Social Science & Medicine 96 (2013) 86e94

Contents lists available at ScienceDirect

Social Science & Medicine

journal homepage: www.elsevier.com/locate/socscimed

Review

A systematic review and meta-analysis of antecedents of blood

donation behavior and intentions
Timothy C. Bednall a, *, Liliana L. Bove b, Ali Cheetham b, Andrea L. Murray c
a
The Australian School of Business, School of Management, The University of New South Wales, Level 5, Building E12, New South Wales 2052, Australia
b
Department of Management and Marketing, The University of Melbourne, Level 9, 198 Berkeley Street, Victoria 3010, Australia
c
Research and Development, The Australian Red Cross Blood Service, Australia

a r t i c l e i n f o a b s t r a c t

Article history: This meta-analysis sought to identify the strongest antecedents of blood donation behavior and in-
Available online 1 August 2013 tentions. It synthesized the results of 24 predictive correlational studies of donation behavior and 37
studies of donation intentions. The antecedents were grouped into six research programs: (1) the Theory
Keywords: of Planned Behavior (TPB) and its extensions, (2) prosocial motivation, (3) affective expectations, (4)
Blood donation donor site experience, (5) past donation behavior, and (6) donor demographics. Antecedent categories
Blood donors
were cross-validated by multiple coders, and combined effect sizes were analyzed using a random-
Meta-analysis
effects model. For donation behavior, medium positive associations were found with five of the con-
Theory of Planned Behavior
Prosocial behavior
structs from the extended TPB: intentions to donate, perceived behavioral control, attitude toward
Prosocial motivation donation, self-efficacy and donor role identity. Other antecedents displaying a positive association with
donation behavior included anticipated regret for not donating, number of past donations and donor age.
Donor experiences at the collection site in the form of temporary deferral or adverse reactions had a
medium negative association with behavior. For donation intentions, strong positive associations were
observed for perceived behavioral control, attitude, self-efficacy, role identity and anticipated regret.
Medium positive associations were observed for personal moral norm, subjective norm, satisfaction, and
service quality. All other potential antecedents had weak or non-significant associations with behavior
and intentions. Several of these associations were moderated by between-study differences, including
donor experience, the period of data collection in which donation behavior was observed, and the use of
a nominal (yes/no return) versus a ratio measure of donation behavior. Collectively, the results under-
score the importance of enhancing donors’ attitudes towards donation and building their perceived
behavioral control and self-efficacy to donate. Further, minimizing the risk of adverse reactions and
enacting re-recruitment policies for temporarily deferred donors will help protect future donation
behavior. Implications of these findings for blood collection agencies and researchers are discussed.
Ó 2013 Elsevier Ltd. All rights reserved.

Since the 1950s (Walsh & Clemens, 1956), hundreds of studies
conducted worldwide have investigated factors that motivate and
discourage blood donation. Given the large quantity of studies yet
mixed findings, synthesizing the results is essential to both re-
searchers and strategists within blood collection agencies (BCAs). A
robust synthesis guides researchers to invest in areas that show
promise or are underexplored, assists BCAs in their donor man-
agement practices, and informs public health campaigns (Adams,
Erwin, & Le, 2009).
* Corresponding author. Tel.: þ61 2 9385 9713; fax: þ61 2 9313 6775.
E-mail addresses: t.bednall@unsw.edu.au, tbednall@gmail.com (T.C. Bednall),
lbove@unimelb.edu.au (L.L. Bove).
At least six major research programs have been undertaken to
understand the antecedents of donation behavior and intentions.
First, many studies have applied the Theory of Planned Behavior
(TPB; Ajzen, 1991), proposing that the decision to donate depends
on a positive outlook towards blood donation (attitude), perceived
social pressure to donate (subjective norm), and beliefs that
donation is under the individual’s control (perceived behavioral
control). This basic model has subsequently been extended with
additional factors that have each increased the model’s predictive
power (e.g., Masser, White, Hyde, Terry, & Robinson, 2009).
Notably, these factors include the extent to which individuals are
confident in their capability to give blood (self-efficacy), and
perceptions of whether significant others (family, friends or work
colleagues) are donating (descriptive norm). The importance of

