Measuring Your Organization’s Boardroom Language

Abstract

This chapter endeavors to answer the question, “Can a reliable and valid instrument be developed from concepts outlined in the boardroom boldness model?” By invoking the eight-step guidance of Robert Devellis and soliciting a sampling size of 256, eight sub-scales were produced. Each sub-scale registered an acceptable Cronbach’s alpha and thus provided an affirmative answer to the research question. The utilization of such sub-scales can help organizations make better empirical choices, better manage the ethical health of their cultures, and help to cultivate a climate of bold followership.

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Thus far, this book has explored the boardroom language of David’s advisers, more specifically, the actions of Joab, commander of the royal army, were closely scrutinized to understand better the dilemma of serving king-think . Next, the personality of Gad was explored for how to speak in the life of a person blinded by narcissism . While discussing the ramifications of organizational dis-eases, this book utilized the literature to understand best practices for speaking out against immoral behaviours, particularly in a nonviolent way. Additionally, this study confronted the reality that often the best prescription to break the fever of organizational dis-ease is to shake up the top executives or convince them of the nobility of stepping down. Finally, the role of organizational elders was discussed, and it was pointed out that at the end of the day, the people had the influence to turn the ship around. The question became, “Were they willing to step it up?”

As entities prepare to navigate the plethora of challenges affiliated with a twenty-first-century workforce, it would be an unwise gesture to expect top executives to have all the answers and to be constantly on top of their game. For organizations to thrive in the information age, those who sit around the literal or metaphorical boardroom must find and activate their voice. This voice could very well be the difference between success and defeat, relevance or irrelevance, life and death. If the premise of this book is that boardroom boldness is the ultimate competitive edge, the question before the reader now becomes, “What is the organization’s boardroom language of your team?” It has been determined that there are five concepts affiliated with Boardroom Boldness. The following is an attempt to understand if there are one or more scales that support the theory of this book.

The Boardroom Boldness Language Scale

In the context of the information age and the onslaught of tools such as Total Quality Management, Process Management, and Six Sigma Lean, it can be argued that unless one is adequately taking a measure of work, it is not leadership. To the credit of practitioners and scholars, the literature has shifted from merely watching the bottom line to monitoring the overall organization. To illustrate, Frost suggests that organizations are now inquiring about four key elements, which the Balanced Scorecard Model wants to understand:

• Financial—How do we look to shareholders?

• Processes—Are we improving how work is done?

• Growth—Are we renewing for continued growth?

• Customers—How do we look to our customers?¹

The challenge of this book is to embrace a Balanced Scorecard approach that sufficiently addresses the research question and produces a reliable and a validated scale to measure corporate boardroom boldness languages. Such an endeavor could help to mitigate ethical mishaps and help organizations to optimize performance proactively (Tables 8.1, 8.2, 8.3).

Table 8.1 Boardroom boldness items

Table 8.2 Age

Table 8.3 Gender

Step 1: Determine Clearly What It Is You Want to Measure

The development of boardroom boldness scale(s) utilized the guidance of DeVellis, who contends that the construction of a tool to measure a phenomenon should adhere to eight guidelines. The first step involves determining clearly what it is one wants to measure. Although this is an obvious point, DeVellis encourages the researcher to think through questions such as, “Should the scale be based on theory, or should you strike out in new intellectual directions? How specific should the measure be? Should some aspect of the phenomenon be emphasized more than others?”²

Therefore, this process attempted to gage the most advantageous way to measure the intangibles of boardroom language as delineated in this book to create a scale that evaluated an organizational citizen’s propensity to:

• 1. Shut up —To instantaneously and silently obey orders.

• 2. Speak in —The ability to utilize truth as a tool to transform a leader’s paradigm and their toxic behavior.

• 3. Speak out —The utilization of peaceful and purposeful means to adjudicate a wrong in a public manner.

• 4. Step down —The ability of a leader to remove themselves for the health of the organization as well as themselves.

• 5. Step it up—The ability of the organizational citizen to proactively act to heal and transform the culture.

