Fair weather avoidance: unpacking the costs and benefits of “Avoiding the Ask”

  • Hannah Trachtman
  • Andrew Steinkruger
  • Mackenzie Wood
  • Adam Wooster
  • James Andreoni
  • James J. Murphy
  • Justin M. Rao
Original Paper


If being asked to give to charity stimulates an emotional response, like empathy, that makes giving difficult to resist, a natural self-control mechanism might be to avoid being asked in the first place. We replicate a result from a field experiment that points to the role of empathy in giving. We conduct an experiment in a large superstore in which we solicit donations to charity and randomly allow shoppers the opportunity to avoid solicitation by using the other door. We find the rate of avoidance by store entrants to be 8.9 %. However, we also find that the avoidance effect disappears in very cold weather, suggesting that avoidance behavior is sensitive to its cost.


Altruism Charity Empathy Giving Social pressure Avoiding the Ask Field experiment Experimental economics 

JEL Classification

D03 D64 H41 

1 Introduction

Why do people give to charity? Research has made it clear that pure altruism cannot fully explain many of the giving behaviors that we see, and thus it must be that people derive some positive utility from the act of giving (Andreoni 1989, 1990). In a recent study, “Avoiding the Ask: A Field Experiment on Altruism, Empathy, and Charitable Giving”, Andreoni et al. (2011, henceforth ART) explored a particular channel by which giving might affect utility: the emotional pressure that people experience when asked to give. Psychologists suggest that a request to donate triggers an emotional response, such as empathy, that makes it difficult for people to resist giving (Batson 1991, 2011; Preston and deWaal 2002; Andreoni and Rao 2011; DellaVigna et al. 2012). Avoiding being asked, and thereby evading the emotional response that induces giving, can serve as an important self-control mechanism.

Several recent laboratory studies also find that people do incur costs in order to avoid opportunities to give (Dana et al. 2006; Broberg et al. 2007; Lazear et al. 2012), which appears consistent with the psychological mechanism described above. In a field experiment incorporated into a door-to-door fundraising campaign, DellaVigna et al. (2012) find that when agents are given opportunities to avoid via fliers specifying the time of solicitation, avoidance (not answering the door) increases by between 9 and 25 %.

ART ran a field experiment in partnership with the Salvation Army’s familiar annual Red Kettle campaign. At a supermarket in Boston with three doors, they randomly varied whether solicitors are placed at two doors or at just a single door. ART also varied whether solicitation is active (verbal and with eye contact) or passive (simply ringing the bell) to uncover potential interactions between social pressure and sorting.1 ART find that 32.6 % of patrons avoided active solicitation by using another door when they had the opportunity to do so, suggesting that patrons were using the defense mechanism of avoiding being asked.

We ran a similar experiment that both replicates and expands upon this finding. We solicited donations outside of a supermarket in Anchorage, Alaska with two entrance doors, and randomly varied whether a solicitor was present at the door located on the right side of the parking lot. By chance, we were also able to investigate the effect of increasing the cost of avoidance: one of the 4 days in our sample had extremely cold temperatures, so taking a longer route to avoid solicitation was presumably less comfortable. We find that on average, adding a solicitor to the right door results in a 6.8 % decrease in traffic through the right door. However, this effect depends on the cost of avoidance. When we restrict our analysis to the days with normal temperatures, avoidance jumps up to 8.9 %, and if we look only at the day that was extremely cold, avoidance is negligible.

2 Experiment design

To test the robustness of ART’s results, our experiment was designed to contrast with theirs on several dimensions. ART used a well-known charity, posted college-age females at the supermarket doors, solicitors asked for donations directly, and conducted the study in the eastern United States. In contrast, this study chose a relatively unfamiliar charity, the Polycystic Kidney Disease (PKD) Foundation. Our solicitors were college-age males. Rather than asking for unspecified donations, we sold pins, shown in Fig. 1, for $1.00. Finally, our study was conducted in Anchorage, Alaska.
Fig. 1

Pin-back button for PKD Foundation

We conducted the experiment over 4 days in November 2013. We divided each of the 4 days into three one-hour time blocks: 12:00–1:00, 1:05–2:05, and 2:10–3:10 p.m. Finally, we assigned each of the 12 total blocks to be control or treatment, making sure that the treatment was distributed as evenly as possible across days and time blocks. The exact schedule is depicted in Table 1. The layout of the supermarket and its two doors is depicted in Fig. 2.
Table 1

Experiment schedule


Saturday 11/9

Thursday 11/14

Saturday 11/16

Wednesday 11/20

Block 1: 12:00–1:00 p.m.

Control (639)

Treatment (388)

Treatment (678)

Control (524)

Block 2: 1:05–2:05 p.m.