0277-9536/$ e see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.socscimed.2013.07.022
 T.C. Bednall et al. / Social Science & Medicine 96 (2013) 86e94
being a blood donor to the self (donor role identity) has also been
included in extended TPB models. Based on Identity Theory
(Stryker, 1980), individuals are attracted to the favorable image of
donors, believing them to be altruistic and responsible (Piliavin &
Callero, 1991).
Second, motivational differences in altruism and felt obligation
have been investigated as possible drivers of donations. Altruism
has typically been defined as a general desire to increase the wel-
fare of other people with no explicit benefit to oneself (Batson,
Ahmad, & Tsang, 2002). Conversely, obligation has been defined
as a moral belief that one ought to help others (Callero, Howard, &
Piliavin, 1987). Measures of obligation have typically been based on
Schwartz’s (1977) theory of personal moral norms: that people
possess internal standards of how they should behave, which are
activated by particular situations (e.g., a request to donate). Both
altruism and personal moral norms can be thought of as forms of
prosocial motivation, in that they represent a desire to have a
positive impact on other people or social collectives through blood
donation.
Third, several studies have examined the influence of antici-
pated affective reactions. While the TBP assumes people make
decisions rationally, emotional outcomes are known to play an
important role (Conner & Armitage, 1998). Studies on emotional
outcomes have primarily focused on pre-donation anxiety (e.g.,
concern about pain), which has frequently been reported as a
barrier to donation (e.g., Lemmens et al., 2005; Robinson, Masser,
White, Hyde, & Terry, 2008). Other studies have investigated
whether anticipated regret for not donating influences return
behavior (Godin et al., 2005). The anticipation of positive emotions
such as satisfaction or pride following donation has also been
studied. This concept has been termed “benevolence”: an “impure”
type of altruism motivated by the prospect of feeling good about
oneself (Ferguson, Farrell, & Lawrence, 2008).
Fourth, studies have examined the influence of positive and
negative experiences at the collection center. In terms of positive
experiences, perceived service quality (including competence and
courtesy of collection staff) and overall satisfaction with the service
experience are thought to be important in encouraging donors to
return (France, Ditto, France, & Himawan, 2008). Conversely, the
experience of adverse reactions such as dizziness, nausea, or
fainting, discourages donors from returning (Ferguson & Bibby,
2002). Temporary medical deferrals, such as for low hemoglobin,
also reduce the likelihood of repeat donations (Custer et al., 2011;
Hillgrove, Moore, Doherty, & Ryan, 2011), although this outcome
can be redressed through telephone or mail re-recruitment of do-
nors (Delage & Dubac, 2007).
Fifth, research has investigated how donor career stage in-
fluences return donation behavior. Stage models of donation
behavior (Ferguson, 1996; Piliavin, Howard, & Callero, 1991) as-
sume that novice donors carefully weigh up the perceived benefits
and costs of donating, while donating becomes habitual for
experienced donors. Thus, experience with blood donation in-
fluences both intentions and actual donation behavior (Piliavin
et al., 1991).
Finally, studies have investigated how donor demographics (e.g.,
gender, age, culture and education) influence donation behavior.
The typical donor has been described as an educated, middle-aged
Caucasian male (Shaz, James, Hillyer, Schreiber, & Hillyer, 2011;
Veldhuizen, Doggen, Atsma, & De Kort, 2009). The prevalence of
males is due to females being more frequently deferred (Misje,
Bosnes, & Heier, 2010).
To date, several reviews of the donor motivation literature have
been produced (Ferguson, France, Abraham, Ditto, & Sheeran,
2007; Gillespie & Hillyer, 2002; Masser, White, Hyde, & Terry,
2008; Piliavin, 1990). However, these reviews have all been
87
qualitative summaries of the literature, which may lead to a dis-
torted subjective summary of the empirical data (Hedges, 2009).
In contrast, a meta-analysis provides a highly systematic and
objective approach to reviewing empirical literature, by outlining
a strategy for collecting primary studies, applying selection
criteria, developing and cross-validating variable categories, and
synthesizing findings across studies. By combining study results, a
meta-analysis has high statistical power to detect small effect
sizes; it can reveal diversity of findings across studies and explain
between-study differences.
Meta-analytic reviews of the blood donor literature have
examined the efficacy of interventions for promoting blood dona-
tion (Godin, Vézina-Im, Bélanger-Gravel, & Amireault, 2011), and
the prevalence of self-reported motivators and deterrents for
donating (Bednall & Bove, 2011). However, with the exception of an
early meta-analysis (Ferguson, 1996), which focused primarily on
factors associated with the TPB, none of the aforementioned meta-
analyses have reviewed literature investigating factors that influ-
ence subsequent donation behavior. The current study updates
Ferguson’s review by incorporating the substantial body of litera-
ture on donation antecedents produced in the last two decades. It
additionally incorporates findings from the five other research
streams, including factors related to donor motivational differ-
ences, affective reactions, collection center experiences, career
stage, and demographics. Moreover, it assesses the impact of
methodological and contextual differences between studies via
moderator analyses.
The first aim of this review is to identify the strongest ante-
cedents of donation behavior and intentions. We present findings
associated with each outcome in separate analyses; we then
compare each set of findings and discuss their joint implications for
each of the six research programs. The second aim is to assess the
impact of between-study differences on the effects of each ante-
cedent. The first moderator we examine is donor experience (i.e.,
donors vs. non-donors, and mean number of past donations), as
donors’ career stage is thought to influence their motivation to
donate (Piliavin et al., 1991). For studies of donation behavior, we
additionally examine the impact of the length of data collection
(i.e., the number of months between the assessment of each
antecedent and subsequent donation behavior), as intentione
behavior relationships are thought to diminish over longer periods
(Sutton, 2006). We also assess the impact of using retention (i.e.,
donated or did not donate) versus frequency (i.e., number of do-
nations) as the measure of donation behavior, given that dichoto-
mization can reduce the size of correlations (Sheeran, 2002).
Method
Fig. 1 presents a summary of the procedure used for selecting
the studies for this meta-analysis, including the literature review,
application of exclusion criteria, and development and cross-
validation of predictor categories.
Literature search
Our review attempted to achieve an exhaustive coverage of
studies investigating antecedents of blood donation behavior (as
measured by actual attendance records) and self-reported in-
tentions to donate. We investigated populations of current donors
and eligible non-donors who had the opportunity to donate for the
first time during each study’s data collection period.
We collected both published and “grey” literature (Rothstein &
Hopewell, 2009), including dissertations and conference papers.
Studies were collected using multiple methods: (1) a manual search
through the Australian Red Cross Blood Service electronic library of
88 T.C. Bednall et al. / Social Science & Medicine 96 (2013) 86e94

1) Manual search of blood service database

Literature 2) Search of databases Web of Science, MEDLINE, PsycINFO and Dissertations Abstracts
3) Reference lists from previous reviews of blood donor motivation
Search 4) Cited reference search of each review
(N = 156) 5) Manual searches of Social Science and Medicine, Transfusion, Vox Sanguinis, Health
Psychology
6) Requests to authors for additional studies

Exclusion criteria
Application of 1) Study was not an investigation of the antecedents of blood donation behavior or
intentions (60 studies excluded)
exclusion 2) The outcome measure was not donation behavior, or intentions to donate blood (12
criteria studies excluded)
3) There was not enough information to calculate a Pearson correlation coefficient (44
studies excluded)
Note: Some studies met multiple exclusion criteria

1) Peer reviewed journal articles (34)

Included 2) Doctoral dissertations (6)
Studies 3) Conference paper (3)
4) Book chapters (2)
(N = 47) 5) BCA research reports (2)

21 antecedents of donation behavior

Development and 20 antecedents of donation intentions
cross-validation Cohen’s kappa = .866
of predictor
categories

Fig. 1. Flowchart of study selection for systematic review and meta-analysis.