Step 2: Generate an Item Pool

DeVellis contends that the second step of scale development is to generate an item pool and offers several practical recommendations, which include: (a) devise a large pool of items; (b) utilize language that is easy for a reading level between the fifth and seventh grades (this is the level for newspapers); and (c) write positively worded items.³ Taking account of DeVillis’ insights, Table 8.1 shows the initial item pool for the five scales.

Step 3: Determine the Format for Measurement

The third step of scale development is to establish the format for measurement. Although DeVellis suggests that steps two and three are related, he stresses that careful consideration should be given to the format. Thus, this endeavor proposed utilizing a seven-item Likert design. DeVellis indicates that “a good Likert item should state the opinion, attitude, belief, or other construct under study in clear terms.”⁴ As such, the following composition guided the items.

1	–	7
Not at all		All the time

Step 4: Have the Initial Item Pool Reviewed by Experts

The fourth step of scale development is to have the original item pool reviewed by a panel of experts. DeVellis defines an expert as “colleagues who have worked extensively with the construct in question or related phenomena.”⁵ The panel for this study consisted of five scholars with a strong command of instrument development. Their task, as delineated by DeVellis, was multifaceted and included:

• 1. Confirm or invalidate the selected definitions of the phenomenon. More specifically, the experts were asked to rate online how relevant they thought each item was with regards to measuring the various phenomena (1 = very relevant, 2 = somewhat relevant, 3 = neutral, 4 = related to the concept, 5 = very related to the concept).

• 2. Comment freely on individual items for improvement.

• 3. Evaluate each item’s clarity and conciseness.

• 4. Point out additional ways to tap into the phenomena that the researcher may have failed to include.

• 5. In addition to rating the items, the major feedback from the panel was included.

Step 5: Consider Inclusion of Validation Items

The fifth step of scale development revolved around the inclusion of validated items. To this end, a decision was made to include all items that received a rating of neutral or better from the panel of scholars. This process discarded item 47 due to redundancy and wording.

Step 6: Administer Items to a Development Sample

The sixth step of a scale development as prescribed by DeVellis is to administer the scale to a sample. There is much debate on what constitutes an adequate number for a sample size. Nummally and Bernstein contend that the sample should include at least 300 people, since such a figure will proactively defuse the unstable factor regarding patterns of covariation among the myriad items.⁶ Whereas, DeVellis suggests that 5–10 participants per item is acceptable. To this end, the sample size peaked at 340.

A web-based company was utilized to help randomly solicit participants who were informed that they were invited to take a survey that would take approximately 10–15 minutes to complete, and that their participation would help to understand a follower’s propensity better to speak out or to obey an unethical order. The survey was available to those with access to the Internet and who lived in the United States. The participants understood that if they did not feel comfortable completing the confidential survey, they could opt out at any time.

Step 7: Evaluate the Items

DeVellis suggests that the seventh step of scale development is to evaluate the items. The primary intent of item analysis is to identify entries that form a consistent internal scale and to eliminate other items. This study adhered to such guidance by employing version 25 of the Statistical Package for Social Scientists (SPSS) to understand if there were one or more scales affiliated with the five concepts of this book. First, a decision was made to remove items that were either incomplete or contained flawed data. Although 340 participants initially engaged in the study, 84 entries were discarded, which reduced the sample size to 256. It should be noted that the new sample size, N = 256, remained within DeVellis’ five per item guidance and is therefore adequate for scale development.

The demographics of the sample as depicted in Tables 8.2, 8.3, 8.4, and 8.5 was 37.5 percent between the ages of 18–29, 35.2 percent were 30–44, 12.9 percent were 45–60 and 14.5 percent were 60 or older. There were 67.2 percent female and 32.8 percent male, with a household income that ranged from $0–$9999 to $200,000+. The sample were located across the USA, 8.6 percent of the sample were from New England, 10.5 percent from the Middle Atlantic, 11.7 percent from the East North Central, 3.5 percent from the West North Central, 21.9 percent from the South Atlantic, 6.6 percent from the East South Central, 9.8 percent from the West South Central, 12.9 percent from the Mountain and 14.5 percent were from the Pacific region. Ethnic demographics were not collected in the survey.

