Understanding the Correlates of Donor Intention: A Comparison of Local, National, and International Charity Destinations

The United Kingdom is generous toward charitable donations, and this commitment appears robust against a background of economic uncertainty. While prior work has identified a clear preference for domestic over international causes, research has yet to identify the range of variables that significantly correlate with this important element of charitable choice. A survey of 1,004 U.K. residents was designed to assess willingness to donate to local, national, and international causes. For each destination, stepwise multiple regression analysis identified the key variables that correlate to an individual’s willingness to donate. Findings suggest that donor willingness correlates with levels of trust, preferred types of charitable cause, and donation channels. In contrast, the role of donor demographics is relatively limited. The findings suggest some commonality in the variables that associate most significantly with willingness to donate locally and nationally, but those relating to international donation intention are relatively distinct.


Introduction
The increasingly competitive landscape faced by charities is widely acknowledged (e.g., Ein-Gar & Levontin, 2013;O'Hara, 2014). In the last decade, the United Kingdom has experienced the global economic recession followed by economic austerity, with many of its citizens being subject to wage freezes or subinflation salary increases. Such trends put charities under greater pressure to understand not just why people donate (see Bekkers & Wiepking, 2011a), but also how donors choose between the ever-increasing numbers of alternatives. Work focusing on what correlates with donations to certain types of charity is surprisingly limited (Bennett, 2003).
A common technique used by donors to segment charities is to distinguish between home and overseas causes (Breeze, 2013). Although national level charities may also provide local services (Hall et al., 2013), there are calls to distinguish between causes which are local, national, and international in nature (Grau & Folse, 2007;Varadarajan & Menon, 1988). In addition to social distance (which refers to the physical and emotional distance between donors and recipients: Strombach et al., 2014), choosing which charity to support is further complicated by the plethora of causes actively seeking donations, ranging from medical research through to animal welfare, poverty alleviation, and environmental projects.
The current study provides a comprehensive analysis of the correlates of donation intention to local, national, and international charities (we term this donation destination). This builds upon recent calls for a greater understanding of how donors choose between charities based upon destination (Hart, 2016) and previous work on domestic versus international giving (Casale & Baumann, 2015;Knowles & Sullivan, 2017;Micklewright & Schnepf, 2009). Existing research provides an understanding of why donors support charitable causes (with reasons straddling personal values and experiences, faith, sense of moral obligation, and warm-glow effects: Ottoni-Wilhelm et al., 2017), but research on preferred donation destination is largely lacking (a recent exception being Knowles & Sullivan, 2017). This study investigates whether the specific correlates of donation intention (proven to play a significant role in actual donation behavior: Kashif et al., 2015) differ by local, national, or international destination.
The study draws from the psychological literature on the theory of planned behavior (TPB: Ajzen, 1991) and social identity theory (SIT). These two theoretical perspectives have previously been brought together to understand health behaviors (Chatzisarantis et al., 2009), recycling (Terry et al., 1999), and sustainable agriculture engagement (Fielding et al., 2008). TPB has been found to predict prosocial behaviors, be it more traditional forms of charitable giving like financial donations (Smith & McSweeney, 2007) or volunteering (Warburton & Terry, 2000). In the current study, we will assess an individual's donation intentions for local, national, and international charities. Previous TPB studies have consistently indicated that donation intentions are powerful predictors of actual donations (France et al., 2007;Smith & McSweeney, 2007).
As donations to these three categories of charities allude to issues of group membership, this study also contributes to our broader understanding of SIT in a charitable context. First developed by Tajfel (1974), SIT considers how an individual identifies based upon group membership, be it friendship, sports team affiliation, or nationality. SIT relates to issues such as prejudice, ethnocentrism, and discrimination (M. A. Hogg, 2006), all of which are potentially relevant to donating to beneficiaries in different geographic locations. Our social identities refer to issues of "us" and "them" (Fielding et al., 2008) that are determined by two processes: social categorization (where boundaries between groups are established) and self-enhancement, where norms are shaped to benefit in-group members. These social identities result in groups wishing to minimize in-group differences and maximize intergroup differences (Terry et al., 1999). In a charitable context, this would result in donors prioritizing charities that aid fellow in-group members (i.e., local and national charities). Of course, the distinction between charities which assist in-groups versus out-groups is complicated by the fact that many serve both (Erlandsson et al., 2019).
Research suggests higher levels of trust and support for domestic causes (Casale & Baumann, 2015;Charity Commission, 2016). We extend the consideration of trust by assessing to what extent trust for specific destinations correlates with donation intention. Equally, the demography of donors represents a core driver of giving. There is evidence that those with higher education and income are more likely to support overseas causes (Bennett, 2003). Further work in this area may aid fundraisers in the effective targeting of donors.
The final two variables considered represent areas of charitable giving notably underrepresented in research, the type of charitable causes supported, and the use of specific donation channels. We argue that the types of charitable cause a donor supports will correlate with their donation destination (for example, those who support charities for ex-military personnel may prefer domestic causes). Equally, certain donation channels (e.g., entering raffles) rely on localized community networks (Schlegelmilch et al., 1997).
The article will next introduce the extant knowledge on donation destination before providing a review of literature covering the variable sets introduced above. The article outlines a quantitative methodology, which leads to the development of separate regression models for local, national, and international charities. The conclusions will summarize the core findings and discuss implications for fundraisers.

