Essay: Multicultural Study of Biometric Privacy Concerns in a Fire Ground Accountability Crisis Response System

Abstract

Biometric technology is rapidly gaining popularity as an access control mechanism in the workplace. In some instances, systems relying on biometric technology for

access control have not been well received by employees. One potential reason for resistance may be perceived privacy issues associated with organizational collection and use

of biometric data. This research draws on previous organizational information handling and procedural fairness literature to frame and examine these underlying privacy issues.

Perceived accountability, perceived vulnerability, and distrust were distilled from the previous literature as the primary dimensions of employee privacy concerns related to

biometric technology. This study assesses the effects of these privacy concerns, how they vary based on the cultural influences of Anglos and Hispanics.
Fire ground accountability is a critical management objective in the firefighting domain. In multi-unit or multi-agency crisis response scenarios, the on-scene incident

commander tracks and accounts for each first responder. This research designed and deployed a new fire ground accountability system that tracked firefighters through biometric

logins to their assigned positions on the firefighting apparatus. An instrument measuring level of privacy concern on three underlying dimensions and demographic data was

developed, validated and administered in a quasi-experimental field study. A pre-test ‘ post-test survey methodology was employed to detect potential differences in privacy

concerns as familiarity with the system increased. The study shows that Anglo and Hispanic subjects frame privacy issues differently associated with use of biometric

technology in a fire ground accountability system. Finally, the study showed that some privacy concerns such as distrust and perceived vulnerability can be alleviated through

system use with changes in post-use privacy concerns moderated by ethnic affiliation.

Keywords: crisis response systems, biometrics, privacy, case study, ethnicity, human-computer interaction

1. Introduction

The use of biometric technology has grown substantially in recent years. In many instances, this technology has been well-received. One study reported that only eight

percent of users who were required to have a biometric embedded in their driver’s license thought the technology invaded their privacy (Jain et al. 2002). Despite the success

of biometrics in some contexts, anecdotal evidence from privacy organizations (Abernathy et al. 2006; EFF 2006; O’Donoghue 2001) suggests that use of biometrics in the

workplace results in significant employee privacy concerns. Additionally, case law shows that the deployment of biometrics in the workplace may lead to a number of undesirable

outcomes, including enticing employees to unionize (York and Carty 2006) or seek renegotiation of existing labor contracts (Kelly and Herbert 2004).
Employee privacy concerns may be rooted in the uncertainty of how biometric data may be used. For instance, it is well-known that carriers of certain genetic disorders

including Down’s syndrome can be detected from fingerprint patterns (Faundez-Zanuy 2005; Zhai and Qui 2010). Additionally, some forms of skin cancer can be detected through

vascular scans (Hay 2003). Using biometric data for such functions could allow systematic discrimination of people who carry certain genetic markers or diseases.
An employee’s trust in the organization to act ethically is also at issue. While organizations may claim they’ll only use biometric technology for system access, there

is little legal protection for employees if the organization decides to expand its usage of collected biometric data (Levinson 2009). While the stated purpose of a biometric

authentication system may be securing the organization’s information technology resources, unscrupulous organizations could surreptitiously use biometric data for purposes

unrelated to this espoused objective. Examples of potential surreptitious use include criminal history inquiries (Cole 2004) and immigration status checks (Bump 2008; Zureik

2004).
The specific privacy concerns associated with biometric technology in the workplace have not been fully explored and defined. This research identifies three core

dimensions of biometric privacy concern and assesses their impact on acceptance and use of a crisis response system. Improved knowledge of employee privacy concerns will help

organizations address those concerns and mitigate negative impacts on the acceptance of systems incorporating biometric technology.
The effect of cultural differences between nations has been studied extensively with most research building upon Hofstede’s (1980) definitions of cultural dimensions.

Generally, this body of research shows that cultural influences are an important predictor of behavior (Hofstede 2001; Oyserman et al. 2002; Triandis 1995). Within this body

of national-level research, there are several works suggesting that privacy concerns are intricately tied to cultural influences (Bellman et al. 2004; Hann et al. 2002). While

considerable effort has been aimed at national cultural differences, there is considerably less research studying cultural differences at other strata, particularly ethnic

groups within the same country. However, Hofstede (1994) states that culture, as he defines it, may also apply to such ethnic groups.
When studying cultural influences at an ethnic group level, rather than a national level, some researchers have advocated context-dependent methods (Markus et al.

1991; Triandis et al. 1998) or individual-level measurements (Gudykunst et al. 1996; Singelis 1994; Singelis et al. 1995). Studies adopting these ethnicity-centric approaches

are still somewhat limited in scope. There are no known studies examining variations in employee privacy concerns among different ethnic groups within the same country.

However, studies in other contexts such as Internet use further suggests that the level of privacy concerns may vary based on ethnic culture (Introna et al. 1999; O’Neil

2001). This research examines the influence of Anglo and Hispanic ethnic cultural differences on privacy concerns within the context of a biometrically-enabled fire ground

accountability system.

2. Literature Review
2.1 Dimensions of Information Privacy Concern in the Employment Context
The introduction of new technologies that are construed to impact privacy may lessen employee acceptance of new systems. Jones et al. (2007) found that there was

substantial uncertainty surrounding the privacy impact of digital authentication technologies. They further found that a third of respondents were concerned with the privacy

implications of fingerprint scans and that the level of concern was highly context dependent. A review of the extant literature on employee privacy reveals three primary

themes that provide insight on the issue of workplace privacy: 1) the scope of monitoring, 2) whether the system is procedurally just or fair, and 3) whether the users trust

the organization to use monitoring data only for its intended purpose. These themes describe the organizational practices that generate concern and are re-framed below in an

employee-centric context.
2.2 Employee accountability perceptions

Perceived accountability is the degree to which employees believe they will be held more accountable for their actions when they log in with a biometric sample than when

they log into a system via other means. Employees expect to provide limited personal information and accept some level of job performance monitoring in exchange for

appropriate compensation. The collection of routine personal information and performance-related monitoring is usually viewed as being relevant to the employers’ business

objectives. Previous studies have found that employees believe it is acceptable for management to monitor employees (Grant and Higgins 1991, Oz et al. 1999). However, other

studies have argued that employers need to look beyond their legal rights and consider the adverse effects on employee morale when considering monitoring efforts, especially

when monitoring personal conduct during break periods or activities outside the work environment (Friedman and Reed 2007, Tolchinsky et al. 1981). When organizational

monitoring activities are not directly related to performance, employees tend to perceive the monitoring as an invasion of privacy (Alder 2001; Alder et al. 2005; Alge 2001;