Table 8.4 Income

Table 8.5 Region

Data Analysis of the Shut-Up Concept

SPSS version 25 was employed to perform an analysis of the data. Specifically, Pearson correlation was applied to items SU1, SU2, SU3, SU4, SU5, SU6, SU7, SU8, SU9, SU10, SU11 and SU12 (see Table 8.1) to measure the degree of linear relationship between two or more variables.

As depicted in Table 8.6, there was evidence of a positive relationship between the variables at the 0.01 level (one-tailed) and the 0.05 level (two-tailed). Hair et al. contend that the correlated variables suggest the direct oblique rotation solution is appropriate for exploratory factor analysis in such a case. Moreover, the literature suggests that items that load at 0.40 or above are acceptable in factor analysis.⁷ To this end, loadings that fail under this threshold were suppressed.

Table 8.6 Shut-up correlation matrix

The Kaiser-Meyer-Olkin (KMO) test and the Bartlett test of sphericity were conducted. KMO accesses how suitable data is for factor analysis, and measures sampling adequacy for each variable in the model. Additionally, the KMO measures the proportion of variance among variables that might be common variance.⁸ The value returns of the KMO range from 0 to 1. Kaiser provides the following rule of thumb for the values returned (0.00–0.49 unacceptable, 0.50–0.59 miserable, 0.60–0.69 mediocre, 0.70–0.79 middling, 0.80–0.89 meritorious and 0.90–1.00 marvelous).⁹ The KMO returned a value of 0.860. The Bartlett test of sphericity is “a statistical test for the presence of correlations among variables… It provides statistical significance that the correlation matrix has significant correlations among at least some of the variables.”¹⁰ Thus, the KMO and the p-value that registered at 0.000 suggest there was enough evidence to conduct a factor analysis.

A principle component analysis was conducted on items SU1–SU12. O’Rourke and Hatcher posited that the best method to understand oblique rotation is to, “always review the pattern matrix to determine which groups of variables are measuring a given factor, for purposes of interpreting the meaning of that factor.”¹¹ To this end, a pattern matrix was generated and two factors for the shut-up concept were identified. The analysis also identified cross-loadings on items SU5 and SU6. Hair et al. maintain that when a variable is found to have more than one significant loading, it becomes a candidate for deletion.¹² As such, these items were deleted, and a component analysis was employed on the remaining ten items.

An interpretation of Table 8.7 reveals no additional cross-loadings and the existence of two factors. Component 1 factored items SU1, SU2, SU3, SU4, SU7 and SU8 that were labeled shut-up and comply. Component 2 was comprised of items SU9, SU10, SU11, and SU12 that were labeled shut-up and sabotageas depicted in Table 8.9. A reliability analysis was conducted that produced a Cronbach’s alpha, which “is a single correlation coefficient that is an estimate of the average of all the correlation coefficients of the items within a test. If alpha is high (0.80 or higher), then this suggests that all the items are reliable, and the entire test is internally consistent.”¹³ To this end, Cronbach’s alpha with no alterations for shut-up and comply rendered a score of 0.93 with N = 6. DeVellis asserts, however, that the last step in scale development is to maximize the scale length. Once the item reliability has been established, DeVellis posited that a researcher should spend time thinking about brevity, “when the researcher has ‘reliability to spare,’ it may be appropriate to buy a shorter scale at the price of a little less reliability.”¹⁴ As such, the item-total statistic matrix was inspected, and several items were recommended for removal. More specifically, it was found that the deleted αs were the same for two items—SU1 and SU2, which were removed—and the renewed Cronbach alpha for shut-up and comply became 0.88 with N = 4. Cronbach’s alpha with no alterations rendered a score of 0.86 with N = 4 for Shut-up and sabotage. While Cronbach’s alpha for shut-up and sabotage could be improved slightly, a decision was made not to remove an article so that both factors had four items.