research papers, (2) a search of the databases Web of Science,
MEDLINE, PsycINFO, and Dissertation Abstracts using a Boolean
combination of keywords,1 (3) an examination of reference lists in
previously published reviews of blood donor motivation (Ferguson,
1996; Ferguson et al., 2007; James, Hewitt, & Barbera, 1999; Masser
et al., 2008; Newman, 2001; Oswalt, 1977; Piliavin, 1990; Piliavin &
Callero, 1991), (4) a cited reference search of each review article
using Web of Science, (5) a manual search of the journals Social
Science & Medicine, Transfusion, Vox Sanguinis and Health Psychology
from their earliest to most recent issues, and (6) a request for
further published and unpublished studies to the author(s) of each
study already collected. The search concluded in February 2012,
and generated 153 studies for possible inclusion.
Exclusion criteria and coding procedure
After compiling the initial list of studies, one investigator
entered information about each study into an electronic coding
form, including the study design, sample characteristics, and type
of predictor and outcome variables. A second investigator cross-
checked this information, with inconsistencies resolved through
discussion. A study was excluded if both investigators agreed it met
any one of the following criteria:
 It was not an empirical investigation of the antecedents of
donation behavior or intentions (e.g., it was a retrospective
correlational study, a descriptive study, or qualitative study).
 It did not use records of actual donation behavior (including
retention or frequency) or self-reported intention to donate as
a dependent variable.
 It reported insufficient information to compute a Pearson cor-
relation effect size (e.g., the results were only presented in a
1
The keyword combination was: (blood OR aphaeresis OR apheresis) AND (donor
OR donation) AND (psychology OR motivation OR intention OR attitude OR behavior OR
behaviour OR recruitment OR retention OR antecedent OR predictor OR altruism).
regression analysis), and the information could not be obtained
from the authors.
In total, there were 24 studies of donation behavior (comprising
36 samples and 1,285,256 participants) and 37 studies of donation
intentions (comprising 45 samples and 15,352 participants). Four-
teen of these studies investigated both behavior and intentions.
Non-donors were only present in studies of donation intentions. A
full list of studies and associated samples is presented in the
Supplementary materials.
Cross validation of predictor categories
In order to develop a cross-validated set of predictor variables, a
list of factors from each study was first compiled, which included
the labels, definitions, and measures of each factor. Two researchers
then independently inspected the list, and assigned each factor to a
category they had developed. Each category was based on a
construct from an established theoretical framework (e.g., subjec-
tive norm from the TPB), or a recognized concept from the blood
donor literature (e.g., adverse events; temporary deferrals). The
researchers used both the construct name given in each study (e.g.,
“subjective norm”) and associated items (e.g., “My friends think I
should donate blood”) to determine the most appropriate category.
Together, agreement was reached about the category labels, defi-
nitions and associated predictors. Inter-rater agreement was very
high (Cohen’s kappa ¼ .866), and all inconsistencies were resolved
through discussion together with a third researcher. The categories
are described in the Supplementary materials.
Effect size calculation and aggregation
In this phase, information from each study was recorded,
including effect sizes, sample characteristics, and reliability co-
efficients for each measure. A second researcher cross-checked the
accuracy of the data entry process, and resolved inconsistencies
through discussion with the first coder. If multiple effect sizes were
 T.C. Bednall et al. / Social Science & Medicine 96 (2013) 86e94
reported (e.g., two measures of the same factor), an average was
calculated and used in the analysis. All effect sizes e including
correlations, odds ratios, return rates, and mean differences e were
transformed into Pearson’s r correlation coefficients using the
conversion formulas described in Borenstein (2009). We used
Cohen’s (1992) guidelines for the interpretation of correlation co-
efficients, with .1, .3, and .5 indicating small, medium and large
effect sizes respectively.
If multiple studies reported results from the same dataset, the
effect sizes were included only once in the meta-analysis. A factor
was only included if it was reported in at least three studies. Some
studies reported inexact descriptions of their samples and mea-
sures (e.g., the smallest reliability coefficient from a set of measures,
sample size ranges for a set of correlations presented in a table). If
exact values could not be obtained from the researchers, the
smallest reported value was incorporated into the analysis.
Based on Schmidt, Le, & Oh (2009), each observed correlation
coefficient (robs) was adjusted for unreliability of the measures to
produce corrected correlation coefficients (rc). If reliability co-
efficients were not reported or provided upon request from the
study’s authors, the average coefficient reported in the other
studies was substituted. Measures derived from objective BCA re-
cords (e.g., donations, temporary deferrals) were assumed to have
perfect reliability (i.e., equal to 1).
The effects of each antecedent were aggregated separately for
donation behavior and intentions. Individual study findings were
aggregated by converting each Pearson correlation coefficient to
Fisher z scores, and weighting each result by the reciprocal of the
effect size variance (Shadish & Haddock, 2009). Using these con-
verted scores, weighted means, standard deviations, and confi-
dence and credibility intervals around each mean were then
calculated. Confidence intervals, which are based on the standard
error estimates of the mean effect sizes, reflect the level of certainty
about the value of the estimated weighted mean. Conversely,
credibility intervals are based on the distribution of effect sizes
across studies, and therefore reflect heterogeneity in findings
(Whitener, 1990). In the final step, these estimates were re-
converted into Pearson r scores. As this transformation is non-
linear, confidence and credibility intervals are asymmetrical
around the mean.
The data were analyzed using a random effects model, based on
the assumption there would be considerable heterogeneity among
the reported effect sizes due to each sample being drawn from
different donor populations across the world. Unlike fixed effects
models, which account for only within-study variability, random
effects models additionally incorporate between-study variability
(i.e., study population differences) into their standard error esti-
mates. Thus, they provide more conservative effect size estimates
and allow broader generalizations to be made (Schmidt, In-Sue, &
Hayes, 2009).
Assessing the impact of between-study differences and publication
bias
The studies differed in terms of donors’ experience, and for
studies of donation behavior, the length of the data collection
period, and type of outcome variable. To assess the impact of these
differences, a series of moderator analyses were conducted using
SPSS macros for weighted least squares regression (Wilson, 2005),
which were performed on the rc coefficients that had been con-
verted to Fisher z scores.
The threat of publication bias to the meta-analysis findings was
also assessed. A fail-safe Nþ score for each factor was calculated
(Rosenberg, 2005), which indicates the number of studies with an
effect size of zero that could be added without reducing the effect
89
size to non-significance. The scores reflect the robustness of the
meta-analysis findings against any systematic omission of studies
with non-significant results in the published literature.
Results
Across studies of donation behavior and intentions, the majority
took place in the USA, Australia and the UK. Owing to the large
number of university studies, approximately a third of the samples
comprised participants whose reported age was less than 25 years.
Most samples comprised repeat whole-blood donors. A detailed
summary of the demographic characteristics of each sample is
presented in the Supplementary materials.
Tables 1 and 2 present the associations between each ante-
cedent and donation behavior and intentions, respectively. The fail-
safe statistics indicate that all of the significant effect sizes are
robust to publication bias, with the exceptions of donation anxiety,
satisfaction with self and marital status in the case of donation
behavior, and knowledge about blood donation in the case of
donation intention.
Antecedents of donation behavior
With regard to the factors associated with the core TPB, dona-
tion intentions had the strongest association with donation
behavior, followed by perceived behavioral control and attitudes.
Conversely, the effect of subjective norm was small. Among the
extended TPB factors, medium positive effects were observed for
self-efficacy and donor role identity, whereas descriptive norm was
non-significant.
With regard to prosocial motivation, personal moral norms
showed a small positive association, while altruism was non-
significant. In terms of affective expectations, anticipated regret
had a positive medium association, whilst anxiety had a negative
small association. Satisfaction with self had a weak but significant
positive association.
In terms of collection site experiences, satisfaction and service
quality experiences were non-significant predictors of donation
behavior. Medium negative associations were observed for both
temporary deferrals and adverse reactions. In the case of temporary
deferrals there was substantial heterogeneity across the effect sizes,
as reflected in the wide credibility interval. While the data reported
in each study made it impossible to decompose the overall effect
sizes into specific types of deferrals (e.g., low hemoglobin versus
travel-based deferrals), we examined whether differences in BCA
approaches towards deferred donors were responsible for these
heterogeneous effects. A moderator analysis revealed that BCAs
that attempt to actively re-recruit temporarily deferred donors
(Custer, Chinn, Hirschler, Busch, & Murphy, 2007 [repeat donors];
Delage & Dubac, 2007) had significantly higher return rates than
BCAs without this policy, B ¼ 0.478, Z(1, 5) ¼ 2.387, p ¼ .017. Of the
donor characteristics, the number of previous donations had a
medium positive association, and age (older) had a small positive
association.
Moderator analyses revealed that between-study differences
influenced the effects of some of the antecedents. With regard to
the mean number of past donations for each sample, experienced
donors were more strongly influenced by self-efficacy (B ¼ 0.103,
Z ¼ 3.062, p ¼ .002) and regret for not donating (B ¼ 0.117,
Z ¼ 2.056, p ¼ .040). Conversely, experienced donors were less
likely to be deterred from subsequent donations by temporary
deferrals (B ¼ 0.149, Z ¼ 2.463, p ¼ .014). With regard to study time
frame, studies with longer periods of data collection showed
stronger effects of satisfaction with the service experience
(B ¼ 0.064, Z ¼ 3.071, p ¼ .002), temporary deferrals (B ¼ 0.009,
90 T.C. Bednall et al. / Social Science & Medicine 96 (2013) 86e94