Donation Destination
We use "donation destination" to describe the location of the recipients of charitable donations relative to the donor, and categorize these as local, national, or international. The literature suggests that donors typically display a preference for more local causes in line with the principles of SIT (Tajfel, 1974). Focus group research across the United Kingdom and Australasia suggests that donors consider local causes more relevant than causes further afield irrespective of seriousness (Dalton et al., 2008), partially as they represent causes that may be utilized by the donor in the future (Hall et al., 2013). National level charities are often preferred because of a moral obligation to tend to the needs of conationals (Stevenson & Manning, 2010). This links closely to Kessler and Milkman's (2018) investigation into donor identity. Across two experiments, they concluded that charity appeals which center on a donor as a member of a local community generate greater donations. Micklewright and Schnepf (2009) interrogated the Office for National Statistics (ONS) data to uncover that while international causes often receive higher individual donations, these tend to be less frequent than domestic donations. Knowles and Sullivan (2017) provide evidence for preference for national over international causes with data from New Zealand, with 71.6% opting for domestic alternatives. The opposite emerges from an Australian perspective, where Lwin et al. (2014) used survey data to conclude that donors have more positive attitudes toward national and international charities than local alternatives. Based on this literature, the underlying premise of SIT, and the view of Bekkers (2010, p. 370) that "people will be more strongly attracted to collective goods in the local community than to the problems of a third world country," hypothesis H1 suggests the following:

Hypothesis 1 (H1):
There is a greater level of donation intention toward U.K.based charities, either locally or nationally focussed, compared with donation intention toward those operating internationally.

Donor Demographics
Donor demography encapsulates numerous variables associated with donor intention. For age, various perspectives emerge. Knowles and Sullivan (2017) indicate no significant association, in contrast to Lwin et al. (2014) who indicate a positive association between age and donation intention. Bekkers and Wiepking (2011b) suggest the association is nonlinear, with donations starting to decline in donors aged 65 years and older.
There is an equally varied picture relating to gender. Micklewright andSchnepf (2009), Lwin et al. (2014), and Knowles and Sullivan (2017) all reported no statistically significant differences in domestic versus international preferences between gender groups. Interestingly (and after accounting for earnings, educational attainment, and household composition), Piper and Schnepf (2008) used the same ONS data set as Micklewright and Schnepf (2009) and concluded women are more generous in both financial contribution and frequency of donation, with particular dominance in causes relating to animal welfare, education, and the elderly.
Educational participation and attainment are important demographics in understanding donation behavior. Bekkers and Wiepking (2011b) indicate greater donation intention among the more highly educated, driven by enhanced information access and confidence-based trust in charitable organizations. Bennett (2003) suggests less-educated donors demonstrate greater affinity with domestic concerns, with Micklewright and Schnepf (2009) recognizing that higher educational attainment resonates with international giving. Income also plays a significant role in charitable giving, correlating with higher levels of donation (Lwin et al., 2014).
Supposition exists that political attitudes may partly explain charitable giving, especially toward international causes (Rajan et al., 2009). For example, those with a left-wing political orientation are more predisposed to supporting international causes (Wiepking, 2010) and specifically international disaster relief (Manesi et al., 2019), with the opposite true of Conservative donors (a statistically significant finding from Chapman et al., 2018). Based on this collective body of evidence, hypothesis H2 proposes the following: Hypothesis 2 (H2): Donor demographics correlate with donation intentions toward support for the three charitable destinations.