Table 8.7 Regenerated shut-up pattern matrix

Table 8.8 Shut-up and comply scale

Table 8.9 Shut-up and sabotage scale

Data Analysis of the Speak-In Concept

SPSS version 25 was employed to perform an analysis of the speak-in concept. Specifically, Pearson correlation was applied to items SI13, SI14, SI15, SI16, SI17, SI18, SI19, SI20, SI21, SI22, SI23 and SI24 with the intent to measure the degree of linear relationship between two or more variables.

As depicted in Table 8.10, this process revealed that there was evidence of a positive relationship between the variables at the 0.01 level (two-tailed). Hair et al. contend the correlated variables suggest that the direct oblique rotation solution is appropriate for exploratory factor analysis in such a case. Moreover, the literature suggests that items that load at 0.40 or above are acceptable in factor analysis.¹⁵ To this end, loadings that fail under this threshold were suppressed.

Table 8.10 Speak-in correlation matrix

The KMO test and the Bartlett test of sphericity were conducted. The KMO returned a value of 0.89 and the p-value registered at 0.000 which suggest there was enough evidence to conduct a factor analysis.

A principle component analysis was conducted on items SI13–SI24. O’Rourke and Hatcher posited that the best method to understand oblique rotation is to, “always review the pattern matrix to determine which groups of variables are measuring a given factor, for purposes of interpreting the meaning of that factor.”¹⁶ To this end, a pattern matrix was generated and two factors for the speak-in concept identified.

As depicted in Table 8.11, component 1 factored items SI13, SI14, SI15, S16, SI17, SI18, SI19 and SI20 that were labeled speak-in with a parable . Component 2 was comprised of items SI21, SI22, SI23 and SI24 that were labeled speak-in on principles. A reliability analysis was conducted on speak-in with a parable that produced a Cronbach’s alpha score of 0.90 with N = 8. DeVellis again asserts that “when the researcher has ‘reliability to spare,’ it may be appropriate to buy a shorter scale at the price of a little less reliability.”¹⁷ As such, the item-total statistic matrix was inspected, and several items were recommended for removal. A decision was made to remove those items with the lowest αs. Hence, items SI16, SI17, SI19, and SI20 were removed and the renewed Cronbach’s alpha for speak-in with a parable decreased to 0.81 with N = 4. Cronbach’s alpha rendered a score of 0.82 with N = 4 for speak-in on principles. An examination of the item-total statistic matrix showed that the removal of additional items would not improve α for speak-in on principles.

Table 8.11 Speak-in pattern matrix

Table 8.12 Speak-in with a parable scale

Table 8.13 Speak-in on principles

Data Analysis of the Speak-Out Concept

SPSS version 25 was employed to perform an analysis of the speak-out concept. Specifically, Pearson correlation was applied to items SO25, SO26, SO27, SO28, SO29, SO30, SO31, SO32, SO33, S034, SO35 and SO36 with the intent to measure the degree of linear relationship between two or more variables.

Table 8.14 showed there was evidence of a positive relationship between the variables at the 0.01 level (two-tailed). Hair et al. contend the correlated variables suggest that the direct oblique rotation solution is appropriate for exploratory factor analysis in such a case. Additionally, the literature suggests that items that load at 0.40 or above are acceptable in factor analysis.¹⁸ In a similar vein as the shut-up and speak-in concepts, loadings that failed under this threshold were suppressed.

Table 8.14 Speak-out correlation matrix

The KMO test and Bartlett test of sphericity were conducted. The KMO returned a value of 0.85 and the p-value registered at 0.000, which suggested there was enough evidence to conduct a factor analysis.