Table 1
Factors predicting subsequent donations.

Variable k N robs SDrobs robs Intervals rc SDrc rc Intervals Fail Safe N

95% Conf. 80% Cred. 95% Conf. 80% Cred.

Theory of Planned Behavior (TPB)

Intention to donate 15 5911 .362 .004 [.245, .469] [.065, .600] .424 .007 [.282, .549] [.050, .694] 3484
Perceived behavioral control 5 2982 .311 .009 [.130, .472] [.064, .523] .381 .015 [.156, .568] [.062, .629] 608
Attitude toward donation 9 4249 .216 .002 [.135, .295] [.076, .348] .235 .002 [.148, .319] [.082, .378] 768
Subjective norm 10 4864 .165 .001 [.118, .212] [.099, .230] .198 .001 [.143, .250] [.112, .280] 485
Extended TPB
Self-efficacy 5 3088 .352 .012 [.153, .523] [.072, .580] .392 .016 [.164, .581] [.064, .644] 509
Donor role identity 10 4873 .232 .001 [.160, .302] [.100, .356] .263 .002 [.184, .338] [.115, .399] 968
Descriptive norm 3 1021 .027 .001 [.035, .088] [.027, .027] .029 .001 [.032, .091] [.029, .029] 0
Prosocial motivation
Personal moral norm 10 4874 .188 .001 [.117, .257] [.061, .310] .204 .002 [.126, .279] [.059, .340] 834
Altruism 3 885 .015 .001 [.082, .051] [.015, .015] .016 .001 [.082, .050] [.016, .016] 0
Affective expectations
Regret (for not donating) 4 2962 .254 .005 [.119, .379] [.096, .399] .268 .006 [.124, .401] [.096, .425] 320
Anxiety 3 681 .133 .001 [.206, .058] [.133, .133] .149 .001 [.222, .074] [.149, .149] 1
Satisfaction (with self) 3 1333 .097 .001 [.044, .150] [.097, .097] .097 .001 [.044, .150] [.097, .097] 1
Collection site experiences
Temporary deferral 7 473,384 .372 .032 [.633, .035] [.764, .219] .372 .032 [.633, .035] [.764, .219] 7084
Adverse reactions 11 803,733 .326 .002 [.397, .251] [.473, .161] .328 .002 [.397, .256] [.471, .169] 302,565
Satisfaction with service experience 3 3113 .092 .002 [.014, .168] [.022, .160] .105 .003 [.000, .208] [.001, .208] 47
Service quality 3 1012 .052 .001 [.018, .122] [.015, .089] .052 .001 [.018, .122] [.016, .089] 0
Donation experience
Number of past donations 8 3405 .271 .002 [.189, .350] [.133, .399] .271 .002 [.189, .350] [.133, .399] 503
Demographics
Age 7 3173 .251 .002 [.174, .325] [.133, .362] .251 .002 [.174, .325] [.133, .362] 415
Gender 5 2573 .132 .018 [.133, .379] [.245, .474] .132 .018 [.133, .379] [.245, .474] 0
Marital status 4 1145 .088 .001 [.030, .146] [.088, .088] .088 .001 [.030, .146] [.088, .088] 1
Education level 4 1145 .012 .001 [.071, .046] [.012, .012] .012 .001 [.071, .046] [.012, .012] 0

Note. k refers to the number of samples; N is the total number of people across samples, robs refers to the weighted average of the observed Pearson correlation coefficients
across samples; rc refers to the weighted average of the correlation coefficient corrected for study artifacts; SDobs and SDc refer to the standard deviation of robs and rc
respectively. Effect sizes that are significantly different from zero are indicated by a confidence interval that does not include zero. Fail Safe N scores refer to the Rosenberg
(2005) t  Nþ value.

Z ¼ 2.309, p ¼ .021), and age (B ¼ 0.017, Z ¼ 2.487, p ¼ .013). In
contrast, the effect sizes weakened over time for intentions
(B ¼ 0.034, Z ¼ 2.438, p ¼ .015) and gender (B ¼ 0.111,
Z ¼ 4.808, p < .001). When donation behavior was measured as a
dichotomous variable (versus donation frequency), stronger effects
were observed for subjective norm (B ¼ 0.104, Z ¼ 2.208, p ¼ .027),
attitude (B ¼ 0.151, Z ¼ 2.533, p ¼ .011), donor role identity
(B ¼ 0.137, Z ¼ 2.215, p ¼ .027), personal moral norm (B ¼ 0.162,
Z ¼ 3.614, p < .001) and satisfaction with the service experience
(B ¼ 0.128, Z ¼ 3.071, p ¼ .002).
Antecedents of donation intentions
With intention to donate as the dependent variable, similar
relationships were observed for each antecedent, although the ef-
fect sizes were typically larger. With regard to the antecedents
related to the TPB, large positive effects were observed for attitude,
perceived control, self-efficacy, and role identity. A medium posi-
tive association was observed for subjective norm, and a small
positive association was observed for descriptive norm.
In terms of prosocial motivation, personal moral norm showed
a medium positive association and altruism a small positive as-
sociation. With respect to affective expectations, positive associ-
ations were observed for regret (large) and satisfaction (small). A
small negative effect was observed for anxiety. For collection
center experiences, medium positive effects were observed for
satisfaction and service quality perceptions. Adverse reactions
showed a medium negative effect. In regard to donor character-
istics, number of past donations was positively associated with
intention to donate. Knowledge about blood donation showed a
weak, albeit significant association. All demographic antecedents
were non-significant.
Moderator analyses revealed that donors with a greater number
of past donations were more strongly influenced by anticipated
regret (B ¼ 0.163, Z ¼ 2.192, p ¼ .028), but less by donor role identity
(B ¼ 0.008, Z ¼ 1.991, p ¼ .047). When donors and non-donors
were compared, non-donors were found to be more strongly
influenced by descriptive norms (B ¼ 0.206, Z ¼ 5.362, p < .001)
and personal moral norms (B ¼ 0.272, Z ¼ 2.126, p ¼ .034).
Discussion
Incorporating studies of blood donor motivation drawn from the
last four decades, the purpose of this meta-analytic review was to
identify the factors that are most strongly associated with donation
behavior and intentions, and to assess the impact of methodolog-
ical and contextual differences between studies. These factors were
drawn from six major programs of research: the TPB, prosocial
motivation, affective expectations, collection site experiences, past
donation behavior, and donor demographics. Overall, the aggre-
gation of results from individual studies suggests that each pro-
gram offers useful insights. Although there was considerable
variability in the effect sizes, the positive or negative influence of
each factor remained largely consistent across studies.
Our findings suggest that the TPB and its associated factors
explain a large proportion of the variance in donation intentions,
but are less strongly associated with behavior. In particular, self-
efficacy, role identity, perceived behavioral control and attitude
show robust positive correlations with both outcomes. Intentions
in turn are the strongest predictor of behavior. These results were
generally consistent with reviews of the TPB in the context of blood
donation (e.g., Masser et al., 2008).
As shown in the moderator analyses, self-efficacy was more
influential for experienced donors. With greater self-knowledge,
 T.C. Bednall et al. / Social Science & Medicine 96 (2013) 86e94 91