Trust
Trust can be considered both at a sector (where charity regulation breeds confidence: E. Hogg, 2018) and also at an individual charity level. Bekkers (2003) notes that trust (often assessed through formal accreditations or testimonies) correlates significantly with giving. In their Dutch-U.S. comparative study, Beldad et al. (2014) identified that donor affinity, cause reputation, and donor trust combine to explain repeat donation behavior. Similarly, Naskrent and Siebelt (2011) employed structural equation modeling to identify trust and commitment as the two strongest predictors of donor retention.
A cross-national study conducted by Nfp Synergy (2019) found that fewer people trust overseas aid charities than domestic causes, and this distinction was particularly acute in the United Kingdom (where only 36% trusted international causes). As trust levels have been found to be critical to future donation intentions (Charity Commission, 2018), this may explain the general preference for domestic causes demonstrated thus far. Collating the above arguments, hypothesis H3 proposes the following: Hypothesis 3 (H3): Levels of trust in local, national, and international charities correlate positively with donation intention toward the three respective charitable destinations.

Charitable Choice
Research focusing on which charities people choose to donate to remains limited (Wiepking, 2010). Good causes range from small-scale local charities through to global projects (Daly, 1997); however, donors tend not to share their generosity equally across all causes (Strombach et al., 2014). U.K. donors display preferences for charities in the fields of medical research, animal welfare, and children, while religious organizations are one of the most popular categories across North America (Charities Aid Foundation [CAF], 2017). Bennett (2003) has previously underlined the critical role of personal experience in charitable choice. The experiences of close family or friends may result in support for relevant causes (referred to as "friends of victims" by Small & Simonsohn, 2007).
The current study uses 13 charitable categories that were adapted from typologies used by the CAF (2014) and Mintel (2012). While some categories align to local or national level interests (e.g., military and local development), others clearly have a more global reach (e.g., international disaster relief). It follows then that those charitable causes preferred by individuals will relate to their inclination to support local, national, and international charities. Therefore, H4 proposes the following: Hypothesis 4 (H4): Particular charitable choices have positive correlations with donation intentions toward the three respective charitable destinations.

Donation Channel
An area underassessed within charitable giving research is the preferred means of donation. Donors face numerous channels ranging from traditional cash donations through to direct debits, mobile giving, and engagement with charity retail stores (Shier & Handy, 2012). In the United Kingdom, cash donations, donating to charity stores, and buying raffle tickets are the most preferred channels, whereas online and mobile forms of giving are most common in North America (CAF, 2017). Peloza and Hassay (2007) developed a typology of charity support behavior that distinguished between high and low involvement forms of support. Citizenship behaviors such as volunteering represent highest involvement owing to the necessary time commitment, with donating to charity stores and buying raffle tickets being examples of lower involvement behaviors that also brought personal benefits to the donor.
Donation channel research tends to focus on one specific channel rather than investigating donor preferences across channels. For example, sponsorship of individuals to take part in charity events has become a notable growth area, with one fifth of all Canadian charitable donations originating from event sponsorship (Higgins & Lauzon, 2003). Charity store donations are a common means of giving (e.g., Hibbert et al., 2005). However, these are distinct from other channels as the donor arguably benefits from the act of either donating products (removing clutter) or buying from charity stores. Finally, the internet is a particularly attractive channel for charities because of its cost-effectiveness (Shier & Handy, 2012) and viral capability (as evidenced through successful campaigns such as the amyotrophic lateral sclerosis [ALS] Ice Bucket Challenge: Pressgrove et al., 2018). Recent work from Herzog and Yang (2018) demonstrated that having contacts on social media who engage in prosocial actions (either giving to charity or asking others to do so) increases donation intention.
We argue here that the donation channels preferred by individuals will correlate to some extent with their destination preferences. Donation channels such as sponsoring a friend or buying raffles tickets involve either face-to-face contact or have focus in local community institutions (Schlegelmilch et al., 1997), and as such are likely to associate with interest in causes that serve local beneficiaries. Digital forms of giving and direct debits are not constrained by the same geographic boundaries, are less likely to be utilized by smaller charities (Shier & Handy, 2012), and may correlate with more national and international level giving. Therefore, H5 proposes the following: Hypothesis 5 (H5): Particular donation channel access positively correlates with donation intentions toward the three respective charitable destinations.

Differences in Variable Sets Displaying Association With Donation Destination
There are certain variables that associate more strongly with enhancing international donation intention. These include trust, political beliefs, and exposure through travel to developing countries. Certain demographic characteristics also correlate with international preferences (Bennett, 2003;Micklewright & Schnepf, 2009). Combining this evidence with the suite of variables examined in this study, hypothesis H6 proposes the following: Hypothesis 6 (H6): The ranges of measures relating to donor demographics, trust, charitable choice, and donation channel differ in their associations with donation intentions toward local, national, and international charities.