Ambrose et al. 1998).
While employees often perceive a right to privacy in email communications, employers may also assert a right to monitor these systems. In early email system deployments,

the privacy rights of employees with regard to employer-owned email systems were often unclear (Cappel 1995; Oz et al. 1999). More recently, legal rulings have clearly favored

the right of the employer to monitor email if they have reasonable business concerns for doing so (Friedman et al. 2007). However, Sipior and Ward (1995) assert that employees

are likely to perceive email monitoring as invasive. In terms of monitoring scope, the email example supports the notion that employees will not accept broad monitoring

activities even when legitimate business purposes are evident (Cappel 1995).
Although employees may perceive different types of electronic monitoring as invasive, the impact of this invasiveness is sparsely documented through empirical evidence.

Research has found that employees modify their behavior when they are being monitored (Stanton and Wiess 2000). For example, more skilled employees were found to produce more

work when they perceived they were being monitored while low skill employees were found to produce less work when being monitored (Aiello and Kolb 1995, Urbaczewski and Jessup

2002). Both more skilled and less skilled employees reported higher stress when being monitored, and lower job satisfaction (Aiello and Kolb 1995, Urbaczewski and Jessup

2002). Overall, the literature on electronic monitoring supports the notion that employees are concerned about systematic monitoring in the workplace and desire to retain some

level of anonymity (Grant and Higgins 1991, Oz et al. 1999). We posit that electronic monitoring leads to perceptions of increased accountability for an employee’s actions.

This perception, which may result in adverse outcomes for the employee, creates privacy concerns in the employment context.
2.3 Employee vulnerability perceptions

Perceived vulnerability is the degree to which employees believes their stored biometric sample is susceptible to both external threats and internal unauthorized access.

Stone et al. (1983) suggest that individuals are concerned with collection, storage, use and release practices of organizations. Specific areas of concern include

understanding what is being collected, the capability to impact that collection, the opportunity to consent to collection, and the physical or psychological intrusiveness of

the information collection procedures (Stone et al. 1990). Empirical evidence supports this contention through findings that perceptions of privacy invasions were correlated

with consequences after disclosure and level of control over information (Fusilier et al. 1980; Stone et al. 1983; Tolchinsky et al. 1981). Culnan (1993) noted similar

dimensions of acquisition, use and transfer of information as privacy concerns in direct marketing campaigns. There were different nuances to the concerns when considering

internal customers, external customers or prospective customers. Brandimarte et al. (2010) found that users with greater control over publication were more willing to disclose

private information, even when they knew they could not control subsequent access to the published data.
In an empirical study, Smith et al. (1996) found that collection, unauthorized use, errors, and improper access were significant concerns regarding organizational privacy

practices. They also identified reduced judgment (reliance on automated decision-making algorithms) and combining of data (integrating data collected for different purposes)

as secondary concerns related to organizational practices. Smith et al. (1996) also found significant correlations between privacy concerns and personality factors including

trust/distrust, paranoia, and social criticism. Stewart and Segars (2002) validated the Smith et al. (1996) model using confirmatory factor analysis. They suggested procedural

fairness (Culnan and Armstrong (1999), environmental control (Hoffman et al. (1999), and control over secondary use of information (Hoffman et al. 1999) as additional factors

potentially impacting concern for organizational privacy practices.
Procedural fairness is the perception that an individual is treated fairly by organizational processes or automated decision-making algorithms. Organizations that are

perceived to exercise procedural fairness build trust when they act on behalf of individuals who cannot control the organization’s behavior (Culnan and Armstrong 1999). The

establishment of a trusting relationship between the individual and the organization is a key element as it may induce the individual to provide greater levels of private

information (Hann et al. 2002; Josang et al. 2007; Liu et al. 2004; Liu et al. 2005; Walczuch et al. 2004).
A central tenant of procedural fairness is that the individual cannot act in his or her own best interest. Thus, the importance of procedural fairness in a given context

is dependent upon the level of environmental control (Hoffman et al. 1999). Environmental control is the ability of the individual to control the actions of others involved in

a transaction. It is primarily associated with commercial exchanges. In these exchanges, consumers who have a higher level of environmental control are less likely to be

concerned with procedural fairness. However, despite an individual’s ability to control the primary transaction environment, that person may still be concerned about

subsequent uses of the transaction information. Secondary use of information control describes the individual’s ability to influence further dissemination of information

related to a transaction (Eddy et al. 1999; Hoffman et al. 1999; Smith et al. 1996; Stone et al. 1990).
In summary, empirical models have identified the collection of excessive information, unauthorized aggregation of data, lack of control, unauthorized use, data errors,

reduced human involvement in decision making, and the fairness of the organization controlling the data as factors impacting privacy perceptions in various contexts. However,

employees often lack insight into an organization’s specific activities related to these practices and thus the effect may be better captured with a more general sense of

perceived vulnerability to undesirable information handling practices.
2.4 Employee distrust

Perceived distrust refers to the extent to which employees are concerned that the biometric data they provide will be used for purposes other than system

authentication. Such unintended purposes include using a piece of biometric data to gain additional personal data about the subject (Woodward, 1997). As an example, genetic

markers indicating the propensity for disease may be gleaned from some biometric samples (Prabhakar et al. 2003). Other possible scenario includes using biometric samples to

link data in ways not previously anticipated (Woodward, 1997). For example, an organization could use access door biometric login records in conjunction with soft-biometric

camera data such as clothing characteristics or hair style to re-identify an otherwise anonymous person walking through publicly accessible areas of the organization (Bazzani

et al. 2013, Schuman et al. 2014). We posit that these concerns will be captured as a manifestation of the employee’s distrust or trust in the organization.
Although trust / distrust is extensively explored in the literature, its relationship to privacy remains a topic of debate. Most authors choose to work with either

trust / distrust or privacy but not both (Gefen et al. 2003; Hann et al. 2007; Hann et al. 2002; Hui et al. 2007; Jarvenpaa et al. 2000; Josang et al. 2007; Lee et al. 2001;

McKnight et al. 2002; Xu et al. 2008). Some authors characterize privacy as an antecedent for trust (Liu et al. 2004; Liu et al. 2005). Other researchers have reframed this

relationship to focus on distrust (rather than trust) as a dimension of privacy concern. Westin (2003) notes a strong correlation between distrust and privacy concerns in

public surveys measuring trust in both business and technology. Further, he reports that thirty percent of employees are concerned about employer handling of personnel data.

