Escolar Documentos
Profissional Documentos
Cultura Documentos
As with anxiety, uncertainty or ambiguity is not been directly measured and compared with the
always a feature of fright because the harm is type of fear experienced with other health-related dona-
always in the future. However, since the dan- tion gifts. The type of fear experienced when signing
ger is concrete and sudden and there is little an organ donor card relates to Lazarus’s explanation
time to reflect, uncertainty is not always as of anxiety.
prominent in fright as it is in anxiety, where On the other hand, in blood donation, the fears that
the threat is symbolic, existential, and individuals experience are consistent with Lazarus’s
ephemeral.5(p235) explication of fright. These fears are concrete threats
of immediate physical harm. Mathew et al13 and Sojka
Lazarus conceptually distinguishes fright and anx- and Sojka14 explored the specific deterrents associated
iety as fear of the known and unknown, respectively. with blood donation, including fear of needles, pain,
Fright and anxiety provoke the action tendencies of and fear of contracting or finding out about a disease.
escape and avoidance, respectively. Because the harm Researchers have found that the fears experienced with
is understood in fright, the avoidance impulse is direct. blood donation are associated with immediate physi-
With anxiety, the source of the threat is not fully under- cal harm, and this emotional experience fits Lazarus’s
stood because of its existential nature and, therefore, definition of fright.1,15,16
neither is the action tendency and thus avoidance. The purpose of the current study was to test
Lazarus and Lazarus6(p54) described the differences Lazarus’s theoretical explanation of fear types of fright
in how people experience the emotions of anxiety and and anxiety in the contexts of blood and organ dona-
fright. Anxiety is an uncertain threat, often experi- tion and to understand further how the fear types are
enced as “low or moderate intensity, a state of distress distinct.5 Specifically, we set out to create and test an
that remains chronic or recurrent.” Anxiety makes peo- empirically valid measure of the emotions associated
ple uneasy about an existential concern: often fear of with giving blood and signing an organ donor card.
death. Fright is an immediate prospect of injury or
pain. The experience of fright is higher in intensity than Methods and Results
anxiety. Still, Lazarus and Lazarus6(p50) describe the 2 Phase 1: Item Generation
fear types as “close cousins,” thus, the 2 constructs are The initial phase of the scale development was
theoretically related. item generation. Two 5-item questionnaires were gen-
Lazarus’s conceptualization of anxiety is pertinent erated to measure fright and anxiety in the context of
to organ donation as research has shown that fear of blood and organ donation acts (signing an organ donor
death and anxiety is associated with organ donation. card). The items from these questionnaires can be
Amir and Haskell7 found a positive correlation between viewed in Table 1, but the actual measures did not con-
participants who had not signed an organ donor card tain labels. Two emotion researchers created the items
and fear of death. In addition, Smith et al8 created mes- based on Lazuras’s5 theoretical explanation of the 2
sages to reduce anxiety associated with signing an organ fear types, and the items were further scrutinized by a
donor card but did not measure anxiety directly. Lester9 doctoral level research methods class at a large Mid-
found a positive relationship between scores on a fear western University as part of an in-class activity. For
of death scale and an unwillingness to donate organs. each questionnaire, 2 of the items were stated from a
People have described their feelings associated positive perspective (eg, “Needles are not a problem
with signing an organ donor card as uncomfortable.10 for me.”) to reduce the potentially biasing effect of
In their directions for future research, Morgan et al11 responding to all negative statements. Also, positive and
emphasized the importance of studying the relation- negative items were randomly interspersed in each scale.
ship between existential anxiety and organ donation Fear measures include pain associated with donating,
directly. The researchers explained that they had not concern about disease/infection and lowered immune
included the measurement of anxiety in their current resistance, and general anxiety/nervousness. Seven-point
study “because of difficulties involved in the opera- (strongly agree to strongly disagree) scales were used.
tionalization in the face of little extant research and a
dearth of reliable, valid, measures.”11(p655) Furthermore, Phase 2: Initial Data Collection
the researchers stressed the need to study anxiety stat- The item-generation phase of the study resulted in
ing “although this is rarely mentioned directly, exis- a 10-item scale, 5 items meant to measure fear of giv-
tential anxiety almost surely plays a role in personal ing blood and 5 items measuring anxiety associated
reactions to organ donation.”11(p655) Subsequent research with being an organ donor. Although we expect the 2
included 2 questions to measure the anxiety associated factors to correlate positively and significantly with
with the lack of information about family members’ each other, we also posit that the 2 factors will demon-
donation beliefs.12 Still, the anxiety experienced in strate a clear distinctiveness when compared with
relation to the decision to sign an organ donor card has each other.
