You are on page 1of 5

CHAPTER 4: ANALYSIS AND FINDINGS 4.

1 INTRODUCTION The units of analysis in this research are six classes of construction firms which were regrouped into three (3) pairs of sample classes (Construction Firms AB, Construction Firms CD and Construction Firms EF). The three group discriminant analysis was adopted, to explore and test for possible differences in e-tendering readiness among three classes of construction firms. Specifically, the objective is to determine if there are significant mean differences among the three classes of construction firms on E-tendering readiness based on their score profile and discriminatory power of set of attribute variables measuring four critical factors of e-tendering implementation vis-a-vis 1] information communication technologies for e-tendering; 2] people attitudinal readiness 3] e-tendering in construction process; and 3] organisation and financial readiness. Tables 3.2 to 3.4 show the attribute variables for each of the factors and abbreviations used in the research. 4.2: READINESS ON INFORMATION COMMUNICATION TECHNOLOHIES FOR E-TENDERING Table4.1 shows the class means score differences among 3 classes on 9 variables used to determine their level of readiness in requisite technology for E-tendering. In the table the attribute variables registered moderate total means ranging from 3.0140 to 3.4558 and total standard deviation ranging from .98815 to 1.12358. But a pattern seems to exist in the distribution of class means. The means follow a descending order from the highest class of contractors (A-B) to the lowest (E-F). Class A-B registered highest mean in 8 of the 9 of the variables: T_1 ; use of internet tools; T_3, use of integrated system; T_4, adoption of ict improve situation; T_5, adequate IT infrestructure; T_6, focus on ICT skill development; T_7, have basic internet requirements; T_8, willingness to install software to run e-tendering system; T_9, sufficient hardware to run e-tendering system. Consequently, on the bases of requisite information communication technologies for e-tendering calss AB is could be said to be most ready on these 8 variables. Class CD had highest mean on T_2, (use of collaboration system) and could be said to have the best use of collaborativesystem required for e-tendering. In the 9 variables Class CD higher mean than class EF. In other words the higher the class of construction firms the higher their e-readiness on information communication technologies for e-tendering. Test of equity of class mean (Table 4.2) shows that all 9 demonstrated strong discriminatory power were highly significant (.001significant level) in differenting among the three classes 32

of construction firms on the basis of e-readiness on information communication technologies for e-tendering. Table4.1: Class Mean Difference on Information Communication Technology for ETendering Class Perception Variables Total M/D TM T_1 T_2 T_3 T_4 T_5 T_6 T_7 T_8 T_9 3.1070 3.0884 3.0140 3.1907 3.1860 3.0837 3.4558 3.3767 3.2140 TD 1.08201 1.04863 .98815 1.10475 1.08636 1.00581 1.21388 1.12840 1.12358 A-B Class CM 3.9583 3.6458 3.4375 3.6875 4.0000 3.7292 4.1875 3.9375 3.9167 CD .65097 .69923 .61562 .74822 .85053 .73628 .78973 .72658 .82083 C-D Class CM 3.3659 3.3902 3.3415 3.5366 3.5000 3.4024 3.9756 3.9024 3.5732 CD .74556 .76588 .74070 .70615 .68943 .66406 .75319 73046 .91666 E-F CLASS CM 2.3765 2.4824 2.4588 2.5765 2.4235 2.4118 .5412 2.5529 2.4706 CD 1.09083 1.15069 1.10790 1.29446 1.03942 1.02695 1.20072 1.13919 1.03035

TM = Total Mean; TD Standard Deviation; CM = Class Mean and; CD = Class Standard Deviation Table 4.2: Test of Equity Class Mean on Inform-Com Technology for E-Tendering

4.2.1 Predicting discriminant function for class readiness on information communication technologies for e-tendering One of the objective of this study is tu build a model that contain only the most important predictive variables that best diffrentiate among the three classes of constuction firm on the basis of a] readiness on information and communication technology for e-tendering; and d] organisation and financial readiness for e-tendering as well as classify their membeship accordingly To this end, all attributes of readiness in information and communication technology for etendering subjected to stepwise method of enter / remove which selects the variable that maximizes the overall Wilks Lambda at each step. Table 4.3 shows that 18 iterations and