Study Design
An online survey captured data on donor intention by charitable destination, trust perceptions, charitable choice, and channels of charitable donation. The survey instrument first addressed various demographic variables (including age, gender, geographic location, education, income, and voting behavior). The instrument then utilized a battery of items addressing charitable giving, trust levels, preferred causes, and donation channels, utilizing a combination of 7-point scales and multiple-choice questions (Table 1). A pilot survey with 112 participants helped to refine the instrument.
Data collection took place between March and April 2017 utilizing a consumer panel accessed through the market researcher Pickersgill Consultancy and Planning. Respondents were required to be aged 18 years and older and be resident in the United Kingdom (England, Northern Ireland, Scotland, and Wales). They were not required to be active donors as the study sought to generate a representative sample. According to the CAF, 60% of the U.K. population donated money to charity in 2018, with a further 59% donating products and 35% sponsoring a charitable activity. Our approach allowed us to capture data from donors and nondonors, following the premise that fundraisers may be equally interested in the donation intentions of those who do not currently support charitable causes.
The sampling frame included individuals who have previously signed up to take part in online surveys via consumer panels deployed by Pickersgill Consultancy and Planning. Emailing consumer panel members achieved agreed quotas, with 1,141 responses received, 137 of which being rejected through incompletion, missing data, or straight-lining (Johnson, 2016). To ensure respondents were considering items fully, a time check for completion was undertaken. The pilot survey indicated the average British Annual income band: less than £10K, £10-20K, £20-30K, £30-40K, £40-50K, £50-75K, £75-100K, more than £100K, Prefer not to say Note. EU = European Union; RSPCA = Royal Society for the Prevention of Cruelty to Animals; NSPCC = National Society for the Prevention of Cruelty to Children; NVQ = National Vocational Qualification. completion time was 6 to 7 min. To ensure data validity, we removed responses from individuals who completed the survey in less than 5 min. In return for full survey completion, panel members receive points redeemable for shopping vouchers.

Data Analysis
Three separate assessments of the measures assessing local, national, and international donation intention involved using ordinary least squares (OLS) regression analysis. There are three respective statements to assess willingness to donate to these respective destinations. For example, "I am likely to donate to a charity that helps my local community in the next month," assesses local donation intention. For the three separate models, the dependent variable is the relevant 7-point individual Likert-type scale.
The decision to utilize donation intention as the dependent variables in each multiple regression model reflected two considerations. First, we were concerned with recall accuracy. Respondents indicated their aggregate donations to all charities across the previous 3 months; this longer time period was utilized to minimize the effects of any seasonal fluctuations in donation patterns as identified by the CAF (2018). The down side of such a timeline is difficulty for respondents in accurately recalling the exact charities supported. We felt this was particularly acute for ad hoc lower involvement forms of donation such as street collections. Second, if respondents could recollect the charities they supported, we were concerned they may struggle to suitably categorize these as either local, national, or international in scope (in particular, as national level charities often provide services at a local level: Hall et al., 2013). Therefore, asking respondents to indicate future intentions across these three categories, rather than relying on potentially erroneous historical behavior, appears a more robust approach to grouping future donations by destination.
There is a general acceptance of the value of donor intention as a predictor of actual donations (Kashif et al., 2015), evidenced with financial donors in the United Kingdom (Smith & McSweeney, 2007), mainland Europe (Verhaert & Van den Poel, 2011), and blood donors in the United States (France et al., 2007). Lee et al. (1999) had earlier demonstrated that intention was a powerful predictor of future behaviors spanning the three major forms of giving (money, time, and blood).
For all the three multiple regression models, the potential predictor variables consisted of trust, charitable choice, donation channels, and donor demographics. The first three variable groups were measured on a 7-point scale from 1 = very unlikely through 4 = neither unlikely nor unlikely to 7 = very likely. Multiple-choice demographic questions were prepared for the respective multiple regression models by converting them into appropriate (1, 0) dummy variables. The respective numbers of dummy variables for each were gender (1), age (7), geographic location in the United Kingdom (12), voting behavior at the June 2016 European Union (EU) referendum (4-including did not vote and preferred not say), level of qualifications (9), social class by employment role (7), ethnicity (4), and annual income band (9).
There are potentially 77 independent variables covering trust, charitable choice, donation channel, and demographics presented in an appropriate dummy variable form. To assist in developing a suite of parsimonious regression models and limiting the potential for multicollinearity, a stepwise process of variable selection was adopted. The forward method of stepwise was actioned starting with no independent variables, with sequential variable entry, and based on correlation with the dependent measure donation intention and partial correlation thereafter until further variable addition ceases to improve the module in a statistically significant way. The assessment of each model considers the overall model significance using the analysis of variance (ANOVA) test, model fit by adjusted R 2 , and a residual analysis. For all the three models, issues of multicollinearity involve examination of the variance inflation factor (VIF) for each retained independent variable. Guided by Berenson et al. (2002), variables with a VIF exceeding 5 were removed.