While the literature shows a strong correlation between privacy and trust / distrust, the complexity of the models prevents the authors from drawing reliable conclusions about

any causal nature of the relationship. Other authors prefer to model both privacy and trust / distrust as components of a ‘privacy calculus’ determination (Dinev et al. 2006;

Dwyer et al. 2007; Malhotra et al. 2004). In these studies, the authors find strong statistical evidence that privacy concerns and trust / distrust perceptions combine to form

an individual’s willingness to share private information. The authors found that not all privacy concerns can be overcome by trust in the organization, nor can distrust

account for all privacy concerns. The conceptualization of these models places trust / distrust as a component of overall privacy concerns. Similar to Westin (2003), we adopt

distrust as a form of organizational privacy concern in our study.

2.5 The Impact of Ethnic Cultural Values on Privacy

Social psychology researchers often associate the desire for personal privacy with western, predominantly individualistic cultures (Hann et al. 2002; Markus et al. 1991;

Milberg et al. 1995; Oyserman et al. 2002; Rhee et al. 1995). These individualistic cultures are characterized as emphasizing status (Schwartz 1990), the attainment of

personal goals (Kim 1994) and maintaining individual control over personal matters (Hui et al. 1994; Triandis 1995). In contrast to these independent cultures, collectivist or

interdependent cultures are often portrayed as less concerned with personal interests and more focused on meeting the needs of family or other referent groups (Hui 1988;

Oyserman et al. 2002; Triandis 1995). Asian, African, and Latin American cultures are generally characterized as collectivist-oriented (Gaines Jr. et al. 1997; Hofstede 1980;

Marin et al. 1985; Triandis et al. 1986). Members of collectivist cultures are expected to share their private matters with other members of the group (Triandis 1995). In many

cases the collective is viewed as having a fundamental right to know a member’s private thoughts and dictate how that member should think (Ho et al. 1994).
A relatively small number of studies have empirically examined the cultural differences between Anglo-American and Latin American or Hispanic subjects. Shkodriani and

Gibbons (1995) reported that Mexican university students exhibited lower levels of collectivism than American university students. This study used an older collectivism

instrument designed by Hui (1988) which has frequently produced results at odds with those produced by other instruments. In an unpublished dissertation discussing the

development and validation of a vertical individualism and collectivism instrument Robert (1998) found American subjects significantly more individualistic and significantly

less collectivist than Mexican subjects. Oishi (2000) found that Brazilian, Portuguese, and Puerto Rican subjects had significantly higher collectivist tendencies than Anglo

populations in a study of the effects of goals as a prerequisite to personal well-being.
In summary, cultural influences are widely accepted as factors that influence both thought and behavior. At a national level, differences in privacy concerns have been

observed between collectivist and individualistic cultures. These cultural differences may impact the level and types of privacy concerns expressed a particular context. A

summary of privacy construct definitions is provided in Table 1.

Table 1. Definition of Privacy Constructs

Construct
Operational Definition
Perceived Accountability
The degree to which employees believe they will be held more accountable for their actions when they log in with a

biometric sample than when they log into a system via other means.

Perceived Vulnerability
The degree to which employees believe their stored biometric sample is susceptible to both external threats and internal unauthorized access.

Perceived Distrust
The extent to which employees believe the entity storing their biometric sample will use it for purposes other than system authentication.

2.6 Hypotheses Development
This study examines privacy concerns related to biometric technology in the workplace. The privacy concerns were modeled as three constructs designed to capture user

perceptions of privacy issues through the lens of a fire ground accountability system. The perceived accountability construct is the user’s perception of the covert

recognition / electronic monitoring threat and is defined as the notion that users will be more accountable for their actions when they log in with a biometric sample than

when they log into a system via other means. This is largely due to the user’s belief that a biometric authentication is difficult to refute. The number of alternative

plausible explanations for a disputed log in is substantially lower when biometrics are employed and thus the users may feel that they are being held to a higher level of

accountability. The perceived vulnerability construct is based on concerns about data aggregation, loss of control and unintended disclosure. It also describes the users’

feelings that the entity storing their biometric data may not be completely secure. Regardless of the users’ trust / distrust in the entity, they may question the entity’s

ability to adequately protect the stored biometric samples from unintentional disclosure or compromise.
Both perceived accountability and perceived vulnerability assume implicit trust in the organization to act with due diligence and benevolence toward the individual. A

third and separate privacy concern relates to the users’ distrust of the organization. Distrust related to organizational use of biometric technology is drawn from the

concepts of unintended functional scope and unintended application scope (Woodward 1997) along with concerns for misuse of information. If the users believe that the

organization storing their biometric templates will use them for undisclosed and unanticipated purposes other than system authentication, they are likely to be concerned about

the impact of those other potential uses. The distrust construct captures this concern and is better suited to empirical testing than the unintended functional scope and

unintended application scope constructs.
While familiarity and the opportunity to use new technologies may alleviate some privacy concerns may improve over time, there is little evidence to suggest that the

level of privacy concerns among Anglos and Hispanics would converge. Thus, the literature suggests that Anglos and Hispanics will differ in each of the three privacy

dimensions, both before having opportunity to use the system and after having such an opportunity. There are no known studies that examined differential changes in privacy

concerns among ethnic groups following exposure to a technology or system. Consequently, the limited guidance in the literature suggests that ethnicity will continue to have a

significant impact on privacy concerns after use. Thus, the hypotheses are:

H1: Prior to use of the system, Anglos will have higher perceived accountability concerns than Hispanics toward using the biometric technology.
H2: The use of biometric technology will reduce perceived accountability concerns of both Anglos and Hispanics, but Hispanics will retain higher levels of concern than Anglos.
H2a: The use of biometric technology will reduce the perceived accountability concerns of Hispanics.
H2b: The use of biometric technology will reduce the perceived accountability concerns of Anglos.
H2c: After use, Hispanics have greater perceived accountability concerns toward using biometric technology than Anglos.
H3: Prior to use of the system, Anglos will have higher perceived vulnerability concerns than Hispanics toward using the biometric technology.
H4: The use of biometric technology will reduce perceived vulnerability concerns of both Anglos and Hispanics, but Hispanics will retain higher levels of concern than Anglos.
H4a: The use of biometric technology will reduce the perceived vulnerability concerns of Hispanics.
H4b: The use of biometric technology will reduce the perceived vulnerability concerns of Anglos.
H4c: After use, Hispanics have greater perceived vulnerability concerns toward using biometric technology than Anglos.
H5: Prior to use of the system, Anglos will have higher distrust concerns than Hispanics toward using the biometric technology.
H6: The use of biometric technology will reduce distrust concerns of both Anglos and Hispanics, but Hispanics will retain higher levels of concern than Anglos.
H6a: The use of biometric technology will reduce the distrust concerns of Hispanics.
H6b: The use of biometric technology will reduce the distrust concerns of Anglos.
H6c: After use, Hispanics have greater distrust concerns toward using biometric technology than Anglos.
3. Methodology

This study involved the deployment of an experimental fire ground accountability system that tracked fire fighter work assignments through biometric logons to a specific

duty position. Biometric technology was an option under consideration for improvement of fire fighter safety by providing positive identification of those who responded to

incidents. Fire ground accountability is a significant and well-known occupational safety issue in the firefighting domain.
The study was conducted using a quasi-experimental pretest-posttest design. The operational environment made it impossible to segregate the subjects into appropriate

treatment and control groups without impacting the subjects’ primary duties. Therefore, a quasi-experimental design without a control group was used. This type of quasi-

experimental design is often most appropriate when it is advantageous to observe the phenomena of interest in a natural setting. A major objective of the study was to assess

fire fighter privacy concerns associated with workplace use of biometric technology. To ensure subjects associated their true workplace privacy concerns with the biometric

technology it was desirable to deploy the treatment system in the normal work environment. Additionally, the pretest-posttest design is more robust than quasi-experiments

utilizing only posttest measures of treatment effects. Comparison between pretest and posttest scores supports the argument that the observed changes are directly attributable

to the treatment. To alleviate validity concerns, Cook and Campbell (1979) advise that the single-group pretest-posttest design should only be used when the nature of the

outcome variable makes it unlikely that the observed results will be attributable to influences other than the treatment and when the pretest-posttest time interval is

relatively short. Our experiment was designed with these recommendations in mind and there was no evidence of external events that were likely to impact results during the

study period.
Subjects for this research were firefighters employed by a large municipality in the southwestern United States. Ethnic diversity of the organization facilitated the goal

of studying differences in ethnic cultures within an otherwise homogeneous group. Hofstede (2001) recommends selecting subjects from the same profession or occupation,

socio/economic strata, organization, education level, age, etc. when studying cultural issues. The proposed sample appeared well-suited in this regard. All of the subjects

were from the same profession and were employed by the same organization. The distribution of pay grades, age, and education levels was similar among the ethnic groups of

interest.
The treatment venue was a single fire station within the municipality. Infrastructure modifications required for the pilot system limited the study to a single site. That

site was selected based on the largest number of personnel assigned and the favorable distribution of Hispanic and Anglo subjects. There were approximately 54 personnel

assigned to the research site with approximately half belonging to each ethnic group of interest. Participants had their privacy concerns toward using biometric technology

measured when the system was initially deployed. They were surveyed again after approximately 45 days of system use to see if their concerns had changed.
Although participation in the study was voluntary, the participation rate was expected to be high. A power analysis based on Cohen’s (1992) work suggested that a sample

size of approximately 21 subjects per ethnic group would be sufficient to detect large effects with power = 0.7. Based on the maximum pool of 54 subjects, there was a

significant risk that the number of participants would be insufficient to detect smaller differences and thus the focus of the study was the detection of large effects.
3.1 Instrument development and validation
A new instrument was developed to collect information on privacy concerns associated with using a biometric system, and respondent demographic information. During the

pre-test survey subjects were asked to select a unique identifier that they would be able to remember later. The survey suggested that the subjects use their fire fighter

badge number but they were allowed to choose an alternate identifier if they desired to do so. This unique identifier was used to match pre-test and post-test survey responses

for each participant. The relevant data collected in the pre-test and post-test surveys was identical except that demographic data was not collected in the post-test

instrument.
There were no pre-existing items available to measure the biometric privacy concern dimensions of interest. A pool of at least twelve new items was developed based on the

operational definition for each construct as detailed in Table 1. The goal was to produce scales of approximately four to six items for each privacy dimension prior to

reliability testing. The new items were then presented to a focus group for evaluation. The focus group consisted of experts in biometric systems, privacy issues, information

systems research, firefighting, and instrument development. The focus group was asked to evaluate items based on suitability to measure the construct of interest,

appropriateness for the target audience, and any potential item ambiguity related to other privacy constructs. The focus group also provided suggestions related to item

clarity, length, wording, and redundancy. Items identified by any focus group member as ambiguous, not clearly related to the construct of interest, or otherwise unsuitable

were rejected. All other items were retained for validity testing. Additional new items were created as necessary to maintain an acceptable scale length. Some items in each

scale were intentionally developed as reverse-coded items for the purpose of validity checks. After several rounds of revisions, a sufficient number of quality items had been

developed for each privacy construct and the focus group agreed that the items were ready for pilot testing. The final instrument selected for pilot testing had six items

measuring Perceived Distrust, five items measuring Perceived Accountability, and four items measuring Perceived Vulnerability. Items for each scale were rated on a 7-point