32

.05 significant level, 6 out of the 9 variable entered the model in the following descend order of magnitude of stepwise Wilks Lambda T-7, T_1, T_8, T_9, T_3 and T_5. Table 4.3: Stepwise Statistics for Infomation CommunicationTechnologies for E-Tendering (Variables Entered/Removeda,b,c,d)

At each step, the variable that mainimizes the overall Wilks Lambda is entered. a. Maximum number of steps is 18. b. Minimum partial F to remove is 3.84. c. Maximum partial F to remove is 2.71. d. F level, tolerance, or VIN insufficient for futher computation

Testing the significance of the model: Table4.4 presents statistics for evaluating the model as whole verifying the hypothesis that the three classes of construction firms have equal mean discriminant function score (U1=U2=U3 otherwise U1 U2 U3) on the attributes of readiness in information and comunication technology for e-tendering. The result indicates that discriminant function (DF1) has the strongest power of dicrimination on the three class of construction firms because 1) it has eigenvalue of .827 and accounts for whooping dispersion ratio (variance) of 84% in the class means, 2) it also has stronger canonical correlation coefficient (CCr2=45.3%, that is (.6732)2) which depicts the degree of relationships between the discriminant functions and the dependent variables (three class of constructuon firms). On the other hand, DF2 exhibits weaker power of discriminatio, having eigenvalue of .157, dipertion ratio of 16%, and CCr2 of 14% (.368)2. Verifying the hypothesis: Consistent with the eigenvalues and canonical correlation, the values of Wilks Lambda (=.473, X2(dif = 10) = 30.58) for test of D1 through DF2 has a P-value mush less than .05 significant level. While the value Wilks Lambda (=.864, X2(4) = 30.58) for test of DF2 has P-value less than .05 significant level. As a result we rejects the hypothesis of equal class mean and conclude that there is at least one significant unequal discriminant function that separates the three classes of construction firms based on their rediness in infomation and communication technology for e-tendering Table 4.4: Summary of Canonical Discriminant Function Information Communication Technology for E-Tendering 32

Eigenvalues

a. First 2 canonical discriminant functions were used in the analysis.

Wilks Lambda

Important Discriminanting Variables: Standardised discriminant function coefficient is used to assess the unique contributation of each variable to discriminant function and to the model. From Table 4.5, it could be seen that the variables that

contributed are most to the DF1 in order of magnitude are T_3, = -.730; T_5, = .648; T_1, = .562. To DF2 the variable are T_8, = 1.766; T_9, = -.103; T_1, = .840, T_3, = .551. Table 4.5: standardized Canonical Discriminant Function Coefficients for Information Communication Technology

Function 1 T_1 T_3 T_5 T_8 T_9 .562 -.730 .648 .300 .275 2 -.840 .551 -.315 1.766 -1.030

The function of group centroids provides optimal centroid on which the three classes of construction firms could classify or preceivd as being ready for e-tendering on the basis of information and comunication technology for e-tendering. Taking the cut-off value or mid-point for three classes to be zero. A movement above zero (+) is considered in relative terms be ready while a movement below zero (-) is considered in relative terms to be not ready.

Table 4.6: Functions at Group Centroids for Information Communication Technology

32

Function C_1 Class AB Class CD Class EF 1 1.182 .403 -1.056 2 -.522 .469 -.157

Unstandardized canonical discriminant functions evaluated at group means

Table 4.6 shows that at the first function which most significant function, class AB and class CD have positive (+) centroids and therefore predicted to be significantly ready or equiped for e-tendering. Class AB is however more equiped than class CD. Class EF. This is a statisfactory hit ratio and suggest that the model has practical significance in differentiating the 3 classes of construction firms based the readiness in information and communication technology for e-tendering.

Table 4.7: Classification Accuracy Results for Information Communication Technology Predicted Group Membership C_1 Original Count Class AB Class CD Class EF % Class AB Class CD Class EF Class AB 26 9 3 54.2 11.0 3.5 Class CD 18 54 19 37.5 67.9 22.4 Class EF 4 19 63 8.3 23.2 74.1 Total 48 82 85 100.0 100.0 100.0

a. 66.5% of original grouped cases correctly classified

32

You might also like