a sample from each class. In cluster analysis, the data do not include information about class membership; the purpose is to construct a classification. See Chapter 11, Introduction to Clustering Procedures. Canonical discriminant analysis is a dimension-reduction technique related to principal components and canonical correlation, and it can be performed by both the CANDISC and DISCRIM procedures. A discriminant criterion is always derived in PROC DISCRIM. If you want canonical discriminant analysis without the use of a discriminant criterion, you should use PROC CANDISC. Stepwise discriminant analysis is a variable-selection technique implemented by the STEPDISC procedure. After selecting a subset of variables with PROC STEPDISC, use any of the other discriminant procedures to obtain more detailed analyses. PROC CANDISC and PROC STEPDISC perform hypothesis tests that require the within-class distributions to be approximately normal, but these procedures can be used descriptively with nonnormal data. Another alternative to discriminant analysis is to perform a series of univariate one-way ANOVAs. All three discriminant procedures provide summaries of the univariate ANOVAs. The advantage of the multivariate approach is that two or more classes that overlap considerably when each variable is viewed separately might be more distinct when examined from a multivariate point of view. Multiple linear regression method is used to predict an outcomes. However, multiple linear regression is limited to cases where the dependent variable on the Y axis is an interval variable so that the combination of predictors will, through the regression equation, produce estimated mean population numerical Y values for given values of weighted combinations of X values. But many interesting variables are categorical. For Eg. Marine/non-marine status, making a profit or not, holding a particular credit card, renting or paying a mortgage for a house, employed/unemployed, satisfaction v/s dissatisfaction of employees, which customers like to buy a product or not to buy and satisfaction or not satisfaction of employee branding. http://www.iaeme.com/IJM/index.asp 396 editor@iaeme.com
Comparison of Employee Branding Through Discriminant Analysis In such cases it is not advisable to use multiple linear regressions. Instead of multiple linear regressions one must use another method for prediction. One of such method is discriminant analysis method. In other words we can say that discriminant analysis can also be used for predicting an outcome considering the above facts we can say that discriminant analysis is used for the following situations: • The dependent variable is categorical at an interval level such as age, sex, income attributes, year of education, and perception etc. It is also to be noted that as in the case of multiple linear regression, the dummy variable also can be used as predictors in case of discriminant analysis. Moreover logistic discriminant analysis method also can be used for any level of measurements. • There are more than two dependent categorical unlike logistic regression method which is limited to a dichotomous dependent variable. In our investigation our problem is to find whether the employee branding is satisfying the organizational brand and creates a competitive tool or not i.e. our dependent variables are of two categories (1) yes or (2) no. Hence it will be more appreciable to apply discriminant analysis method for such problems. Null hypothesis (H0): Organizational branding does not have a significant difference in the working conditions of the employee in their respective organization. Alternate Hypothesis (H1): Organizational branding have a significant difference in the working conditions of the employee in their respective organization. Next we will discuss the linear equation. 1.1. Discriminant Analysis Linear Equation Discriminant analysis method, requires the determination of a linear equation like regression that will predict which group the case belongs to. The methodical model of discriminant analysis linear equation is given by: Where D = discriminate function v = the discriminant coefficient or weight for that variable X = respondent’s score for that variable a = a constant i = the number of predictor variables From figure 1 it is obvious that discriminant analysis linear equation or function is similar to a linear regression equation or function. In case of linear regression b is called regression co-efficient, where beta is called unstandardised coefficient. This maximizes the distance between the means of the dependent variables. Standardized co-efficient can also be used like b wigs in regression. However it is remarkable to note that good predictors tend to have large weights. Also number of discriminant function is always 1, less than number of groups. Here in our problem we have two group discriminant analyses (yes or no) so there is only 1 discriminant function. 1.2. Assumptions of Discriminant Analysis The major underlying assumptions of DA are: • The observations are a random sample. • Each predictor variable is normally distributed. • Each of the allocations for the dependent categories in the initial classification are correctly classified. • There must be at least two groups or categories, with each case belonging to only one group so that the groups are mutually exclusive and collectively exhaustive (all cases can be placed in a group). http://www.iaeme.com/IJM/index.asp 397 editor@iaeme.com