Table 1 Items from questionnaires about fear of blood donation and anxiety about signing an organ donor carda
The following items are to assess your feelings about blood donation and signing an organ donor card. Your answers will be used
only for research purposes. Please answer the items honestly, in terms of how you really feel, not how you would like to be.
Strongly disagree Strongly agree
1 2 3 4 5 6 7
Blood donation: fright
1. Donating blood would involve too much pain for me.
2. I am generally not frightened to donate blood.
3. Needles are not a problem for me.
*4. I can pinpoint my fears toward donating blood.
*5. I feel that real harm will result from my involvement in blood donation.
Signing an organ donor card: anxiety
*6. I am not sure why I feel uncomfortable with the idea of donating organs.
7. The idea of donating organs makes me anxious.
8. The idea of donating organs makes me nervous.
9. I feel uneasy about donating organs.
10. The thought of donating organs gives me an uneasy feeling of what might happen in the future.
a Items with an asterisk were removed because of poor fit of the statistical model.
To test the different types of fear in the contexts of poor factor loadings and low reliability estimates.
of blood and organ donation, 509 undergraduate stu- Specifically, 2 items from the fear of giving blood fac-
dents from a large Midwestern university voluntarily tor and 1 item from the anxiety associated with being
agreed to participate in the study. The university’s an organ donor factor were omitted (Table 1). The
institutional review board approved the study. Partici- reliability coefficient for the fear of giving blood fac-
pants were recruited from introductory communica- tor was 0.826, and the reliability coefficient for anxiety
tion classes, and research participation was voluntary. associated with being an organ donor factor was 0.832.
All participants were handed measures with an informed At this point, we subjected these data to a confir-
consent sheet as the top page and asked to fill them out matory factor analysis (CFA) in order to verify statis-
anonymously. Students reviewed the informed con- tically the structure of the reduced, 7-item instrument
sent sheet and elected whether or not to participate. and the accompanying factors. The test of a measure-
Those who chose to participate filled out the measure ment model with the 2 hypothesized latent variables
(see Table 1) and returned their responses to a research produced a very good fit, χ 13 2 = 11.0, P < .61, compar-
assistant. To protect anonymity, all informed consent ative fit index [CFI] = 1.0, incremental fit index [IFI]
documents and measures were disassembled before = 1.0, and root-mean-square error of approximation
data entry. [RMSEA] = 0.0 (the χ 21 2 for the null model was 946).
This student population is desired because healthy In addition to the nonsignificant χ 2 estimate, all
young adults are ideal donors for both blood and sign- parameter estimates loaded significantly onto their
ing an organ donor card. This population is more often respective factors, P values < .001) with an average
engaged in more reckless behavior than other seg- loading of 0.76 (ranging from 0.56 to 0.93). As
ments of the population, making them ideal candidates expected, the 2 factors—fear of giving blood and anx-
for donation of healthy organs. Also, previous research17 iety associated with being an organ donor—were pos-
has shown that college students account for a large itively and significantly correlated (r = 0.28, P = .01),
percentage of the blood donor population. Participants but at the same time, demonstrated a fair degree of
in this study were 61.6% female, predominantly white, distinctiveness.
and ranged in age from 18 to 27 years old. The dimension variance of each scale item was
In the initial data analysis, we tested the scale with also computed. These components of variance provide
a sample of all 314 female participants taken from the additional evidence of the reliability of scores on indi-
total 509 participants in the study. Initial analyses con- vidual items. A widely accepted critical value, 0.50,
sisted of both exploratory factor analysis (principal specifies that at least 50% of the variance in an item is
components analysis) and reliability estimates (Cron- due to the hypothesized dimension.18,19 The remainder
bach α) and revealed a more parsimonious scale. of variance is systematic nondimension-related and
Although a 2-factor solution adhering to a priori expec- random error. Approximately 71% of the individual
tations (a fear of giving blood factor and an anxiety item reliabilities (5 of 7) either met or exceeded the
associated with being an organ donor factor) explained critical value of 0.50 with an average variance of 0.60
the data, 3 items were deemed inappropriate because (range, 0.31-0.86). The dimension-related indices of
Root-mean-square
Hypothesis χ2 df Comparative fit index Incremental fit index error of approximation
Baseline 1352.9 42
1 36.1 26 0.99 0.99 0.04
2 57.0 31 0.98 0.98 0.06
3 59.8 32 0.97 0.98 0.06
average variance extracted are considered additional associated with being an organ donor factors were
indicators of reliability. Because each of the 2 factors 0.69 and 0.82, respectively. All parameter estimates
explained at least 50% of the variance (0.62 and 0.58 loaded significantly on to their respective factors (P
for fear of giving blood and anxiety associated with values < .001) with an average loading of 0.70 (rang-
being an organ donor, respectively) in the average ing from 0.51 to 0.86). The 2 factors were again posi-
item, the items were deemed satisfactory. tively and significantly correlated (r = 0.21, P = .05)
yet conceptually distinct.