Sample Overview
The sample comprises 1,004 U.K. respondents, demonstrating some resonance with the wider U.K. population (Table 2). In summary, 51.7% of respondents were female and 92.0% reported their ethnicity as White, with the most commonly read national newspapers being the Mail, the Sun, and the Mirror. In the 2016 referendum, 51.2% voted to leave the EU. The breakdown by age band is 18 to 24 (8.6%), 25 to 34 (16.5%), 35 to 44 (16.7%), 45 to 54 (18.9%), 55 to 64 (15.6%), and 65 and older (23.6%; of which 3.4% of the total data set was aged 75 years or older). Gender and age band are representative of the wider population data (ONS, 2017), as is referendum voting declaration. The profile based on ethnicity represents an underrepresentation of participants from the Black and Minority Ethnic groupings, while there is some overrepresentation of Scotland, Wales, and Northern Ireland in the sample. Around half of the sample indicated earnings between £10,001 and 30,000 per annum, which is in line with wider economic data.
More than 80% of the sample reported donating to charity within 3 months of data collection, the majority of these supporting two or three charities. The most common donation amounts in the time period were £11 to 20 (17.5%), £6 to 10 (14.2%), £1 to 5 (13.9%), and £21 to 30 (13.7%). The most common charitable causes supported by the sample were health, children's, and animal causes. The most common forms of assisting charities were donating to/buying from charity stores, cash donations, and sponsorship, aligning closely with CAF (2018) giving report.

Assessment of Donor Intentions
The means for donation intention to local, national, and international concerns are 4.36, 4.58, and 3.61, respectively (Table 3). The first two statistics are significantly greater in value than the midpoint of 4.0, the converse being the case for the item assessing international donation intention (for each, p < .001). For the pairwise assessment of donation intention, significant differences between the pairs of donation destination were statistically significant (each p < .001). The strongest level of donation intention relates to country-level alternatives, followed by local charities, which in turn are significantly more likely to receive donations than international charities.