Treatment (628)

Control (412)

Control (685)

Treatment (680)

Block 3: 2:10–3:10 p.m.

Control (716)

Treatment (467)

Treatment (863)

Control (596)

Number of observations in parentheses

In blocks assigned to control, there was no solicitation.2 In blocks assigned to treatment, we placed a male solicitor at the right door who actively solicited sales of the PKD Foundation pins.3 He stood by a small table that had an informative poster about the PKD Foundation and a collection box for proceeds. In all blocks, two experimenters sat in vehicles and counted the number of customers entering the store. Each experimenter was assigned to monitor traffic through a single door.4 Finally, we only counted incoming shoppers, and ignored outgoing shoppers.
Fig. 2

Supermarket layout

Fig. 3

Mean percentage of passings through right door by condition and weather. Bars give 1.96 standard errors from identical regressions restricted to normal days and the cold day, respectively (for the cold day, standard errors are not clustered by block since only one block received the treatment). Treatment indicates fund-raisers were placed at the right door of the store, and Cold indicates an exceptionally cold day

3 Results

Over the 4 days, we observed 7276 entrances through both doors. During the first three days of the study, temperatures were within the normal range for Anchorage, between 20 and 30 \(^{\circ }\)F (\(-\)7 to \(-\)\(^{\circ }\)C). However, on the fourth day, November 20, temperatures were unseasonably cold, with the daytime temperature of 0 \(^{\circ }\)F (\(-\)18 \(^{\circ }\)C) approaching a new record for that date. The cold weather presumably reduced the appeal of walking to the less convenient door, allowing us to test whether the avoidance effect is sensitive to its cost. As shown in Fig. 3, on the three normal days combined, 46.2 % \((n=1397)\) of customers entered through the right entrance in the treatment, relative to 51.2 % \((n=1256)\) in the control. On the fourth day, the extremely cold temperature had a significant impact on entrances through the right door, yielding no discernible difference between the treatment (50.9 %, \(n=346\)) and control (51.6 %, \(n=578\)). The nonparametric Fisher exact test confirms that avoidance is statistically significant on the days with normal temperatures \((p=0.000)\), but not on the cold day \((p=0.771)\). A nonparametric chi-square test yields the same conclusions.5

To test whether the lack of avoidance on the cold day is due to the time of day that the treatment was implemented, and not temperature, we compare the outcomes on Days 1 and 4, which follow the same sequence. On Day 1, with normal temperatures, 47.3 % \((n=297)\) of customers entered through the right entrance in the treatment, compared to 52.2 % \((n=707)\) in the control. This difference is statistically significant \((p=0.048)\), which supports the hypothesis that the lack of avoidance on Day 4 is due to temperature differences, and not the time of day.

We use a probit regression to confirm these findings, and report marginal effects in Table 2.6 Column (1) shows that adding the solicitor to the right door reduces the number of entrances through the right door by 3.5 percentage points, amounting to a 6.8 % decrease. When we control for cold and its interaction with the treatment in Column (2), the effect jumps to 4.5 percentage points, or an 8.9 % decrease in traffic through the right door. Both results are significant at the 1 % level. Moreover, cold weather dampens the effect of the treatment, increasing the number of entrances through the right door by 4.2 percentage points. This positive effect is significant at the 1 % level. Thus, the extremely cold weather essentially neutralized the effect of the treatment on avoidance.
Table 2

Probability of choosing the right door: marginal effects of a probit regression















Treatment \(\times \) Cold








Mean of control



Date and time fixed effects



Treatment indicates fund-raisers were placed at right door of the store, and Cold indicates an exceptionally cold day. Standard errors clustered by block in parentheses

*** p \(<\) 0.01, ** p \(<\) 0.05, * p \(<\) 0.10

4 Conclusion

We conducted an experiment that was similar to (Andreoni et al. 2011), but which also differed in many key respects. We solicited money for a relatively unknown charity, while they used the Salvation Army. Our solicitors were male while theirs were female. We asked for $1 in exchange for lapel buttons rather than an unspecified donation. We ran our study in Anchorage, Alaska, while they ran theirs in Boston. The fact that both studies found evidence of avoidance is notable. In addition to the above differences in context, there were also two key differences in experiment design, which can likely explain the large differences in observed magnitudes (8.9 % avoidance in our study, relative to 32.6 % avoidance in ART). First, the ART study site had three entrances, one of which was conveniently located next to the entrance that always had a solicitor. The addition of a more convenient, less costly escape route would be expected to generate more avoidance, and the fact that most of the observed avoidance in ART occurred through this door is consistent with the large difference in magnitudes. Second, ART counted outgoing shoppers as well as incoming shoppers, and therefore were able to detect any additional avoidance behavior that occurred on the way out of the store.