Likert scale with responses for each statement ranging from 1 (Strongly Disagree) to 7 (Strongly Agree).
In order to receive meaningful results for validation testing it was necessary to prime subjects so they would answer as if they were subject to using biometric technology

in the workplace. To do this, each fire fighter was sent an email announcing an upcoming pilot project that would evaluate using a biometric system to keep track of who was on

duty. The email was sent by a senior department official and indicated management’s support of an upcoming pilot project designed to improve fire fighter tracking through

biometric technology. Fire fighters were encouraged to complete a web-based survey to provide anonymous feedback on privacy concerns associated with the system.
The instrument was validated using firefighters assigned to stations other than the designated treatment venue. As with the target research sample pool, the validation

sample pool was divided nearly equally among Hispanic and Anglo subjects. Fire fighters within the organization work a 24-hour shift. There are three shifts that rotate on a

daily basis. To ensure all potential participants had equal opportunity to complete the survey, it was offered to each shift an equal number of times. After two complete

cycles of the three shifts, more than 300 responses had been received. It was determined that sufficient validation data had been collected and access to the validation

instrument was withdrawn.
The post-collection analysis revealed that 303 responses were received, with 15 declining to take the survey after reading the instructions, and 27 incomplete responses.

The high number of incomplete responses can be attributed, in part, to the operational environment. Several fire fighters provided feedback stating that they started the

survey but were forced to abandon it when an emergency call was received. These fire fighters were instructed to take the survey again if they desired to do so. After removing

declined / incomplete responses, a total of 261 usable responses were obtained. The number of fire fighters employed by the municipality varies daily, but was approximately

1,200 at the time of the survey. Thus, the overall response rate was approximately 25.2 percent. A 20 percent response rate is average for surveys where the user is given a

single notification (Kaplowitz et al. 2004). Given the short period of time the survey was available, the single email notification, and the lack of any incentive for

completion, the response rate for this survey was somewhat higher than expected. Response demographic percentages were in line with the organization’s actual workforce

structure, indicating that a non-response bias based on demographic factors was unlikely.
The validation data was subjected to an exploratory factor analysis using principle component analysis with varimax rotation and Kaiser Normalization to assess convergent

validity. Convergent validity examines the degree to which multiple measures of a construct are similar (Kerlinger et al. 2000). Varimax rotation is an orthogonal scheme that

adjusts the coordinates to maximize squared variance sums. It is often a preferred rotation method for examining convergent validity because each factor receives equal weight

in determining the amount of rotation (Lattin et al. 2003).
The factor analysis identified the five expected factors with eigenvalues greater than 1. The rotated component matrix is provided in Table 2.

Table 2: Validation Sample Factor Loadings

Component

3
4
5
PV-1-R

.546
PV-2-R

.605
PV-3-R

.640
PV-4

.628
PA-1

.645

PA-2

.767

PA-3

.711

PA-4

.743

PA-5-R

.756

PT-1-R
.636

PT-2
.688

PT-3
.776

PT-4-R
.734

PT-5-R
.662

PT-6
.776

A Cronbach’s coefficient alpha of greater than .7 is generally considered acceptable for internal consistency (Crano et al. 2002). Scale reliabilities for the constructs

of interest are Perceived Vulnerability = .741, Perceived Accountability = .873, and Perceived Distrust = .834, thus all scale reliabilities are acceptable and the result

indicated a relatively parsimonious instrument that adequately assessed the desired constructs. Overall, validation testing indicated that the instrument was suitable to

measure the constructs of interest. The instrument exhibited good psychometric properties related to convergent validity and scale reliability. Completion percentages

indicated that the survey length was acceptable to potential respondents and that there was reasonable likelihood that sufficient responses would be obtained from the

treatment group.
3.2 The Research Treatment
The treatment for this experiment was the introduction of a custom-designed, biometrically-enabled fire ground accountability application at the research site. The

application was developed using business rules provided by senior fire department management. Chief among these rules were the requirement for positive log on and log off as

well as the ability for the duty officer to override assignments. For the purpose of this experiment, additional rules such as the rank requirements to be an apparatus (fire

truck) commander and special qualifications for certain apparatus were not enforced. The application was installed on existing computers within the fire station and on

firefighting apparatus assigned to the research site. The sole stated purpose of the application was to track the personnel assigned to each position at any point in time for

fire ground accountability.
The application was designed to visually depict the fire fighter’s job assignment paradigm. It allowed users to select from graphical representations of the various

firefighting apparatus assigned to the station. Once the fire fighters selected an apparatus, they could choose any available position on the apparatus. If the position was

unoccupied, the fire fighters could enter their badge number and provide a biometric finger pattern sample. If the finger pattern sample matched the stored template for that

badge number, the fire fighter was assigned to that position. Successful assignment to a position was indicated with the fire fighter’s name shown in the position on the

apparatus with a green background. If a log on was successful but the position was already occupied, the log on would show as a red ‘pending logon’ until the previous occupant

of the position logged out. Usage was generally twice daily as the fire fighters reported to duty and completed their shift. In some instances, where fire fighters had

appointments, used personal time off, or swapped shifts with other workers, a particular fire fighter may have had several log-ins / log-outs in a single day. The duty officer

could override the log out requirement as necessary.

3.3 Data Collection
The research data was collected in two phases. The administration of pre-test surveys was scheduled to coincide with the introduction of the treatment system at the

research site. Each fire fighter at the research site was sent an email from senior management asking them to complete the pre-test survey and participate in the experiment.