Regression Models by Donation Destination
Local charities. The stepwise multiple regression model developed to explain local donation intention comprises 16 predictor variables being statistically significant in combination (Table 4).
There is a significant correlation for each of the predictor variable groups identified (trust, charitable choice, donation channels, and donor demographics). The predictor variables indicate the multiple role of trust. This points to positive correlation with trust in local causes x 1 (b = 0.26, t = 7.59, p < .001) and trust in national causes x 4 (b = 0.21, t = 6.12,  p < .001), but a negative association with trust in international causes x 5 (b = −0.11, t = −3.80, p < .001). Trust developed for causes close to home has a positive association on local intentions compounded by a lack of trust for causes based more remotely.
Various charitable causes also contribute to local donation intention. These comprise local development charities x 2 (b = 0.24, t = 7.76, p < .001) and education training charities x 6 (b = 0.16, t = 5.00, p < .001). There is a negative correlation with = 53.90, p < .001. Adjusted R 2 = 45.8%, standard error of the estimate = 1.12 Y = willingness to donate to local causes x 1 = Trust local charities to use donations wisely x 2 = Donate to local development charities x 3 = Buy raffle tickets/enter competitions x 4 = Trust national charities to use donations wisely x 5 = Trust international charities to use donations wisely x 6 = Donate to education and training charities x 7 = Donate to health charities x 8 = 55 to 64 years old (1 = yes, 0 = no) x 9 = Donating items to charity x 10 = Donate to international disaster relief charities x 11 = Salary deductions via employer x 12 = West Midlands (1 = yes, 0 = no) x 13 = Wales (1 = yes, 0 = no) x 14 = Donate to environmental charities x 15 = Skilled manual workers and equivalent (1 = yes, 0 = no) x 16 = Uncertain about employment status (1 = yes, 0 = no) each of the health charities x 7 (b = −0.10, t = −3.42, p = .001), international disaster relief charities x 10 (b = −0.06, t = −2.19, p = .029), and environmental charities x 14 (b = −0.06, t = −2.23, p = .026). This shows some intuitive resonance with commitment to local issues through local development and education, with a more negative perception of concerns further afield.
The model is statistically significant, F (16, 987) = 53.90, p < .001. The level of fit is moderate with an adjusted R 2 value of 45.8%, albeit based on a large data set. Further analysis of the model's residuals shows no departure from normality, constant variance, and randomness. Only 10 cases recorded high-value standardized residuals, outside of the range ±3 (<1% of the sample). In terms of assessing multicollinearity, none of the 16 independent variables introduced into the multiple regression model have a VIF value above 5 (values range from 1.02 to 2.42- Table 4) and are therefore retained within the model. National charities. The second model developed to explain national donation intention comprises nine predictor variables that are statistically significant in combination (Table 5). Consistent with local donation intentions presented above, there is a role to play for each of trust, charitable choice, donation channels, and donor demographics.
The multiple role of trust mirrors that presented in the explanation of local donation intention. There is a positive correlation with trust in national causes x 1 (b = 0.30, t = 8.72, p < .001) and trust in local causes x 4 (b = 0.22, t = 6.74, p < .001), but a negative correlation with trust in international concerns x 6 (b = −0.15, t = −5.74, p < .001). Various charitable causes also contribute positively to national donation intention. These include armed forces and emergency services charities x 3 (b = 0.09, t = 3.45, p = .001), education and training charities x 7 (b = 0.07, t = 2.80, p = .005), and health charities x 9 (b = 0.07, t = 2.47, p = .014), all of which appear nationally focussed. Two donation channels also correlate positively with increasing national donation intention; these are donating items to charity x 2 (b = 0.14, t = 5.63, p < .001) and direct debit x 5 (b = 0.08, t = 4.81, p < .001), the former again being a channel with potentially high levels of visibility in the donor's immediate locality.
In contrast, associations with demographics is limited, based only on respondents located in the West Midlands x 8 (b = 0.47, t = 2.98, p = .003).
The model for national donation intention is statistically significant, F (9, 994) = 97.76, p < .001. Like the model for local intention, the level of fit is moderate with an adjusted R 2 value of 46.5%, with assessment of the model's residuals again showing no concerns around normality, constant variance, and randomness, with nine cases recording high-value standardized residuals outside of the range ±3 (<1% of the sample). Multicollinearity is of no concern; none of the nine independent variables have a VIF value above 5 (values range from 1.01 to 2.16- Table 5).
International charities. This final model comprises 13 predictor variables that are statistically significant in combination (Table 6). Trust, charitable choice, channels of donation, and demographics again combine to correlate with attitudes toward international donation destination, although there is a more distinct suite of individual predictor variables identified here compared with the local and national alternatives.
The only dimension of trust acting as a significant correlate with international donation intention is that involving international causes x 2 (b = 0.31, t = 12.48, p < .001), with no significant association in either direction for trust in local or national equivalents. Charitable choice also has a greater combined role to play. In a positive sense, these comprise international charities x 1 (b = 0.16, t = 4.93, p < .001), international disaster relief charities x 3 (b = 0.23, t = 7.47, p < .001), religious charities x 6 (b = 0.09, t = 3.95, p < .001), and social services charities x 11 (b = 0.08, t = 2.55, <.014 2.16 F (9, 994) = 97.76, p < .001. Adjusted R 2 = 46.5%, standard error of the estimate = 1.11 Y = willingness to donate to national causes x 1 = Trust national charities to use donations wisely x 2 = Donating items to charity x 3 = Donate to armed forces and emergency services charities x 4 = Trust local charities to use donations wisely x 5 = Donate using direct debit x 6 = Trust international charities to use donations wisely x 7 = Donate to education and training charities x 8 = West Midlands (1 = yes, 0 = no) x 9 = Donate to health charities x 16 = Uncertain about employment status (1 = yes, 0 = no) p = .011), three of which have an explicit international remit. There is a negative association involving armed forces and emergency services charities x 5 (b = −0.11, t = −4.76, p < .001), which have a much stronger national focus.
Donation channels make multiple contributions to explaining international donation intention. There is a positive correlation with salary deductions x 4 (b = 0.06, t = 2.43, p = .015), direct debit x 7 (b = 0.07, t = 3.68, p < .001), and donation via mobile text message x 10 (b = 0.06, t = 2.82, p = .005). Contrasting is the negative association with the more immediate and face-to-face channel of sponsoring a friend or relative x 12 (b = −0.05, t = −2.20, p = .028). In terms of personal characteristics, willingness to donate internationally correlates more positively with those donors earning in excess of £100,000 per year x 8 (b = 0.99, t = 3.27, p = .001). The opposite  (13,990) = 138.25, p < .001. Adjusted R 2 = 64.0%, standard error of the estimate = 1.07 Y = willingness to donate to international causes x 1 = Donate to international charities x 2 = Trust international charities to use donations wisely x 3 = Donate to international disaster relief charities x 4 = Salary deductions via employer x 5 = Donate to armed forces and emergency services charities x 6 = Donate to religious charities x 7 = Donate using direct debit x 8 = Annual income of £100,001+ (1 = yes, 0 = no) x 9 = Vote leave in the EU referendum (1 = yes, 0 = no) x 10 = Donate via mobile text message/online x 11 = Donate to social services charities x 12 = Sponsoring a friend/relative in an event x 13 = Aged 75+ years old (1 = yes, 0 = no) is true for leave voters in the EU referendum x 9 (b = −0.20, t = −2.77, p = .006) and for those in the age range 75 years and above x 13 (b = −0.39, t = −2.02, p = .044).
The model of international donation intention is statistically significant, F (13, 990) = 138.25, p < .001. In comparison with the previous two models, the level of fit is better with an adjusted R 2 value of 64.0%, with no concerns around normality, constant variance, and randomness emerging from the residual analysis, with eight cases recording high-value standardized residuals, outside of the range ±3 (similar to Models 1 and 2). Multicollinearity is again unproblematic; the VIF values for the 13 independent variables range from 1.02 to 3.22 (Table 6), leading to variable retention.