A principle component analysis was conducted on items SO25–SO36. O’Rourke and Hatcher posited that the best method to understand oblique rotation is to, “always review the pattern matrix to determine which groups of variables are measuring a given factor, for purposes of interpreting the meaning of that factor.”¹⁹ To this end, a pattern matrix was generated and two factors for the speak-out concept identified. Moreover, the analysis also identified cross-loadings on items SO29 and SO30. Hair et al. maintain that when a variable is found to have more than one significant loading, it becomes a candidate for deletion.²⁰ To this end, these items were deleted, and a component analysis was employed on the remaining ten items.

As depicted in Table 8.15, component 1 factored items SO31, SO32, SO33, SO34, SO35, and SO36 that were labeled speak-out nonviolently. A reliability analysis was conducted on speak-out nonviolently that produced a Cronbach’s alpha score of 0.82 with N = 6. An examination of the item-total statistic matrix showed that the removal of additional items would not improve α. However, DeVellis posited that a researcher should spend time thinking about brevity.²¹ To this end, an analysis of the item-total statistic matrix revealed that the removal of items with the lowest scores, SO31 and SO32, would not negatively impact α. Once Cronbach’s alpha was recalculated, the score remained at 0.82 with N = 4. Component 2 was comprised of items SO25, SO26, SO27 and SO28 that were labeled speak-out negatively. A reliability analysis was conducted on speak-out negatively that produced a Cronbach’s alpha score of 0.77 with N = 4. An inspection of the item-total statistic matrix showed that the removal of additional items would not improve α.

Table 8.15 Speak-out pattern matrix

Table 8.16 Speak-out nonviolently scale

Table 8.17 Speak-out negatively scale

Data Analysis of the Step-Down Concept

SPSS version 25 was employed to perform an analysis of the step-down concept. Specifically, Pearson correlation was applied to items SD37, SD38, SD39, SD40, SD41, SD42, SD43, SD44, SD45, SD46, and SD47 with the intent to measure the degree of linear relationship between two or more variables.

Table 8.18 showed that there was evidence of a positive relationship between the variables at the 0.01 level (two-tailed). Hair et al. contend the correlated variables suggest that the direct oblique rotation solution is appropriate for exploratory factor analysis in such a case. Moreover, the literature suggests that items that load at 0.40 or above are acceptable in factor analysis.²² In keeping with the other concepts, loadings that fail under this threshold were suppressed.

Table 8.18 Step-down correlation matrix

The KMO test and Bartlett test of sphericity were conducted. The KMO returned a value of 0.85 and the p-value registered at 0.000 which suggest there was enough evidence to conduct a factor analysis.

A principle component analysis was conducted on items SD37–SD47. O’Rourke and Hatcher posited that the best method to understand oblique rotation is to, “always review the pattern matrix to determine which groups of variables are measuring a given factor, for purposes of interpreting the meaning of that factor.”²³ To this end, a pattern matrix was generated, and two factors for the step-down concept were identified.

As depicted in Table 8.19, component 1 factored items SD41, SD42, SD43, SD44, SD45, SD46 and SD47 which were labeled step-down by resigning. A reliability analysis was conducted on step-down by resigning that produced a Cronbach’s alpha score of 0.91 with N = 7.

Table 8.19 Step-down pattern matrix

Table 8.20 Step-down by resigning scale

Table 8.21 Step-down by resisting scale

As alluded to before, DeVellis contends that “when the researcher has ‘reliability to spare,’ it may be appropriate to buy a shorter scale at the price of a little less reliability.”²⁴ Thus, the item-total statistic matrix was inspected, and several items were recommended for removal. A decision was made to remove the three items with the lowest α’s—SD42, SD44 and SD47. Once these items were removed, a Cronbach’s alpha for step-down by resigning regenerated a score of 0.79 with N = 4. Component 2 was comprised of items SD37, SD38, SD39 and SD40 that were labeled step-down by resisting. A reliability analysis was conducted on step-down by resisting that produced a Cronbach’s alpha score of 0.88 with N = 4. Although Cronbach’s alpha for step-down by resisting could be improved slightly, a decision was made not to remove an item for the sake of factor consistency.

