Decision
As in chapter 2, Decision was operationalized as a single, eight (8) point semantic differential scale question in which the subject was asked to choose to “Test Further” or “Move to Production”. Anchor points for the variable were “Definitely Test Further” and “Definitely Move to Production.” Intermediate points were “strongly”, “somewhat” and “slightly” on each side of the scale. I used an eight-point scale to force the subject to select one of the two alternatives presented. A lower value indicated a stronger desire to change the course of action (i.e., to delay for testing). A higher value indicated a
stronger desire to continue the existing course of action (i.e. move into production). See Appendix E, item 1, for the item that collected the Decision data.
Relevance
Here, I utilized the same scales employed in chapter 2. Relevance was operationalized as a set of four, seven (7) point Likert scale questions anchored with “Strongly Agree” and “Strongly Disagree” on the end points and “neutral” in the mid-point. A lower value indicated that the subject held the bad news reporter’s message to be of lower relevance to the decision. The items used to elicit these responses are listed below. See also Appendix E, items 7-10.
24. Sandy’s assessment was highly relevant in forming my decision.
25. Sandy’s assessment was very important in forming my decision.
26. My decision was most influenced by the Sandy’s assessment. 27. My decision was more influenced by the Sandy’s assessment than
any of the other views expressed.
Bad News Reporter Credibility
I utilized the three item measures developed in chapter 2 as a starting point and added two additional questions to improve the scale reliability. As described above, I improved the measurement of this construct by replacing one question, revising one and adding two more items measuring Credibility. I did this to improve the reliability of the construct over that reported in Cuellar, et al. (2006) which was somewhat low at 0.745, although acceptable. I drew on the source credibility literature (Pornpitakpan 2004) to add two questions dealing with the key dimension of Credibility: trustworthiness and expertise. Additionally, I revised a question to reflect the effect of organizational power on
Credibility by asking about the bad news reporter’s perception in the organization. Bad news reporter credibility was operationalized as a set of five, seven (7) point Likert scale questions anchored with “Strongly Agree” and “Strongly Disagree” on the end points and “neutral” in the mid-point. A lower value indicated that the subject believed the bad news reporter to have lower Credibility. The items used to elicit this information are listed below. See also Appendix E, items 15-19.
This construct was previously operationalized at a reflective construct. However, in examining the items in the light of the new advice in Petter, Straub and Rai(2007), I find that I must reclassify this construct as a formative construct. Following their decision rules in table 2 of their paper, I recognize the following: 1) the direction of causality for most of the items is from the item to the construct. Trustworthiness and expertise are dimensions that define the perception of Credibility (Pornpitakpan 2004). Similarly, a perception of a proper motivation and respect are causal factors of Credibility (Near and Miceli 1995). However, the direct question about Credibility is reflective of Credibility existing. 2) the indicators are not interchangeable or separable. To lose one would be to underspecify the construct. 3) These indicators will not necessarily covary with each other. Expertise and trustworthiness may vary separately from each other. The other indicators are similar. 4) Antecedents of the indicators are different from each other. Trustworthiness and expertise arise from two different sets of factors. Similarly, respect and perceived desire will have different antecedents. I therefore conclude that this is a formative construct which I are measuring with four formative indicators and a reflective indicator.
Risk Propensity
I adapted the Risk Propensity measures used in Sitkin and Weingart (1995). Risk propensity was operationalized as a set of three, seven (7) point Likert scale questions anchored with “Strongly Agree” and “Strongly Disagree” on the end points and “neutral” in the mid-point. A lower value indicated increasing reluctance to agree to a risky course of action. The items used to elicit this information are listed below. See also Appendix E, items 21-23.
6. I would choose a risky alternative based on the
assessment of others on whom I must rely
7. I would choose a risky alternative relying on an assessment that is high in technical complexity.
8. I would choose a risky alternative which could have a major impact on the strategic direction of my organization
Risk Perception
For Risk Perception, I adapted the measures used in Sitkin and Weingart (1995). Risk perception was operationalized as a set of four, seven (7) point Likert scale questions anchored with “Strongly Agree” and “Strongly Disagree” on the end points and “neutral”
1. Sandy is the most credible person in the scenario. 2. Sandy is well respected within the company.
3. Sandy is motivated by a desire to see things done correctly for the bank. 4. Sandy is very trustworthy.
as the mid-point. A higher value indicated that the subject believed the project was more likely to be successful, and therefore of lower risk. The items used to elicit this data are listed below. See also Appendix E, items 11-14.
9. This project has a high probability of success. 10. This project is in a positive situation.
11. I believe that there is very little risk in moving this project into production.
12. I believe that there is high potential for a positive result in putting this project into production.
Role Prescription
Role Prescription was measured by a single Likert scale item created for this study, with higher values indicating that the bad news reporter was role prescribed to report bad news. This variable was used as a manipulation check. See Appendix E, item 20. Control Variables
The following constructs were entered into the analysis as control variables.
Gender. Gender was dummy coded (female = 0 and male =1).
Age related information. Age and Years of Full Time Work Experience were entered in chronological years reported by the subjects.
4 EFFECTS OF SOCIETAL COLLECTIVISM ON THE DEAF
EFFECT
The previous studies have developed a model at the individual level of how the Deaf Effect occurs (figure 4.1). At base, the Deaf Effect occurs when the decision maker considers the report of bad news to be not important or irrelevant to be considered in their understanding of the project. When the report of bad news is considered to be not relevant the decision maker tends to decide not to change the course of the project. Also involved in the decision are the decision maker’s propensity to assume risk and the perceived level of risk in the project. The Credibility and Role Prescription of the bad news reporter have been shown to strongly affect the perceived Relevance of the bad news reporter’s message.