Discussion
The relative preference for national causes mirrors prior work in this setting (Micklewright & Schnepf, 2009) and supports hypothesis H1. The additional contribution made here is the delineation between national and local alternatives, with respondents seemingly more positive toward donating to national level causes (which we acknowledge may also provide services on a local level).
The association between donor demographics on willingness to donate to the three distinct destinations is limited. The lack of gender association has some support in literature (Einholf, 2011;Knowles & Sullivan, 2017;Lwin et al., 2014), although the absence of qualification contrasts with previous research (Bekkers & Wiepking, 2011a;Micklewright & Schnepf, 2009;Neumayr & Handy, 2019). Those with higher incomes displayed higher donation intention for international causes, which contradicts recent work from Neumayr and Handy (2019) who instead concluded that income was positively associated with domestic giving. As indicated by existing literature (e.g., Chapman et al., 2018), older respondents and donors with more right-wing political views appear less likely to support international causes. With these limited associations identified, there is only partial support for hypothesis H2.
Donor trust in both local and domestic causes correlates positively with donation intentions toward local and national charities, with donors reporting lower levels of trust in international alternatives. Willingness to donate internationally correlates positively with trust in international causes. Therefore, one or more of the assessments of trust is significantly associated with intention for each donation destination, supporting hypothesis H3. The positive role of trust accords with various previous studies (Bekkers, 2003;Naskrent & Siebelt, 2011).
Charity choice is associated with donation destination in an intuitive manner. Local donation willingness correlates with increased likelihood to donate to local development charities; nationally donation willingness associates positively with greater attachment to armed forces and emergency services charities, while international donation intention increases with support for international disaster relief and religious charities. While many of the associations are arguably unsurprising, the data clearly demonstrates that the role of charity choice correlates positively with donation intention across all three destinations, supporting hypothesis H4.
Donation intention by destination also associates with preferred channels of donation. Donating items to a charity store and buying raffle tickets from family or friends have a positive correlation with donating to local charities. For national charities, donating items has a role, as does the opportunity to set up a direct debit. This visibility also correlates with willingness to donate to international concerns, with salary deductions, direct debit, and donation via mobile device, all having a positive marginal association. With the various donation channels offering significant associations across the three destinations, this supports hypothesis H5.
The willingness to donate to local and national charities shares various common significant associations across the variable sets considered. Willingness to donate to both destinations correlates positively with trust in both local and national concerns wisely, but for both, they are less likely to trust international alternatives. Like all destinations, the donation is independent of gender, donor qualifications, or ethnicity. In this study, it is worth remembering that 92% of the study participants belong to a single ethnic group (a clear sampling limitation), offering no opportunity to differences by ethnic groups.
A distinct set of variables correlates significantly with willingness to donate to international charities. While trust locally and nationally plays no significant part, willingness to donate outside the United Kingdom correlates significantly with trust in international charities to use the donations wisely. Those supporting international disaster relief, social services, and religious charities are more likely to donate internationally, with the same individuals less likely to support armed forces and emergency charities. These donors are more likely to favor the technological/banking channels of donation and exhibit distinct demographics relating to higher income, being anti-Brexit, and being relatively younger. Combining these findings, there is evidence to conclude that the suite of variables differ according to donation destination, thereby supporting hypothesis H6. Table 7 summarizes the similarities and differences between the variables that correlate significantly with donation willingness by destination.