Table 2
Factors predicting intention to donate.

Variable k N robs SDrobs robs Intervals rc SDrc rc Intervals Fail Safe N

95% Conf. 80% Cred. 95% Conf. 80% Cred.

Theory of Planned Behavior (TPB)

Attitude toward donation 28 8925 .536 .002 [.471, .596] [.299, .711] .626 .003 [.552, .691] [.338, .807] 30,304
Perceived behavioral control 14 4762 .415 .008 [.256, .551] [.014, .701] .547 .024 [.298, .727] [.131, .877] 15,530
Subjective norm 27 8972 .358 .001 [.303, .412] [.179, .515] .439 .002 [.375, .499] [.222, .615] 13,857
Extended TPB
Self-efficacy 14 5914 .632 .008 [.518, .724] [.323, .819] .737 .017 [.594, .834] [.305, .917] 12,254
Donor role identity 15 5826 .503 .004 [.408, .588] [.255, .689] .632 .006 [.528, .717] [.342, .812] 8744
Descriptive norm 9 3067 .174 .001 [.102, .245] [.052, .292] .203 .002 [.118, .285] [.052, .345] 262
Prosocial Motivation
Personal moral norm 24 9718 .394 .003 [.300, .481] [.084, .635] .464 .004 [.355, .559] [.093, .721] 19,242
Altruism 7 2292 .221 .004 [.104, .332] [.033, .394] .264 .005 [.137, .382] [.057, .449] 332
Affective Expectations
Regret (for not donating) 7 3675 .549 .009 [.409, .664] [.303, .727] .613 .011 [.467, .726] [.350, .786] 3017
Anxiety 6 1585 .229 .016 [.447, .015] [.551, .153] .267 .021 [.510, .016] [.623, .180] 148
Satisfaction (with self) 7 2119 .185 .002 [.091, .276] [.041, .322] .225 .004 [.111, .333] [.043, .392] 241
Collection Center Experiences
Satisfaction with service experience 3 3113 .290 .004 [.177, .395] [.173, .398] .365 .003 [.263, .459] [.262, .460] 267
Service quality 4 1456 .304 .001 [.250, .356] [.273, .335] .325 .001 [.255, .391] [.260, .386] 159
Adverse reactions 5 1861 .287 .001 [.328, .245] [.287, .287] .318 .001 [.373, .261] [.370, .264] 251
Donation Experiences
Number of past donations 11 2553 .279 .002 [.190, .364] [.106, .436] .317 .003 [.222, .405] [.128, .483] 702
Knowledge about blood donation 4 1638 .050 .001 [.001, .098] [.050, .050] .057 .001 [.009, .105] [.057, .057] 0
Demographics
Education level 4 1145 .049 .001 [.120, .023] [.102, .005] .056 .002 [.150, .039] [.152, .042] 0
Marital status 4 1145 .029 .004 [.095, .152] [.114, .171] .043 .006 [.105, .189] [.134, .218] 0
Gender 4 1284 .043 .007 [.209, .126] [.245, .163] .043 .008 [.212, .129] [.249, .167] 0
Age 7 2051 .018 .002 [.075, .110] [.121, .156] .017 .003 [.092, .125] [.153, .186] 0

Note. k refers to the number of samples; N is the total number of people across samples, robs refers to the weighted average of the observed Pearson correlation coefficients
across samples; rc refers to the weighted average of the correlation coefficient corrected for study artifacts; SDobs and SDc refer to the standard deviation of robs and rc
respectively. Effect sizes that are significantly different from zero are indicated by a confidence interval that does not include zero. Fail Safe N scores refer to the Rosenberg
(2005) t  Nþ value.