This email clearly informed the fire fighters that their participation in both the survey and the experiment was voluntary. It further encouraged each fire fighter to use the

system on a daily basis. Finally, fire fighters were asked to complete the pre-test survey prior to their enrollment and first use of the system.
Each fire fighter that chose to participate in the experiment had the option to use either a biometric finger pattern sample or a traditional password to log on to the

system. A researcher was available each day during the first two weeks of the experiment to assist fire fighters with the biometric enrollment process. This is similar to

industry recommended ‘best practices’ where a technical representative is normally available to assist new biometric system users during the enrollment process. The researcher

specifically refrained from further endorsing the system, encouraging survey participation, or answering privacy-related questions.
The post-test survey was administered approximately 45 days after the introduction of the system. As with the pre-test survey, subjects were given six days to complete the

post-test survey. 53 fire fighters were assigned to the experiment site. 46 fire fighters completed both the pre-test and post-test surveys. The usable response rate for the

matched set was 86.8 percent. The distribution between Anglo and Hispanic subjects was relatively equal and representative of organizational demographics. There were 24

Hispanic respondents and 22 Anglo respondents. The sample size is sufficient to detect large effects between the ethnic groups.

4. Analysis, Results, and Discussion

4.1 Pre-test / Post-test Examination of Ethnicity Effects on Perceived Accountability

The main effect of ethnicity on perceived accountability was significant at p = .092 as shown in Table 5, indicating a notable overall difference in perceived

accountability concerns at T1 that should be assessed by examining the means in Table 4. H1 is assessed by comparing the T1 means between ethnic groups. The T1 mean of

Hispanics (4.692) is greater than the T1 mean for Anglos (4.245) and falls outside of the 95% confidence interval (3.939-4.552) for Anglos. Although the difference between the

means of the two ethnic group is significant, the difference is not in the direction hypothesized, thus, H1 is not supported.
H2a, H2b, and H2c are assessed by comparing the T1 and T2 means as indicated by each of the hypotheses. The effect of time on perceived accountability is not significant

(Table 3), thus H2a and H2b are not supported. H2c is assessed by comparing the T2 means of Anglos and Hispanics as shown in Table 4. The T2 mean of Hispanics (4.654) is

greater than the T2 mean for Anglos (4.332) but falls within the 95% confidence interval (3.960-4.704) for Anglos. Thus, Anglos and Hispanics are not significantly different

at T2 and H2c is not supported.
In summary for H1and H2, although there were significant differences in perceived accountability concerns between Anglos and Hispanics at T1, the groups were not

statistically different following the treatment. Neither group was significantly different from their T1 mean at T2. These effects are visually depicted in Figure 2, which

shows relatively flat lines for each ethnic group from T1 to T2, a significant gap between Anglos and Hispanics at T1, and a smaller, statistically insignificant gap between

the groups at T2.

Table 3: Pre-test / Post-test Differences in Perceived Accountability Concerns
Effect
Value
F
Hypothesis df
Error df
Sig.
TIME
Pillai’s Trace
.004
.194a
1.000
46.000
.662

Wilks’ Lambda
.996
.194a
1.000
46.000
.662

Hotelling’s Trace
.004
.194a
1.000
46.000
.662

Roy’s Largest Root
.

004
.194a
1.000
46.000
.662
TIME * ETHNICITY
Pillai’s Trace
.028
1.317a
1.000
46.000
.257

Wilks’ Lambda
.972
1.317a
1.000
46.000
.257

Hotelling’s Trace
.029
1.317a
1.000
46.000
.257

Roy’s Largest

Root
.029
1.317a
1.000
46.000
.257
a. Exact statistic

b. Computed using alpha = .05

c. Design: Intercept + ETH1
Within Subjects Design: TIME

Table 4: Pre-test / Post-test Perceived Accountability Means
ETHNICITY * TIME
Measure: Perceived Accountability
ETHNICITY

TIME
Mean
Std. Error
95% Confidence Interval

Lower Bound
Upper Bound
Anglo

1
4.245
.152
3.939
4.552

2
4.332
.185
3.960
4.704
Hispanic
1
4

.692
.140
4.410
4.974

2
4.654
.170
4.312
4.996Table 5: Pre-test / Post-test Between Subjects Perceived Accountability Effects
Measure: Perceived Accountability
Transformed Variable: Average
Source
Type III Sum of Squares
df
Mean Square
F
Sig.
Noncent. Parameter
Observed Powera
Intercept
957.055
1
957.055
1606.687
.000
1606.687
1.000
ETH1
1.761
1
1.761

2.957
.092
2.957
.391
Error
27.401
46
.596

a. Computed using alpha = .05
Figure 2: Perceived Accountability Means Plot
4.2 Pre-test / Post-test Examination of Ethnicity Effects on Perceived Vulnerability
The main effect of ethnicity on perceived vulnerability was not significant (p = .282) as shown in Table 8, indicating statistically negligible overall differences in the

perceived vulnerability concerns of Anglos and Hispanics at T1. Thus, H3 is not supported.
H4a and H4b are assessed by comparing the T1 and T2 means for each ethnic group. The main effect of time is significant in Table 6 (p = .009), thus an examination of the

means is necessary to evaluate the specific hypotheses. H4a is assessed by comparing the T1 and T2 means of Hispanics. The T2 mean (4.096) falls within the 95% confidence

interval for the T1 mean (3.811-4.419) therefore H4a is not supported. H4b is assessed by comparing the T1 and T2 means of Anglos. The T2 mean (3.500) falls outside the 95%

confidence interval for the T1 mean (3.897-4.558) therefore H4b is supported, indicating that the treatment significantly reduced the perceived vulnerability concerns of

Anglos. H4c is assessed by comparing the T2 means for Hispanics and Anglos. The T2 mean for Hispanics (4.096) is greater than the T2 mean for Anglos (3.500), and falls outside

the T2 95% confidence interval for Anglos (3.063-3.937). Thus, although the two ethnic groups are significantly different at T2, the difference is not in the direction

hypothesized and H4c is not supported. Hispanics have higher perceived vulnerability concerns than Anglos after using the system.
Although not hypothesized, the main effect of time and ethnicity indicates a significant interaction effect (p = .013) in Table 6. Thus the treatment affects the ethnic

groups differently.
In summary for H3 and H4, Anglos and Hispanics were not significantly different in perceived vulnerability concerns at T1. Although the main effect of ethnicity was not

significant, this finding was likely confounded by an interaction between the treatment and ethnicity. A thorough investigation of the means revealed important insights into

the constructs of interest. The treatment substantially reduced the perceived vulnerability concerns of Anglos while the perceived vulnerability concerns of Hispanics remained

relatively stable. Thus, at T2, the level of vulnerability concerns for Anglos diminished and became significantly less than those of Hispanics.