Dr. D.K. Ghosh and Shweta S. Kulshrestha • Each group or category must be well defined, clearly differentiated from any other group(s) and natural. Putting a median split on an attitude scale is not a natural way to form groups. Partitioning quantitative variables is only justifiable if there are easily identifiable gaps at the points of division. • For instance, three groups taking three available levels of amounts of housing loan. • The groups or categories should be defined before collecting the data. • The attribute(s) used to separate the groups should discriminate quite clearly between the groups so that group or category overlap is clearly non-existent or minimal. • Group sizes of the dependent should not be grossly different and should be at least five times the number of independent variables. 1.3. Process of Discriminant Analysis The main aim of this statistical using discriminant analysis is to combine the variable scores is someway so that a single new composite variable is produce. This is also called discriminant score. At the end of the discriminant analysis process the intention would be made that each group will have a normal distribution of discriminant scores. The degree of overlap between the discriminant scores distribution can then be used as the major of the success of techniques. In our investigation we have two different types of groups, for e.g. let us assume that the two different groups are poor and good. So its discriminant distribution is shown in fig. 1: Figure 1 Discriminanat distribution From the figure 1 it is clear that the top two distributions overlap too much and do not discriminate very good compared to the bottom sets. Hence the misclassification would be minimal in the lower pair. While many will be misclassified in the top pair. By standardizing the variables we ensure that scale differences between the variables are eliminated. So when all the variables are standardized, absolute weights can be used to rank variables in terms of their discriminant powers. Hence the largest weights have been associated with the most powerful discriminative variable. Variables with large weights are those which contribute mostly in differentiating the groups. In a two group situation, predictive membership is calculative by first producing a score or discriminant for each case using the discriminant function. Then the cases with discriminant values smaller than cut off value are classified, as belonging to one group while those with values larger are classified into other group. The groups centriod is the mean value of the discriminant scores for a given category of the dependent variable. There are as many centroids as there are groups or categories. The cut off is the mean of the two centroids. If the discriminant scores of the function is less than or equal to the cut off the case is classified as zero while if it is above then it is classified as one. http://www.iaeme.com/IJM/index.asp 398 editor@iaeme.com
Comparison of Employee Branding Through Discriminant Analysis 1.4. Materials and Method Our aim in this investigation is based on analysis of impact of employees branding on the organizational image and we want to know whether employee branding provides a good image to the organization in the market or not. Thus if the response is yes then the corporate have to give an immense importance to the employee branding to maintain its position as well as ongoing growth. To study the importance of employee branding and its impact on the regular working of the organization we have conducted the research in ten different companies including Amul Industries, Balaji Store, Echjay Industries, Falcon Pumps, Field Marshal, Jyoti CNC, Kadwani Forgien, Patel Brass Company, Saral Industries and Thirth Agrotech Company. The main investigation in my research of these organizations include the working environment been provided to the employees in their respective organization and the support from the management which in turn decides the level of work integrity from the sides of employees. For the further study of my research work we have considered the random sample of 30 employees from each industry working at different functional levels so as to justify my research work. The details of each organization including the various ages of employees and their educational levels are been mentioned below as per the details of the organization. 1.5. Conducting Discriminant Analysis 1.5.1. Discriminant analysis Dialogue Box: • To find the significant result of the responses YES/NO for the employee branding of 10 companies for 19 categories we used SPSS technique where following steps are carried out: • Analyze >> Classify >> Discriminant • Select the employee response as our grouping variable is entered it into the grouping variable box. The below given menu is displayed on the monitor. Figure 2 Discriminant Analysis Dialogue Box: http://www.iaeme.com/IJM/index.asp 399 editor@iaeme.com
Dr. D.K. Ghosh and Shweta S. Kulshrestha • Click Define range button and enter into independents box. • Click >> continue. • Select all independent variable together. • Click on statistics button and select means univarriate invoice, Box M’s, Unstandardised and within group co- relation. The menu at the window displayed the following table (figure 3.) Figure 3 Range Box • Continue >> Classify. Select compute from group assigned, summary table, leap one out classification, within groups and all plots. • Continue >> select predicted group membership and discriminant scored. Finally select OK. 1.5.2. Discriminant analysis Statistics Box Within discriminant analysis our main aim is to predict a group membership. For this purpose, first of all we examine whether there are any significant difference between the groups on each of the independent variables. Figure 4 Discriminant Analysis: Statistics This will be carried out using group means and analysis of variance (ANNOVA) tables. This information we collect from group statistics and test of equality of group mean table 1 and 2. In case no significant group difference is found then we should not proceed for any further analysis. However a rough idea of variables that made important can be obtained by inspecting the group means and its corresponding standard deviation. http://www.iaeme.com/IJM/index.asp 400 editor@iaeme.com