experienced donors make more realistic assessments of their
ability to donate and are more strongly influenced by their self-
evaluations (Gist & Mitchell, 1992). Conversely, role identity was
less influential for experienced donors. It is possible that experi-
enced donors become less influenced by identity and conscious
decision-making as donation becomes more ingrained as a habit
(Ferguson, 1996). Descriptive norms were significantly related to
donation intentions (but not behavior), and this relationship was
stronger for non-donors. Having never donated, such individuals
may be strongly influenced by the opinions and donation experi-
ences of their friends and family (Rivis & Sheeran, 2003).
With regard to prosocial motivation, personal moral norm
influenced both behavior and intentions, and the latter relationship
was stronger for non-donors. This finding may mean that new
donors feel relief from an obligation to donate after their initial
donation. Alternatively, participation in a survey about blood
donation may remind non-donors of their non-participation,
prompting a sense of obligation. Altruism was related to donation
intentions, but its association with behavior was non-significant.
We propose three explanations for this non-significant finding
grounded in measurement issues. First, the content of the items is
subject to a strong social desirability bias, possibly resulting in at-
tempts at impression management by participants. Second, the
variance of the altruism scores is small. For example, Bove and
Bednall (2009) observed a mean altruism score of 6.2 on a 7-
point Likert scale, with a standard deviation of 0.82. Third, studies
differed in how they operationalized “altruism”. The following
items used to measure altruism illustrate this point (parentheses
depict the alternate construct the item could represent). For
example: “If I gave blood I would be fulfilling my duty to society”
(personal moral norm); “Voluntarily helping someone else is very
rewarding” (anticipated satisfaction with self); “I feel proud when I
know that my generosity has benefited a needy person” (antici-
pated pride). Thus this non-significant result should be interpreted
with caution, as it may be attributed to the use of measures of
altruism that are inconsistent in meaning.
Affective expectations influenced both donation behavior and
intentions. Both anticipated regret and satisfaction were positively
associated with donation behavior and intentions, with regret
being the stronger predictor. This pattern is consistent with the-
ories of loss aversion (Tversky & Kahneman, 1991), which posit
that potential losses have a greater impact on preferences than
gains. Consistent with this explanation, regret was more influen-
tial for experienced donors who would be more aware of the
benefits they would sacrifice if they missed opportunities to
donate. Anxiety about donating had a negative impact on both
behavior and intentions, which was consistent for both donors and
non-donors.
Among the collection site experiences, temporary deferrals and
adverse events had the strongest negative impact. Although these
findings have generally been supported (Custer et al., 2011;
Hillgrove et al., 2011; Masser et al., 2009), our study additionally
revealed that temporary deferrals had less impact on experienced
donors. Both satisfaction and perceived service quality showed
medium positive associations with donation intentions, although
only the former was significantly related to behavior. The absence
of effects for service quality was surprising, but may be explained
by the probable asymmetric and nonlinear relationship between
attribute-level performance (service quality) and dependent con-
structs such as overall satisfaction and repatronage intentions/
behavior (cf. Mittal, Ross, & Baldasare, 1998). That is, one unit of
negative performance on an attribute such as waiting time is likely
to have a greater effect on subsequent donation behavior than a
corresponding unit of positive performance.
92 T.C. Bednall et al. / Social Science & Medicine 96 (2013) 86e94
Of the remaining research programs, knowledge about donation
appears only useful in predicting intentions, whereas the number
of past donations predicts both behavior and intentions, suggesting
the benefit of habit formation (Ferguson, 1996). Age and being
married were positively associated with behavior, possibly because
of their correlation with donation experience (Jun, 1994).
Overall most antecedents showed a stronger relationship with
intentions than behavior. One possibility is that some antecedents
(e.g., altruism) prompt contemplation about donating, but do not
actually trigger behavior. Alternatively, common-method and
simultaneous measurement of each antecedent and intentions may
have produced inflated correlations (Podsakoff, MacKenzie, &
Podsakoff, 2012). Thus, the large correlations with intentions
should be interpreted cautiously.
Limitations and future directions
It is recognized that this meta-analysis is limited by the
quality of available data, differences in the operationalization of
each factor, and the potential for publication bias. In particular,
the majority of studies came from relatively few countries and
many relied on university student samples. Thus, our pooled
sample is potentially unrepresentative of the worldwide donor
population. We attempted to address these issues by using a
comprehensive strategy for collecting studies, applying stringent
selection criteria, correcting the measures for unreliability, and
testing the robustness of each effect using the failsafe method.
Notwithstanding these issues, our meta-analysis provides a
comprehensive synthesis of the key influences on donation
behavior and intentions based on all available evidence to date. It
has revealed new insights not present in the original studies or
previous reviews and identified under-researched possibilities
for future exploration.
In our analysis, we were unable to test a multivariate causal
model of blood donation behavior, as most of the studies reported
no empirical links to studies from other research streams. For
example, no study has investigated the impact of temporary de-
ferrals on the TPB factors. The links between these divergent
streams should be investigated in future studies, as this approach
would account for the interrelatedness of the antecedents and
allow a holistic model of donor motivation to be tested.
The results have additional implications for future research.
First, identifying the strongest predictors of donation behavior (i.e.,
self-efficacy, perceived behavioral control, attitude towards dona-
tion, anticipated regret, and temporary deferral) has highlighted
the constructs that offer the most fruitful opportunities for
recruiting and retaining donors. Interventions designed to target
these factors should be a focus of managerial practice. Second, this
synthesis has identified gaps in the existing literature that require
further investigation. For example, while it is plausible that inex-
perienced and non-donors may be deterred from donating because
of inadequate knowledge about blood donation, only a small
number of studies have investigated this hypothesis with in-
tentions to donate (e.g., Lemmens et al., 2005), or have compared
the knowledge of donors versus non-donors (Chliaoutakis, Trakas,
Socrataki, Lemonidou, & Papaioannoue, 1994). No study has
investigated the direct effect of knowledge or learning on subse-
quent donation behavior.
Other areas that require future investigation based on their
theoretical justification and lack of empirical prevalence, include