Table 6: Pre-test / Post-test Differences in Perceived Vulnerability Concerns
Effect
Value
F
Hypothesis df
Error df
Sig.
TIME
Pillai’s Trace
.138
7.365a
1.000
46.000
.009

Wilks’ Lambda
.862
7.365a
1.000
46.000
.009

Hotelling’s Trace
.160
7.365a
1.000
46.000
.009

Roy’s Largest

Root
.160
7.365a
1.000
46.000
.009
TIME * ETHNICITY
Pillai’s Trace
.126
6.625a
1.000
46.000
.013

Wilks’ Lambda
.874
6.625a
1.000
46.000
.013

Hotelling’s Trace
.144
6.625a
1.000
46.000
.013

Roy’s

Largest Root
.144
6.625a
1.000
46.000
.013
a. Exact statistic

b. Computed using alpha = .05

c. Design: Intercept + ETH1
Within Subjects Design: TIME

Table 7: Pre-test / Post-test Perceived Vulnerability Means
ETHNICITY * TIME
Measure: Perceived Vulnerability
ETHNICITY

TIME
Mean
Std. Error
95% Confidence Interval

Lower Bound
Upper Bound
Anglo

1
4.227
.164
3.897
4.558

2
3.500
.217
3.063
3.937
Hispanic
1

4.115
.151
3.811
4.419

2
4.096
.200
3.694
4.498Table 8: Pre-test / Post-test Between Subjects Perceived Vulnerability Effects
Measure: Perceived Vulnerability
Transformed Variable: Average
Source
Type III Sum of Squares
df
Mean Square
F
Sig.
Noncent. Parameter
Observed Powera
Intercept
756.844
1
756.844
1282.608
.000
1282.608
1.000
ETH1
.699
1
.699

1.184
.282
1.184
.187
Error
27.144
46
.590

a. Computed using alpha = .05
Figure 3: Perceived Vulnerability Means Plot
4.3 Pre-test / Post-test Examination of Ethnicity Effects on Distrust
The main effect of ethnicity on distrust concerns was not significant at p = .181 as shown in Table 11, indicating statistically negligible overall differences in the

distrust concerns of Anglos and Hispanics at T1. Thus, H5 is not supported.
H6a and H6b are assessed by comparing the T1 and T2 means as indicated by each of the hypotheses. The effect of time on perceived distrust indicates a significant main

effect (Table 9), thus an examination of the means in Table 10 is warranted. H6a is assessed by comparing the T1 and T2 means of Hispanics. The T2 mean (4.127) falls within

the 95% confidence interval for the T1 mean (3.747-4.676) therefore H6a is not supported. H6b is assessed by comparing the T1 and T2 means of Anglos. The T2 mean (4.561) falls

within the 95% confidence interval for the T1 mean (4.184-5.195) therefore H6b is not supported.
H6c is assessed by comparing the T2 means for Hispanics and Anglos. The T2 mean for Hispanics (4.127) is less than the T2 mean for Anglos (4.561) and falls within the 95%

confidence interval for Anglos at T2 (4.064-5.058). Thus, H6c is not supported.
In summary for H5 and H6, Anglos and Hispanics were not significantly different in distrust concerns either before or after treatment. Figure 4 shows modest reductions in

the distrust concerns of both groups following the treatment. When both ethnic groups were combined there was a significant main effect of the treatment in reducing distrust

concerns. Thus system use alleviated distrust concerns of study participants as a whole. However, this effect was not statistically significant for either ethnic group alone.

Additionally, there was no significant difference between the ethnic groups after treatment.

Table 9: Pre-test / Post-test Differences in Distrust Concerns
Effect
Value
F
Hypothesis df
Error df
Sig.
TIME
Pillai’s Trace
.109
5.645a
1.000
46.000
.022

Wilks’ Lambda
.891
5.645a
1.000
46.000
.022

Hotelling’s Trace
.123
5.645a
1.000
46.000
.022

Roy’s Largest

Root
.123
5.645a
1.000
46.000
.022
TIME * ETHNICITY
Pillai’s Trace
.005
.244a
1.000
46.000
.623

Wilks’ Lambda
.995
.244a
1.000
46.000
.623

Hotelling’s Trace
.005
.244a
1.000
46.000
.623

Roy’s Largest

Root
.005
.244a
1.000
46.000
.623
a. Exact statistic

b. Computed using alpha = .05

c. Design: Intercept + ETH1
Within Subjects Design: TIME

Table 10: Pre-test / Post-test Distrust Means
ETHNICITY * TIME
Measure: Distrust
ETHNICITY

TIME
Mean
Std. Error
95% Confidence Interval

Lower Bound
Upper Bound
Anglo

1
4.689
.251
4.184
5.195

2
4.561
.247
4.064
5.058
Hispanic
1
4.212
.231
3.747

4.676

2
4.127
.227
3.670
4.585
Table 11: Pre-test / Post-test Between Subjects Distrust Effects
Measure: Distrust
Transformed Variable: Average
Source
Type III Sum of Squares
df
Mean Square
F
Sig.
Noncent. Parameter
Observed Powera
Intercept
921.689
1
921.689
687.425
.000
687.425
1.000
ETH1
2.475
1
2.475