Comparison of Employee Branding Through Discriminant Analysis In our case from table 30 we can see that the mean differences between employee branding criteria and gender is very large. 1.5.3. Group Statistics Table In case of No Employee branding criteria and gender is very large. This suggest that there are may be good dissertators as separations are large. Again the mean difference between employee branding criteria and company name is also large. This suggest that is good discrimination between employee branding and company name. Table1 Group Statistics: Group Statistics Valid N (listwise) Response Mean Std. Deviation Unweighted Weighted No Employee Branding Criteria 9.73 5.395 1137 1137.000 Gender .60 .490 1137 1137.000 Company Name 5.15 2.836 1137 1137.000 Yes Employee Branding Criteria 10.07 5.497 4563 4563.000 Gender .86 .345 4563 4563.000 Company Name 5.59 2.875 4563 4563.000 Total Employee Branding Criteria 10.00 5.478 5700 5700.000 Gender .81 .392 5700 5700.000 Company Name 5.50 2.873 5700 5700.000 In case of Yes The mean difference between employee branding criteria and gender is large this suggest there is a good discrimination between this two. The same is true for employee branding criteria and company name. This shows that employee branding is affecting the company and makes it more competitive. 1.5.4. Tests of Equality of Group Means Table Table 2 Tests of Equality of Group Means Table Tests of Equality of Group Means Wilks' F df1 df2 Sig. Lambda Employee Branding Criteria .999 3.408 1 5698 .065 Gender .930 427.150 1 5698 .000 Company Name .996 21.757 1 5698 .000 From the table 2, we can observe that the statistical evidence of significant difference between means of Yes or NO group for all the companies with gender and company name is highly significant which can be seen from its T- value. However the statistical evidence of significant difference between means of Yes and No group for all the companies with employee branding criteria and gender as well as employee branding criteria and company name is near to significant. http://www.iaeme.com/IJM/index.asp 401 editor@iaeme.com
Dr. D.K. Ghosh and Shweta S. Kulshrestha 1.5.5. Pooled within Group Matrices Table 3 Pooled within Group Matrices Pooled Within-Groups Matrices Employee Branding Gender Company Criteria Name Correlation Employee Branding 1.000 -.007 -.002 Criteria Gender -.007 1.000 .021 Company Name -.002 .021 1.000 From table no.3 i.e. the pooled within matrices also supports the same results for all the companies because inter correlations are very low for employee brand criteria and gender, employee branding criteria and company name as well as company name and gender. In analysis of variance case the important assumption is that the variances were equivalent for each group. However this is not true for discriminant analysis case. In discriminant analysis case, the basic assumption is that variance co-relation matrix is equivalent. 1.5.6. Box M Test and Log Determinant: For box M’ Test, the null hypothesis states that the co-variance matrix do not differ between groups formed by the dependent. In case of discriminant analysis the researcher is interested that the test should be insignificant so that the null hypothesis that the groups do not differ can be retained. Table 4 Log Determinants Log Determinants Response Rank Log Determinant No 2 .636 Yes 2 -.021 Pooled within-groups 2 .163 The ranks and natural logarithms of determinants printed are those of the group covariance matrices. To hold this assumption the log determinant value should be equal. But in our case from table no. 4 we can see that the log determinant of Yes and No are not more or less the same. The value of No is 0.636 while that of Yes id -0.021. This shows that the employee branding is not been considered by most of the selected companies which is affecting their working and outcomes. Table 5 BOX M’s Test Test Results Box's M 303.231 F Approx. 101.010 df1 3 df2 60401976.020 Sig. .000 Tests null hypothesis of equal population covariance matrices. http://www.iaeme.com/IJM/index.asp 402 editor@iaeme.com
Comparison of Employee Branding Through Discriminant Analysis Remark With large samples a significant result is not regarded as too important. So considering this if there are 3 or more than 3 groups and Box M is significant, and then the groups with very small log determinants should be deleted from the analysis. In our case, we have only two groups No and Yes and hence relation of groups do not play any role here. Again from table no.5 if we test using Box M method, then we are looking for non- significant differences. In our investigation the log determinants do not act similar. However the value of Box M is 303.231 with the f value 11.010. It is significant as the value is 0.000. This shows that our null hypothesis is rejected. In other words we can say that, organisation and employee branding have a significant difference in the working condition of the employee of their respective organisation. 