In addition to replicating the ART avoidance result in a new context, our findings also shed some light on the cost-benefit calculation that underlies avoidance. As ART hypothesize, if giving only increases utility by resolving the feeling of empathy experienced when asked, then there is a major benefit to avoiding feeling empathy in the first place, and thereby avoiding the guilt or giving that results. ART make the argument (but do not verify) that empathy is the most likely explanation behind the avoidance effects they see. Our study does not provide any new evidence regarding the mechanism, but it does strengthen the case for avoidance as an important phenomenon. Thus, insofar as we are convinced by the ART argument that empathy is the most likely mechanism for avoidance, this study gives additional credence to the empathy story. Both experiments indicate that the benefits of avoidance are big enough under most circumstances to induce changes in behavior. But the fact that avoidance behavior is also sensitive to its cost triggers further questions about the precise trade-offs faced by individuals, providing important direction for future research.


  1. 1.

    While this paper is most similar to ART, it can also be seen as a more indirect confirmation of the finding of significant avoidance by DellaVigna et al. (2012).

  2. 2.

    In the ART experiment, blocks assigned to control had solicitors at both doors, but there is no reason why the distribution of traffic should differ between the two.

  3. 3.

    Since ART did not see avoidance for passive solicitation, we chose to focus our efforts only on active solicitation. Following the procedures in ART, the solicitor attempted to make eye contact with each customer and said “Please give today”.

  4. 4.

    To minimize counting errors, future research in this area may want to consider having multiple people assigned to each entrance.

  5. 5.

    Since there was only a single day with extremely cold temperatures, it is possible that some other factor unique to that particular day could explain behavior.

  6. 6.

    A linear probability model gives almost exactly the same results.


  1. Andreoni, J. (1989). Giving with impure altruism: applications to charity and ricardian equivalence. Journal of Political Economy, 97, 1447–1458.CrossRefGoogle Scholar
  2. Andreoni, J. (1990). Impure altruism and donations to public goods: a theory of warm-glow giving. Economic Journal, 100, 464–477.CrossRefGoogle Scholar
  3. Andreoni, J., & Rao, J. M. (2011). The power of the ask: how communication affects selfishness, empathy and altruism. Journal of Public Economics, 95, 513–520.CrossRefGoogle Scholar
  4. Andreoni, J., Rao, J. M., & Trachtman, H. (2011). Avoiding the ask: a field experiment on altruism, empathy, and charitable giving. NBER Working Paper Series 17648.Google Scholar
  5. Batson, C. (1991). The altruism question: toward a social psychological answer. New Jersey: Lawrence Erlbaum Assoc Inc.Google Scholar
  6. Batson, C. (2011). Altruism in humans. Oxford: Oxford University Press.Google Scholar
  7. Broberg, T., Ellingsen, T., & Johannesson, M. (2007). Is generosity involuntary? Economics Letters, 94(1), 32–37.CrossRefGoogle Scholar
  8. Dana, J., Cain, D., & Dawes, R. (2006). What you don’t know won’t hurt me: costly (but quiet) exit in dictator games. Organizational Behavior and Human Decision Processes, 100(2), 193–201.CrossRefGoogle Scholar
  9. DellaVigna, S., List, J., & Malmendier, U. (2012). Testing for altruism and social pressure in charitable giving. Quarterly Journal of Economics, 127, 1–56.CrossRefGoogle Scholar
  10. Lazear, E., Malmendier, U., & Weber, R. (2012). Sorting in experiments with application to social preferences. American Economic Journal: Applied Economics, 4, 136–163.Google Scholar
  11. Preston, S. D., & deWaal, F. B. M. (2002). Empathy: its ultimate and proximate bases. Behavior and Brain Sciences, 25, 1–72.Google Scholar

Copyright information

© Economic Science Association 2015

Authors and Affiliations

  • Hannah Trachtman
    • 1
  • Andrew Steinkruger
    • 2
  • Mackenzie Wood
    • 2
  • Adam Wooster
    • 2
  • James Andreoni
    • 3
    • 4
  • James J. Murphy
    • 2
    • 5
    • 6
  • Justin M. Rao
    • 7
  1. 1.Yale UniversityNew HavenUSA
  2. 2.University of Alaska AnchorageAnchorageUSA
  3. 3.University of California San DiegoSan DiegoUSA
  4. 4.NBERCambridgeUSA
  5. 5.Nankai UniversityTianjinChina
  6. 6.Chapman UniversityOrangeUSA
  7. 7.Microsoft ResearchNew YorkUSA

Personalised recommendations