Phase 3: A Second Data Analysis Effort The dimension variance of each scale item indi-
In a second data analysis effort, we tested the cated that approximately 57% of the individual item
scale with all 195 male participants from a total sam- reliabilities (4 of 7) either met or exceeded the critical
ple of 509 participants. We chose males such that we value of 0.50 with an average variance of 0.50 (rang-
could subsequently compare both male and female ing from 0.26 to 0.74). Further, each factor explained
samples on the newly devised scale as we have no rea- or came very close to explaining at least 50% of the
son to believe that the 2 sexes should differ according variance in the average item (0.44 and 0.55 for fear of
to their fear or anxiety levels associated with giving giving blood and anxiety associated with being an
blood and being an organ donor, respectively. In this organ donor, respectively), and, as before, the items
way, the scale can be compared directly across 2 sim- were deemed satisfactory. Next, we conducted the
ilar yet independent samples for structural invariance. multisample analysis.
For the scale to be generally applicable, it should have
similar structural properties across comparable yet Phase 4: A Multisample Analysis
independent samples. If a scale is not nearly equiva- As alluded to earlier, it was predicted that the
lent across similar yet independent samples, then the structure of the “blood-fear, organ-anxiety” scale would
relation of fear and/or anxiety levels to other variables be similar for the 2 sexes (samples). In this case, we
can be confounded by scaling differences, and the estimate simultaneously a measurement model for each
comparison of standings on the scale across samples independent sample. As a rule, multisample models
becomes superfluous. It was determined a priori to not have large χ2 values and can be difficult to estimate.21
only test statistically the structure of the scale with the Further, it is unrealistic to expect a multisample analy-
newly collected data (eg, conduct a CFA with the sam- sis to produce a better fitting model than any single-
ple of males), but also to compare statistically the struc- sample model, given that the lack of fit in 1 sample
ture of the scale across the 2 samples (the female and leads to lack of fit in the multiple-group model, and
male samples) by using a structural equation, multiple- the lack of fit in any 1 sample contributes to the over-
group modeling approach (ie, simultaneously estimat- all χ2 test.
ing a measurement model across 2 independent groups Three specific hypotheses were tested in a multi-
or samples). sample analysis (Table 2). Overall, the hypotheses eval-
The results of the CFA for the male data revealed uate the extent to which the item responses are similar
a well-fitting model estimating the same 2 hypothe- across the 2 samples (ie, invariance). The test for each
sized latent variables measured with the male data, hypothesis imposes increasingly more stringent con-
2 = 25.05, P = .02, CFI = 0.97, IFI = 0.97, RMSEA =
χ 13 straints, forcing the 2 models to be more and more sim-
0.07 (the χ 221 for the null model was 407.11). ilar. The extent to which these constraints are shown
Although the χ2 was significant in this model at P<.05, to be statistically appropriate further validates the
the remaining fit indices suggest a fairly well-fitting invariant nature of the scale.
model. For instance, RMSEA values up to 0.08 usu- First, a test of form (H1) in the present context is
ally signal acceptable model fit.20 The reliability whether the model for the 2 samples (corresponding to
coefficients for the fear of giving blood and anxiety the female and male samples) has the same number of
Table 3 Items from questionnaire about anxiety about blood donation and fear of organ donation
The following items are to assess your feelings about blood donation and signing an organ donor card. Your answers will be used
only for research purposes. Please answer the items honestly, in terms of how you really feel, not how you would like to be.
Strongly disagree Strongly agree
1 2 3 4 5 6 7
Blood donation: anxiety
1. The idea of donating blood makes me anxious.
2. The idea of donating blood makes me nervous.
3. I feel uneasy about donating blood.
4. The thought of donating blood gives me an uneasy feeling of what might happen in the future.
Signing an organ donor card: fright
5. Donating organs would involve too much pain for me.
6. I am generally not frightened to donate organs.
7. I believe that signing an organ donor card would not affect the care I receive at a hospital.
latent variables (factors) with the same indicators, and continued support for the invariance prediction. An
the same specification of fixed and free parameters. examination of interfactor correlations revealed simi-
Table 2 presents the χ2 estimates and summary fit indices lar values across the samples, and the Δχ2 of H3 and
generated by this test of form. The insignificant χ2 H2 (2.8 with 1 degree of freedom) is not significant (P
value (36.1, df = 26, P = .09) and high values for the fit = .05). In the end, ample evidence suggests that the
indices lead to the conclusion that the model form holds factor intercorrelations are equivalent across the 2 inde-
similarly for the 2 samples. Thus, H1 is supported. pendent groups or samples, and H3 is supported.