Data Analysis of the Step-It-Up Concept

SPSS version 25 was employed to perform an analysis of the step-it-up concept. Specifically, Pearson correlation was applied to items SIU48, SIU49, and SIU50 with the intent to measure the degree of linear relationship between two or more variables.

Table 8.22 showed that there was evidence of a positive relationship between the variables at the 0.01 level (two-tailed). Hair et al. contend the correlated variables suggest that the direct oblique rotation solution is appropriate for exploratory factor analysis in such a case. Moreover, the literature suggests that items that load at 0.40 or above are acceptable in factor analysis.²⁵ In keeping with the other concepts, loadings that fail under this threshold were suppressed.

Table 8.22 Step-it-up correlation matrix

The KMO test and Bartlett test of sphericity were conducted. The KMO returned a value of 0.67 and the p-value registered at 0.000 which suggest there was enough evidence to conduct a factor analysis.

A principle component analysis was conducted on items SIU48–SIU50. O’Rourke and Hatcher posited that the best method to understand oblique rotation is to, “always review the pattern matrix to determine which groups of variables are measuring a given factor, for purposes of interpreting the meaning of that factor.”²⁶ To this end, a pattern matrix was generated, and two factors for the step-down concept were identified.

As depicted in Table 8.23, component 1 factored items SIU48 and SIU49 that were labeled step-it-up morally. Component 2 factored item SIU50 that were labeled step-it-out with reflective leadership. Step-it-up morally and step-it-out with reflective leadership were deemed empirically unsuitable for scale development because there were two items or fewer in the components.²⁷ Thus, no further analysis was warranted for the step-it-up concept.

Table 8.23 Step-it-up pattern matrix

Step 8: Optimize Scale Length

The last step in scale development according to DeVellis is to optimize the scale length. Once the item reliability has been established, DeVellis’ guidance that a researcher should spend time thinking about brevity was followed. Although shortness of the scales may potentially threaten reliability, it may also increase the probability of participation due to time constraints. This point may particularly resonate within today’s high-paced culture. Upon removal of the “bad” items as driven by statistical examination, the ensuing items upheld as sub-scales. The brevity of such instruments, to conclude this analysis, may be sufficiently tailored for a twenty-first-century organization that is constantly competing for time.

Discussion

The chief hope of this chapter was to understand if the concepts affiliated with boardroom boldness language could be developed into a scientific instrument. The findings of this study can potentially help decisions-makers do three things: (1) make better empirical choices; (2) better manage the ethical health of cultures; and (3) help decision-makers to understand the climate of followership better. Moreover, the empirical establishment of the eight sub-scales as outlined in Tables 8.24, 8.25, 8.26, 8.27, 8.28, 8.29, 8.30, and 8.31 can help to advance a reseacher’s understanding of an influencer’s propensity to follow unethical orders blindly or to utilize their moral imagination to stop king-think.

Table 8.24 Shut-up and comply scale

Table 8.25 Shut-up and sabotage scale

Table 8.26 Speak-in with a parable scale

Table 8.27 Speak-in on principles scale

Table 8.28 Speak-out negatively scale

Table 8.29 Speak-out nonviolently scale

Table 8.30 Step-down by resisting scale

Table 8.31 Step-down by resigning scale

Limitations

This aspect of the book had several limitations. First, the study did not collect demographic data about the participants’ levels of followership or their ethnic data. This ommision could have potentially skewed the data. In a similar vein, participants in the study were overwhelmingly female. A more balanced data collection could have provided a different outcome. Although the Cronbach’s alpha score for step-down by resigning was lower than the other sub-scales, at 0.79 with N = 4, the brevity of the scale may be worth the exchange. Moreover, this study could be improved by generating a larger and better quality of pool items for the step-it-up concept. This, coupled with the inclusion of a more purposeful demographic, could bring more empirical rigor to the study. As the construct of followership continues to develop, this section should not be viewed as an exhaustive attempt to explore the spiritual facet of leading upward, but as an initial attempt to understand and scientifically codify the phenomenon.