1.5.7. Eigen Value Table The eigen value provides information on each of the discriminant function produced. The maximum number of discriminant function produced no. of groups – 1. In our investigation we have two groups namely Yes and No. so number of discriminant function is 2- 1= 1. The canonical correlation is defined as the multiple co-relations between the predictors and the discriminant functions. Table 6 Eigen value Eigen values Function Eigen value % of Variance Cumulative % Canonical Correlation 1 .078a 100.0 100.0 .269 a. First 1 canonical discriminant functions were used in the analysis. In our case we have only one function so it provides an index of overall model fit, which is interpreted as being the proportion of variance explained ( R2). From table 6 we observe that there is only one function where eigen value is 0.078a 1.5.8. Wilks Lambda Table Table 7 Wilks Lam.bda Value Wilks' Lambda Test of Function(s) Wilks' Lambda Chi-square df Sig. 1 .928 428.374 2 .000 Wilks Lambda criteria describes the significance of the discriminant function. From the table 7 we observe that the p value is 0.000 and hence the discriminant function is highly significant. This provides the proportion of total variability not explained, which is the converse of the squared canonical correlation. Moreover from table no.7, it is observed that the Wilks table value is 0.928, so we have 92.8% unexplained. 1.5.9. Standardized canonical Discriminant Analysis Co-efficient Table Table 8 Standardized Canonical Discriminant Function Coefficients Standardized Canonical Discriminant Function Coefficients Function 1 Gender .975 Company Name .200 http://www.iaeme.com/IJM/index.asp 403 editor@iaeme.com
Dr. D.K. Ghosh and Shweta S. Kulshrestha 1.5.10. Structure Matrix Table Table 9 Structure Matrix Value Structure Matrix Function 1 Gender .980 Company Name .221 Employee Branding Criteriaa -.007 Pooled within-groups correlations between discriminating variables and standardized canonical discriminant functions Variables ordered by absolute size of correlation within function. a. This variable not used in the analysis. 1.5.11. Canonical Discriminant Function Co-efficient Table Table 10 Canonical Discriminant Function Coefficients Canonical Discriminant Function Coefficients Function 1 Gender 2.578 Company Name .070 (Constant) -2.472 Unstandardised coefficients The unstandardised co-efficient (b) are used to obtain the discriminant function. On this investigation the discriminant function can be expressed as D = -2.472+ (2.578*Gender) + (0.70*company name). The discriminant function coefficient (b) or standardized form of beta both indicates the partial contribution of each variable to the discriminant function controlling for all other variables in the equation. This can also be used to assess unique contribution of each variable to the discriminant function and hence provide information on the relative importance of each variable. 1.5.12. Functions at Group Centroids Table Table 11 Functions at Group Centroids Functions at Group Centroids Response Function 1 No -.560 Yes .139 Unstandardised canonical discriminant functions evaluated at group means http://www.iaeme.com/IJM/index.asp 404 editor@iaeme.com
Comparison of Employee Branding Through Discriminant Analysis 1.5.13. Classification Result Table Table 12 Classification Results Classification Resultsa,c Predicted Group Membership Response Total No Yes Original Count no 174 963 1137 Yes 301 4262 4563 % no 15.3 84.7 100.0 Yes 6.6 93.4 100.0 Cross-validatedb Count no 174 963 1137 Yes 301 4262 4563 % no 15.3 84.7 100.0 Yes 6.6 93.4 100.0 a. 77.8% of original grouped cases correctly classified. b. Cross validation is done only for those cases in the analysis. In cross validation, each case is classified by the functions derived from all cases other than that case. c. 77.8% of cross-validated grouped cases correctly classified. 2. CONCLUSION The classification table is simply a table in which the rows are the observed categories of the dependent, while the columns are the predicted categories. When prediction is perfect all cases will lie from the diagonal. The % of cases on the diagonal is the % of correct classification. The cross validated set of the data is the most honest presentation of the power of the discriminant function. Then that provided by the original classification and often to this is a very poor result. The cross validation is often called a jack knie classification. In this classification if successively all cases but want to develop a discriminant function and then categories the case that were left out. This process is repeated with each case left out in turn. However this cross validation produces a more reliable function In case of no were classified with 15.3%, then the yes were classified with 93.4%. From the table number 12 we observe that 54.35% (15.3+93.4/2) of responses were classified into yes/no groups. REFERENCES [1] SPSS software for analysis. [2] Anderson, T. W. 1958. Introduction to multivariate statistical analysis. New York: John Wiley & Sons, Inc. [3] Cooley, W. W., and P. R. Lohnes. 1971. Multivariate data analysis. New York: John Wiley & Sons, Inc. [4] Dempster, A. P. 1969. Elements of Continuous Multivariate Analysis. Reading, MA: Addison- Wesley. [5] Dixon, W. J. 1973. BMD Biomedical computer programs. Los Angeles: University of California Press. [6] Tatsuoka, M. M. 1971. Multivariate analysis. New York: John Wiley & Sons, Inc. [7] Gaurav Gupta and Urvashi Singh, A Study of ERP System for Employee Satisfaction in An It organization. International Journal of Management 5(10), 2014, pp. 25–32. http://www.iaeme.com/IJM/index.asp 405 editor@iaeme.com