1.846
.181
1.846
.265
Error
61.676
46
1.341

a. Computed using alpha = .05

Figure 4: Distrust Means Plot
5. Conclusion

This study examined privacy concerns associated with the use of biometric technology in the workplace and how those privacy concerns vary based on ethnic affiliation. The

study showed that some, but not all, privacy concerns were alleviated by allowing the subjects to use the biometric technology. In particular, distrust concerns were

significantly reduced for all participants after using the biometric technology and perceived vulnerability concerns for Anglo subjects were also diminished. The large change

in perceived vulnerability scores for Anglos may be attributable to system familiarity. The initial concerns of Anglos may have been alleviated once they were able to see how

the actual system worked. The high level of initial concerns may have been primarily due to unfamiliarity with the system. By contrast, the concerns of Hispanics appear to be

more enduring and may represent a systemic fear of technological security weaknesses.
The study extracted three new constructs for privacy concerns related to biometric technology from the extensive literature on organizational privacy, trust, user

resistance, and technology adoption. Perceived accountability, perceived vulnerability, and distrust were determined to reflect the primary concerns associated with this

emerging technology. A new instrument with scales for the biometric privacy concern dimensions was developed. It was refined with the aid of a focus group, subjected to pilot

testing, and ultimately validated through factor analysis and scale reliability measures. The successfully validated scale provides a new resource for researchers in the

privacy domain.
Although only one of nine hypotheses was supported, the study provides significant insights into the privacy concerns of different ethnic groups. Previous research

suggested that privacy concerns might vary based on ethnic affiliation. Contrary to previous literature that suggests predominantly independent cultures will have more privacy

concerns than predominantly collectivist cultures, the study showed that the reverse was often true with regard to specific privacy dimensions. One potential explanation for

this contradictory finding may lie in this study’s examination of ethnic cultural groups within a single country. The vast majority of previous research that informed the

hypotheses development relied on national level measures of culture. This study provided strong empirical support for the notion that different ethnic groups within a country

may share some privacy concerns but differ in other privacy dimensions. Finally, the study shows that some privacy concerns can be alleviated through use of the technology but

not all ethnic groups react the same way after exposure. Future studies should consider the implications of this finding during the design phase.
Our results suggest several factors that should be taken into account by system designers, analysts, and integrators. Practitioners would be wise to consider the three

privacy sub-dimensions delineated here and carefully assess their impact on the various ethnic groups affected by a new system deployment. Additionally, system usage was shown

to lessen some privacy concerns, although the effects of use were not uniform across all ethnic groups. Therefore, system integrators may find it advantageous to provide

opportunities for use early in the implementation process. Tactics such as introducing trial systems and soliciting user feedback may help in this regard.
Although not empirically tested here, a review of the extant literature suggests that transparency (Awad et al., 2006), employee voice in system and policy design

(Armitage et al. 1999, Culnan and Armstong 1999), and written policies governing use (Jensen et al. 2005) have the potential to improve employee acceptance of biometric

technology. Taken together, the adoption of the strategies outlined in the preceding paragraphs may improve the likelihood of system success.
5.1 Limitations
Although the experiment sample size was carefully calculated as being large enough to reliably detect large effects, it may not have been sufficient to detect smaller size

effects. This may account for some of the insignificant results in the pre-test / post-test hypotheses. Another limitation was the lack of female participants. Researchers

should exercise care when interpreting the results as they may not hold for predominantly female or mixed gender populations. The lack of female participation was an artifact

of the organization selected for the study as firefighting is generally a male-dominated field. However, the organization had many other desirable attributes and was an

appropriate setting for an initial study on biometric privacy concerns and ethnicity in association with biometric technology in the workplace.
5.2 Directions for Future Research
The results of this study suggest several avenues for future research. The first such avenue would be to replicate the study in an organization with a mix of female and

male subjects. Confirmation of the findings from this study with a mixed gender population would provide considerable additional insight into the study of ethnic influence and

behavior.
Another avenue for future research is refinement of the newly developed privacy constructs and their associated scales. Although the scales have been subjected to a

rigorous validation protocol with a large pool of respondents, there may be additional privacy dimensions that were not identified in the initial scale development process.

Identification of additional concerns could improve the predictive power of the instrument. One factor that merits consideration may be computer anxiety, or a more specific

form of anxiety related to biometrics / privacy-related technologies. The significant reductions in the perceived vulnerability concerns of Anglos after use and the

significant reduction in perceived distrust for the combined study sample after use suggest that an examination of computer anxiety or system familiarity factors is warranted

in future studies.
Finally, determining the relative importance of items and constructs may be useful in enhancing the validity of both the instrument and theoretical model. Assessment via

partial least squares (PLS) regression techniques may help in this regard. PLS regression is particularly suitable when there is some degree of multicollinearity in the data

as may be expected when investigating dimensions of a multi-faceted construct such as privacy.
This study showed that perceived accountability, perceived vulnerability, and distrust concerns are affected by ethnic cultural differences in the context of a crisis

management system. It further showed that Anglos and Hispanics are relatively equal in their overall level of privacy concern but the underlying dimensions of that concern

vary considerably based on ethnic affiliation. These findings have important implications for the design and implementation of future crisis response systems as ethnic

affiliation and privacy concerns may both affect user acceptance of these systems.

Acknowledgements

Portions of this research were supported by the National Institute of Standards and Technology Building and Fire Research Laboratory via Fire Research Grant #05-388, Award

# 70NANB6H6124 for a project entitled ‘The Integration of Biometric Technologies with Fire Fighting Information Systems’. Support included funding for the pilot system

implemented in the study and faculty support during the research effort.

Source: Essay UK - http://doghouse.net/essays/information-technology/essay-multicultural-study-of-biometric-privacy-concerns-in-a-fire-ground-accountability-crisis-response-system/


Not what you're looking for?

Search our thousands of essays:

Search:


About this resource

This Information Technology essay was submitted to us by a student in order to help you with your studies.



Word count:

This page has approximately words.


Share:


Cite:

If you use part of this page in your own work, you need to provide a citation, as follows:

Essay UK, Essay: Multicultural Study of Biometric Privacy Concerns in a Fire Ground Accountability Crisis Response System. Available from: <http://doghouse.net/essays/information-technology/essay-multicultural-study-of-biometric-privacy-concerns-in-a-fire-ground-accountability-crisis-response-system/> [21-02-19].


More information:

If you are the original author of this content and no longer wish to have it published on our website then please click on the link below to request removal:


Essay and dissertation help


Latest essays in this category:


Our free essays:

badges