perceptions of the need for blood, reciprocity-related motives, and
the influence of collection agency reputation (Bednall & Bove,
2011). In addition, although research has examined the donation
intentions of non-donors, there have been no predictive correla-
tional studies that have investigated antecedents of a person’s
initial donation. With few exceptions (e.g., Masser, Bednall, White,
& Terry, 2012), donor motivation changes have not been tracked
over time. Given the influence of some factors (e.g., donation in-
tentions) has been shown to change over time, more longitudinal
studies of donors are needed. Finally, most of the samples included
in this meta-analysis were drawn from the general population.
Further research involving ethnic minorities is warranted, given
that they often report additional barriers to donating, such as
perceived social exclusion and mistrust of the healthcare system
(Polonsky, Brijnath, & Renzaho, 2011).
Implications and conclusion
The findings of this meta-analysis have important implications
for BCAs. In particular, interventions designed to build donor self-
efficacy, perceived behavioral control, a positive attitude towards
donation and a desirable donor identity will improve donation
behavior and intentions. Making regret more salient is also likely to
be effective (especially for experienced donors), although such a
strategy must be carefully managed by BCAs. Regret could be
evoked by focusing on potential negative outcomes of inaction (i.e.,
fewer people will be helped). However, given the voluntary nature
of donation, a more appropriate strategy may be to use a benevo-
lent message focused on the personal benefits of donating
(Ferguson et al., 2008). This message may be framed in terms of
loss; that is, missing out on positive emotions from donating, such
as happiness, pride, satisfaction or the composite “warm glow”. In
line with this idea, BCAs can encourage individuals to imagine the
positive feelings they would experience post-donation, which has
been shown to increase positive-outcome behaviors (Sandberg &
Conner, 2008).
Given their negative impact, it is important for BCAs to combat
the effects of adverse reactions and temporary deferrals, especially
for early career donors. Reducing the likelihood of adverse re-
actions could be achieved by focusing recruitment efforts on
healthy populations, training staff to be vigilant to the first signs of
an episode, using more experienced collection staff to handle first
time donors, and encouraging non-problematic donors to return.
With respect to temporary deferrals, although they are necessary to
protect the health of donors and blood product recipients, a pro-
active approach to donor re-recruitment can weaken the damaging
effects of deferrals on subsequent donation behavior. Indeed, in
addition to promptly contacting donors following their period of
deferral, social media provides an opportunity to keep donors
engaged with the BCA during the period of deferral (Briones, Kuch,
Liu, & Jin, 2011).
In summary, this meta-analysis has provided a systematic re-
view of six programs of research into blood donor motivation, and
has revealed insights about the factors that most strongly drive
donation behavior and intentions. It has also identified some of the
conditions that moderate such relationships. These findings pro-
vide researchers with insights to build more informed and robust
models of donation behavior for empirical investigation. The meta-
analysis also provides BCAs with strategic guidance for effective
investment of scarce resources. It is hoped that this review will
stimulate further research into blood donation and improved donor
management practices, ultimately leading to a safe, stable supply of
blood products across the world.
Acknowledgments
We would like to acknowledge Australian governments that
fully fund the Australian Red Cross Blood Service for the provi-
sion of blood products and services to the Australian community.
This study was also supported by a Business and Economics
 T.C. Bednall et al. / Social Science & Medicine 96 (2013) 86e94
Faculty research grant from the University of Melbourne. The
authors would like to acknowledge Jo En Yap for her assistance in
collecting research papers for this synthesis. We would also like
to formally acknowledge the Australian Red Cross Blood Service
library staff, Tara Bansal, Anna Stamatelos, and Janette Smith for
their assistance in the collection of secondary data. We extend
our appreciation to all of the blood researchers who responded to
our call for studies.
Appendix A. Supplementary material
Supplementary material related to this article can be found at
http://dx.doi.org/10.1016/j.socscimed.2013.07.022.
References
Adams, V., Erwin, K., & Le, P. V. (2009). Public health works: blood donation in
urban China. Social Science & Medicine, 68(3), 410e418.
Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human
Decision Processes, 50(2), 179e211.
Batson, C. D., Ahmad, N., & Tsang, J.-A. (2002). Four motives for community
involvement. Journal of Social Issues, 58(3), 429e445.
Bednall, T. C., & Bove, L. L. (2011). Donating blood: a meta-analytic review of self-
reported motivators and deterrents. Transfusion Medicine Reviews, 25(4), 317e
334.
Borenstein, M. (2009). Effect sizes for continuous data. In H. Cooper, L. V. Hedges, &
J. C. Valentine (Eds.), The handbook of research synthesis and meta-analysis (2nd
ed.). (pp. 221e235) New York: Russell Sage Foundation.
Bove, L., & Bednall, T. C. (2009). Pilot donor satisfaction survey. Melbourne, Australia:
The Australian Red Cross Blood Service.
Briones, R. L., Kuch, B., Liu, B. F., & Jin, Y. (2011). Keeping up with the digital age: how
the American Red Cross uses social media to build relationships. Public Relations
Review, 37(1), 37e43.
Callero, P. L., Howard, J. A., & Piliavin, J. A. (1987). Helping behavior as role behavior:
disclosing social structure and history in the analysis of prosocial action. Social
Psychology Quarterly, 50(3), 247e256.
Chliaoutakis, J., Trakas, D. J., Socrataki, F., Lemonidou, C., & Papaioannoue, D. (1994).
Blood donor behaviour in Greece: implications for health policy. Social Science
& Medicine, 38(10), 1461e1467.
Cohen, J. (1992). A power primer. Psychological Bulletin, 112(1), 155e159.
Conner, M., & Armitage, C. J. (1998). Extending the theory of planned behaviour: a
review and avenues for further research. British Journal of Social Psychology,
28(15), 1429e1464.
Custer, B., Chinn, A., Hirschler, N. V., Busch, M. P., & Murphy, E. L. (2007). The
consequences of temporary deferral on future whole blood donation. Trans-
fusion, 47(8), 1514e1523.
Custer, B., Schlumpf, K. S., Wright, D., Simon, T. L., Wilkinson, S., & Ness, P. M. (2011).
Donor return after temporary deferral. Transfusion, 51(6), 1e9.
Delage, G., & Dubac, S. (2007, June). Comparison of return rates of temporarily
deferred donors with those of undeferred donors. In Paper presented at the 17th
regional congress of the international society of blood transfusion, Madrid, Spain.
Ferguson, E. (1996). Predictors of future behaviour: a review of the psychological
literature on blood donation. British Journal of Health Psychology, 1(4), 287e308.
Ferguson, E., & Bibby, P. A. (2002). Predicting future blood donor returns: past
behavior, intentions, and observer effects. Health Psychology, 21(5), 513e518.
Ferguson, E., Farrell, K., & Lawrence, C. (2008). Blood donation is an act of benev-
olence rather than altruism. Health Psychology, 27(3), 327e336.
Ferguson, E., France, C. R., Abraham, C., Ditto, B., & Sheeran, P. (2007). Improving
blood donor recruitment and retention: integrating theoretical advances from
