System Design Features and Repeated Use of Electronic Data Exchanges

Download 269.61 Kb.

Page	5/5
Date	09.08.2017
Size	269.61 Kb.
	#29289

1 2 3 4 5

TABLE 6

Measured Model Results

Panel A: T1 Measured Model Results

	Including PIQ Effects on Intent		Excluding PIQ Effects on Intent
	PiQ	Intent	PiQ	Intent
-Control Transparency - Outcome Feedback - Perceived Info Quality -Structural Assurance	.45;7.16** .11;1.43ns	.04;0.66ns .25;4.47 .40;4.89 .37;4.34**	.49;10.07** .10;1.13n.s.	.16;2.68 .29;5.15 ------ .59;11.39**
R²	21.4%	58.4%	25%	50.5%

Panel B: T2 Measured Model Results

	Including PIQ Effects on Intent		Excluding PIQ Effects on Intent
	PiQ T2	Intent T2	PiQ T2	Intent T2
-Control Transparency - Outcome Feedback - Perceived Info Quality -Structural Assurance	.50;8.02 .21;2.97	.04;0.74ns .30;4.99 .48;6.46 .31;5.12**	.51;8.98 .21;3.02	.21;3.77 .39;7.10 ------ .57;9.91**
R²	30.0%	70.7%	31.1%	60.8%

Numbers in cells represent path coefficients and corresponding t-values after the semi-colon.

One-tailed Significance Levels: *: p<.05 (2.326>t>1.645); **: p<.01 (t>=2.326)

TABLE 7

Differences In Path Coefficients Between T1 And T2 Measured Model Results

Perceived information quality^ξ

Intent to Use

- Control Transparency

- Outcome Feedback

- PIQ

- Structural Assurance

-7.20 **

-12.48 **

-0.31 n.s.

-7.78 **

-8.78 **

7.03 **

^ξ: Numbers in cells represent t-values on differences of respective path coefficients as measured in the T1 and T2 research models (as shown on left half of Table 6).
Note: Negative t-values denote increases from T1 to T2; positive t-values denote decreases.

One-tailed Significance Levels: *: p<.05 (2.326>t>1.645); **: p<.01 (t>=2.326)

TABLE 8

Summary of Hypothesis Testing Results

Research Hypothesis: Test (table reporting results):	H1a (PIQ will not be affected by positive outcome feedback at T1; but it will be affected at T2)	H1b (High PIQ at both time periods under high control transparency)	H2 (At T2, negative outcome feedback will eliminate effects of control transparency on Intent to Use)	H3 (mediation hypothesis: PIQ will mediate effects of control transparency on Intent to Use at both T1 and T2)
MANOVA (Table 4)	Supported	Supported	n/a	n/a
Planned Contrasts (Table 5)	Supported	Supported	Supported	n/a
Structural Model – PLS (Table 6)	n/a	n/a	n/a	Supported
Structural Model – differences in PLS path coefficients (Table 7)	n/a	Supported	n/a	n/a
Sobel mediation tests (untabulated)	n/a	n/a	n/a	Supported

TABLE 9

Areas for Future Research

Topic	Author(s)	Issues/Models
Cross-Theory Tests	Petty and Cacioppo [69] Karney and Bradbury [42]	Elaboration Likelihood Model and similar multi-process models. Vulnerability-stress-adaptation model.
Cross-IS Research Domain Tests	Rustagi et al. [75] Robey et al. [73]	Outsourcing. Information systems development.
Cross-Methods Tests	Cole and Maxwell [14] Karney and Bradbury [42]	Expand measurement to more time periods. Use methods other than self-reports.

System Design Interventions:

- Control Transparency (T1, T2) - Outcome Feedback (T1, T2)

autoshape 10

Perceived Information Quality T2

Perceived Information Quality T1
oval 13

Intention to Use T1
oval 9

Intention to Use T2

autoshape 37

Structural Assurance T1

Structural Assurance T2

Figure 1. Research Model

rectangle 40

Data Validation:

-under high control transparency only

Validity

Parse XML file
group 65

Data Input

Generate XML
group 72

Completion

Content

rectangle 63

Feedback

to Customer:

-General

-Positive

-Negative

rectangle 57

Database

Processing

Figure 2. Workflow of Simulated Data Exchange System

Figure 3. Screenshot of Validation Screen -- High Control Transparency Condition

Figure 4. Screenshot of Outcome Feedback -- Negative Feedback Condition

Personality variables measurement

Training

autoshape 344

autoshape 341

Experimental Treatments (T1, T2)

Measurement Instrument

(T1, T2)

2 weeks interval

autoshape 339

Training

Figure 5. Progression of Experimental Steps

1 Subject attrition was mainly due to work responsibilities of participants that caused their absence from their location of contact within the two-week interval in which our study was conducted.

2 We also tested our models by splitting the sample between students (N=55) and non-students (N=90) with no changes in our hypothesis test conclusions.

3 The test statistic is estimated as T= (b₁-b₂) / [ S_p √ (1/N1 + 1/N2) ] (1a)

where, T has the t-distribution with (N1+N2-2) degrees of freedom;

b₁ and b₂ are the T1 (phase 1) and T2 (phase 2) path coefficients being compared;

N1 and N2 are the sizes of the T1 and T2 samples;

S_p is the estimator of the pooled sample variance, which is constructed using the standard deviations of the T1 and T2 samples and equals the square root of:

S²_p = [(N1-1) S²_T1 + (N2-1) S²_T2] / [N1+N2-2] (1b)

Download 269.61 Kb.

Share with your friends:

1 2 3 4 5