Theoretical and Managerial Implications
Donation destination intention appears only marginally associated with donor demographics, with gender, ethnicity, and qualifications playing no part whatsoever. Age and social class (defined by income and employment category) play some role, with the younger and more affluent tending to be more international in their donation focus. Such findings suggest that charities may wish to limit their dependence on classic demographic data as a means of identifying potential donors. Voting behavior in the EU membership referendum suggests that political voting data may provide a useful alternative means of targeting potential donors. The accessibility of charities to donors' voting behaviors and record is unlikely, but there are indirect means to targeting (or avoiding) potential donors based on such measures (e.g., through specific newspapers or targeting certain geographic regions).
Respondents appear loosely segmented into two groups; those willing to support local or national level causes and those with a predisposition toward international concerns. This aligns with the principles of SIT, whereby individuals possess an inherent desire to minimize inequality between group members and will subsequently support charitable (i.e., domestic) causes that enable this. Previous research has indicated that out-groups are typically less trusted (Tanis & Postmes, 2005) which aligns with the trust levels reported for international charities in this study. SIT recognizes the role of power and status in intergroup relations and suggests that members of a group with greater power will act to maintain the status quo (Fielding et al., 2008).
Education and training causes appear to resonate locally, in contrast to environmental causes and international disaster relief. Nationally, education and training have a positive role, as do initiatives focusing on the armed forces and emergency services. Those who report intention to support international causes typically identify preferences for disaster relief, social services, and religious causes (but also national options relating to social services). In summary, a donor's preference for donation destination correlates significantly with the type of charitable causes they opt to support. The findings suggest that donation channels may be associated with donation preferences. Face-to-face channels tend to appeal more to those respondents favoring local and national level causes, while more remote forms of giving such as mobile applications relate more to international charities. The role played by trust is central to donation intention across all destinations. Both local and national charities can leverage higher levels of public trust in their future fundraising efforts and may benefit from the fact that many people hold a more cynical view toward international charities. Conversely, those individuals with higher trust in international causes are in turn more likely to support them independent of whether they trust local or national charities.

Limitations and Future Research
While we are content that the sample is broadly representative of the U.K. population for particular measures, this study group may not reflect donor preferences in other parts of the world. More comparative research across nations (particularly those characterized by varying levels of nationalistic and patriotic tendencies) would facilitate a more global picture of preferred donation destination. One specific sampling limitation identified was the representation of donors from Black and Minority Ethnic groups (8.0% of the sample presented compared with 14.0% of the U.K. population, ONS, 2017). Given the increased mobility of various populations, there exists a need to understand the giving patterns of migrant populations and their attitudes to supporting causes based in their home and host countries.
Notwithstanding our prior justification of basing our multiple regression models on future donation intentions, we acknowledge that collecting data on past giving patterns would add further nuance to the research area. In this study, our rationale for opting for intentions accounted for accurate recall and categorization of donations by respondents over a 3-month period. Future work may wish to address this issue by capturing intentions and behaviors via a more longitudinal format. Research which builds upon Fajardo et al. (2018) work, which calls for a distinction between charitable choice and amount donated, also appears worthwhile, given previous assertions that international charities receive fewer but higher value donations (Micklewright & Schnepf, 2009). It would also be beneficial to add context to the findings through qualitative work, particularly among those predisposed to local and national concerns who self-report greater nationalistic tendencies. Such research may add further insights from those who support domestic over international causes and will aid fundraisers in producing appropriate campaign messaging.
The key message emerging from this study is that an individual's intention to support local, national, and international causes is significantly associated with a range of issues spanning trust, charitable type, and donation channel. The finding that demographic variables largely fail to correlate with preferred destination donations highlights the need for further work to help fundraisers truly understand how donors feel about charitable causes in different parts of the world.

Author Biographies
Andrew Robson is a professor in Business Modeling in the Department of Marketing, Operations and Systems at Northumbria University. His teaching focuses on business modeling and associated quantitative methods, with research interests spanning across marketing, pedagogy, and management.

David John
Hart is an associate professor in Marketing in the Department of Marketing, Operations and Systems at Northumbria University. His primary research interests are charitable marketing, customer loyalty, and customer complaint behavior. LinkedIn: https://uk.linkedin. com/in/david-hart-phd-08a47433.