Having received support for this initial test of Overall, the multisample analyses corroborate the
invariance, we conducted 2 subsequent tests of invari- consistency of the empirically derived models across
ance in a hierarchical fashion. The hypothesis that the the 2 independent samples of male and female respon-
factor coefficients were equal for the 2 groups (H2) dents. Such findings give us the confidence that the
was tested next. This hypothesis restricted the regres- factors derived from the present inventory are valid,
sion of the items on the factors (ie, slopes relating the reliable, and robust. Next, we estimate an alternative
measures to the latent variables) to be equal for the 2 model for comparison purposes.
samples. Although the χ2 estimate of 57.0 with 31
degrees of freedom was statistically significant (P = Phase Five: An Alternative Model
.003), the relatively high goodness-of-fit measures In a final attempt to verify the current “blood-
accompanying this test (CFI = 0.98, IFI = 0.98, fear, organ-anxiety” scale, we estimated an alternative
RMSEA = 0.06) provided evidence that the factor struc- model. Specifically, we derived a blood-anxiety, organ-
ture is largely invariant across the 2 groups. The χ2 fear scale and compared its measurement model to the
difference (Δχ2) between H2 and H1 is 20.9 with 5 current blood-fear, organ-anxiety measurement model.
degrees of freedom. When compared with a cutoff We created the new scale by replacing the words “donat-
critical χ2 value of 20.5 (P < .001), this test suggests ing blood” with the words “donating organs” across the
caution in accepting the added restriction (ie, equal entire 7-item scale (Table 3). Thus, the scales were con-
slopes). The individual parameter estimates for the 2 sistent in theme and addressed the same dependent vari-
samples (models), however, indicate that the 2 sets of ables. We then had the same group of male and female
factor coefficients are in fact highly similar. The aver- respondents reply to the newly devised inventory.
age difference in standardized factor coefficients across Having found consistent evidence that the male and
the 2 samples, for instance, is 0.077 with the single female samples were largely invariant with respect to
largest difference equal to 0.172. This result suggests the structure of the blood-fear, organ-anxiety scale, we
that the Δχ2 test is very sensitive.22 combined the 2 samples (males and females) for this
The final test of invariance added the restriction phase of the research. Thus, we endeavor to compare the
that the covariance matrix for the 3 empirically derived original blood-fear, organ-anxiety measurement model
factors was invariant across the 2 samples (H3). This (Table 1) by using a combined male-female sample with
implies that the factor covariances (or interfactor cor- a blood-anxiety, organ-fear measurement model (Table
relations) are equal in the 2 samples. Table 2 shows 3) that uses the same combined sample. We assume the
that H3 had a χ2 estimate of 59.8 with 32 degrees of latter model will evidence a worse fit with the data
freedom (P = .002) and fit indices that were very high according to extant theory (see earlier portions of this
(CFI=0.98, IFI=0.98, RMSEA=0.059), thus providing article). Because the 2 scales are not equivalent, we do
14. Sojka BN, Sojka P. The blood donation experience: self-reported elation, gladness, and joy. J Personality Soc Psychol. 1991;
motives and obstacles for donating blood. Vox Sanguinis. 61(1):98-104.
2008;94(1):56-63. 19. Fornell C, Larcker DF. Structural equation models with unob-
15. Duboz P, Cunéo B. How barriers to blood donation differ servable variables and measurement error. J Marketing Res.
between lapsed donors and non-donors in France. Transfusion 1981;18(3):382-388.
Med. 2010;20(4):227-236. 20. Browne MW, Cudeck R. Alternative ways of assessing model
16. France CR, Montalva R, France JL, Trost Z. Enhancing atti- fit. In: Bollen KA, Long JS, eds. Testing Structural Equation
tudes and intentions in prospective blood donors: evaluation Models. Newbury Park, CA: Sage; 1993:136-166.
of a new donor recruitment brochure. Transfusion. 2008; 21. Bentler PM, Bonett DG. Significance tests and goodness of fit
48(3):526-530. in the analysis of covariance structures. Psychol Bull.
17. Nonis SA, Ford CW, Logan L. College students’ blood dona- 1980;88(3):588-606.
tion behavior: relationships to demographics, perceived risk, 22. Hair JF, Anderson RE, Tatham RL, Black WC. Multivariate
and incentives. Health Marketing Q. 1996;13(4):33-46. Data Analysis. 5th ed. Upper Saddle River, NJ: Prentice-
18. Bagozzi RP. Further thoughts on the validity of measures of Hall; 1998.