social and behavioral science research agendas. Transfusion, 47(11), 1999e2010.
France, C. R., Ditto, B., France, J. L., & Himawan, L. K. (2008). Psychometric properties
of the Blood Donation Reactions Inventory: a subjective measure of presyncopal
reactions to blood donation. Transfusion, 48(9), 1820e1826.
Gillespie, T. W., & Hillyer, C. D. (2002). Blood donors and factors impacting the blood
donation decision. Transfusion Medicine Reviews, 16(2), 115e130.
Gist, M. E., & Mitchell, T. R. (1992). Self-efficacy: a theoretical analysis of its de-
terminants and malleability. Academy of Management Review, 17(2), 183e211.
Godin, G., Sheeran, P., Conner, M., Germain, M., Blondeau, D., Gagné, C., et al. (2005).
Factors explaining the intention to give blood among the general population.
Vox Sanguinis, 89(3), 140e149.
Godin, G., Vézina-Im, L. A., Bélanger-Gravel, A., & Amireault, S. (2011). Efficacy of
interventions promoting blood donation: a systematic review. Transfusion
Medicine Reviews, 26(3), 224e237.
Hedges, L. V. (2009). Statistical considerations. In H. Cooper, L. V. Hedges, &
J. C. Valentine (Eds.), The handbook of research synthesis and meta-analysis (2nd
ed.). (pp. 37e47) New York: Russell Sage Foundation.
Hillgrove, T., Moore, V., Doherty, K., & Ryan, P. (2011). The impact of temporary
deferral due to low hemoglobin: future return, time to return, and frequency of
subsequent donation. Transfusion, 51(3), 539e547.
93
James, V., Hewitt, P. E., & Barbera, J. A. J. (1999). How understanding donor
behavior should shape donor selection. Transfusion Medicine Reviews, 13(1),
49e64.
Jun, S. (1994). Cultural differences in altruistic action: Comparing Korean, American,
and Polish blood donors (Unpublished doctoral dissertation). Wisconsin: Uni-
versity of Wisconsin-Madison.
Lemmens, K. P. H., Abraham, C., Hoekstra, T., Ruiter, R. A. C., De Kort, W., Brug, J.,
et al. (2005). Why don’t young people volunteer to give blood? An investigation
of the correlates of donation intentions among young nondonors. Transfusion,
45(6), 945e955.
Masser, B. M., Bednall, T. C., White, K. M., & Terry, D. J. (2012). Predicting the
retention of first-time donors using an extended Theory of Planned Behavior.
Transfusion, 52(6), 1303e1310.
Masser, B. M., White, K. M., Hyde, M. K., & Terry, D. J. (2008). The psychology of
blood donation: current research and future directions. Transfusion Medicine
Reviews, 22(3), 215e233.
Masser, B. M., White, K. M., Hyde, M. K., Terry, D. J., & Robinson, N. G. (2009).
Predicting blood donation intentions and behavior among Australian blood
donors: testing an extended Theory of Planned Behavior model. Transfusion,
49(2), 320e329.
Misje, A. H., Bosnes, V., & Heier, H. E. (2010). Gender differences in presentation
rates, deferrals and return behaviour among Norwegian blood donors. Vox
Sanguinis, 98(3p1), e241ee248.
Mittal, V., Ross, W. T., & Baldasare, P. M. (1998). The asymmetric impact of negative
and positive attribute level performance on overall satisfaction and repurchase
intentions. Journal of Marketing, 62(1), 33e47.
Newman, B. (2001). Blood donor suitability and allogeneic whole blood donation.
Transfusion Medicine Reviews, 15(3), 234e244.
Oswalt, R. M. (1977). A review of blood donor motivation and recruitment. Trans-
fusion, 17(2), 123e135.
Piliavin, J. A. (1990). Why do they give the gift of life? A review of research on blood
donors since 1977. Transfusion, 30(5), 444e459.
Piliavin, J. A., & Callero, P. L. (1991). Giving blood: The development of an altruistic
identity. Baltimore: Johns Hopkins University Press.
Piliavin, J. A., Howard, J. A., & Callero, P. L. (1991). Is the road to helping paved with
good intentions? Or inertia? In J. A. Howard, & P. L. Callero (Eds.), The self-society
dynamic: Cognition, emotion, and action (pp. 259e279) New York: Cambridge
University Press.
Podsakoff, P. M., MacKenzie, S. B., & Podsakoff, N. P. (2012). Sources of method bias
in social science research and recommendations on how to control it. Annual
Review of Psychology, 63, 539e569.
Polonsky, M. J., Brijnath, B., & Renzaho, A. M. N. (2011). “They don’t want our blood”:
social inclusion and blood donation among African migrants in Australia. Social
Science & Medicine, 73(2), 336e342.
Rivis, A., & Sheeran, P. (2003). Descriptive norms as an additional predictor in the
theory of planned behaviour: a meta-analysis. Current Psychology, 22(3), 218e
233.
Robinson, N., Masser, B. M., White, K. M., Hyde, M. K., & Terry, D. J. (2008). Predicting
intentions to donate blood among nondonors in Australia: an extended Theory
of Planned Behavior. Transfusion, 48(12), 2559e2567.
Rosenberg, M. S. (2005). The file-drawer problem revisited: a general weighted
method for calculating fail-safe numbers in meta-analysis. Evolution, 59(2),
464e468.
Rothstein, H. R., & Hopewell, S. (2009). Grey literature. In H. Cooper, L. V. Hedges, &
J. C. Valentine (Eds.), The handbook of research synthesis and meta-analysis (2nd
ed.). (pp. 103e128) New York: Russell Sage Foundation.
Sandberg, T., & Conner, M. (2008). Anticipated regret as an additional predictor in
the theory of planned behaviour: a meta-analysis. British Journal of Social Psy-
chology, 47(4), 589e606.
Schmidt, F. L., In-Sue, O., & Hayes, T. L. (2009). Fixed- versus random-effects models
in meta-analysis: model properties and an empirical comparison of differences
in results. British Journal of Mathematical and Statistical Psychology, 62(1), 97e
128.
Schmidt, F. L., Le, H., & Oh, I.-S. (2009). Correcting for the distorting effects of study
artifacts in meta-analysis. In H. Cooper, L. V. Hedges, & J. C. Valentine (Eds.), The
handbook of research synthesis and meta-analysis (2nd ed.). (pp. 317e333) New
York: Russell Sage Foundation.
Schwartz, S. H. (1977). Normative influences on altruism. In L. Berkowitz (Ed.),
Advances in experimental social psychology (Vol. 10; pp. 221e279). New York:
Academic Press.
Shadish, W. R., & Haddock, C. K. (2009). Combining estimates of effect size. In
H. Cooper, L. V. Hedges, & J. C. Valentine (Eds.), The handbook of research syn-
thesis and meta-analysis (2nd ed.). (pp. 257e277) New York: Russell Sage
Foundation.
Shaz, B. H., James, A. B., Hillyer, K. L., Schreiber, G. B., & Hillyer, C. D. (2011). De-
mographic patterns of blood donors and donations in a large metropolitan area.
Journal of the National Medical Association, 103(4), 351e357.
Sheeran, P. (2002). Intentionebehavior relations: a conceptual and empirical re-
view. European Review of Social Psychology, 12(1), 1e36.
Stryker, S. (1980). Symbolic interactionism: A social structural version. Palo Alto:
Benjamin Cummings.
Sutton, S. (2006). Predicting and explaining intentions and behavior: how well are
we doing? Journal of Applied Social Psychology, 28(15), 1317e1338.
Tversky, A., & Kahneman, D. (1991). Loss aversion in riskless choice: a reference-
dependent model. The Quarterly Journal of Economics, 106(4), 1039e1061.
94 T.C. Bednall et al. / Social Science & Medicine 96 (2013) 86e94

Veldhuizen, I. J., Doggen, C. J., Atsma, F., & De Kort, W. L. (2009). Donor profiles:
demographic factors and their influence on the donor career. Vox Sanguinis,
97(2), 129e138.
Walsh, R. J., & Clemens, K. (1956). A survey of some voluntary blood donors in New
South Wales. Medical Journal of Australia, 43(6), 205e209.
Whitener, E. M. (1990). Confusion of confidence intervals and credibility intervals in
meta-analysis. Journal of Applied Psychology, 75(3), 315e321.
Wilson, D. B. (2005). Meta-analysis macros for SAS, SPSS, and Stata (Version
2005.05.23) [Computer software]. Thousand Oaks, CA: Sage Publications.
Retrieved from http://mason.gmu.edu/wdwilsonb/ma.html.