Good Dog: Aspects of Humans' Causal Attributions for a Companion Animal's Social Behavior

D. W. Rajecki 1 and Jeffrey Lee Rasmussen
Indiana University - Purdue University

Clinton R. Sanders
University of Connecticut

Susan J. Modlin and Angela M. Holder
Indiana University - Purdue University

Lay theories or assumptions about nonhuman animal mentality undoubtedly influence relations between people and companion animals. In two experiments respondents gave their impressions of the mental and motivational bases of companion animal social behavior through measures of causal attribution. When gauged against the matched actions of a boy, as in the first experiment, respondents attributed a dog's playing (good behavior) to internal, dispositional factors but a dog's biting (bad behavior) to external, situational factors. A second experiment that focused on a dog's bite revealed clear attributional process on the part of observers. Higher ratings of a dog as the cause of a victim's distress predicted higher ratings of a dog's guilt. Higher ratings that a dog had an excuse predicted stronger recommendations for forgiveness. Individual differences in seeing the actor as a "good dog" systematically predicted judgments of severity of the outcome and recommendations for punishment. Discussion of these attributional findings referred to tolerance for companion animal misbehavior and relinquishment decisions. This article illustrates the utility of causal attribution as a tool for the study of popular conceptions of nonhuman animal mind and behavior.
In mainstream causal attribution theory and research, the goal is to identify cognitive processes by which a human observer will credit the behavior of a human actor either to external environmental pressures or constraints or to the actor's internal personal motivations or goals (Hewstone, 1989; Ross & Fletcher, 1985). To put it another way, to which type of cause--situational or dispositional--can one attribute a specific act? Causal attributions can allow for "judgments of responsibility" for an actor's behavior (Weiner, 1995). By extension, the same issues should be raised in the study of the human experience of other animals. Given the opportunity, would people be inclined to make dispositional or situational causal attributions for the behavior of a companion animal?

Research Review

To date, we know of only a single report on the application of the formal principles and methods of causal attribution to the question of humans' perceptions of the bases of companion animal behavior. Rajecki, Rasmussen, Modlin, and Holder (1998) derived measures from a human model for the attribution of blame, originally proposed by Shaver and Drown (1986), and applied them to the matched misbehavior (biting) of a dog compared with that of a boy. Supporting the general attribution model, results showed that people's impressions of the extent of the dog's and the boy's personal dispositions -- volition, intent -- for the act predicted levels of judgment blame. The nonhuman and human actors, however, were not perceived as equals. The dog, seen as having more excuse and less intent for biting, was judged less guilty and blameworthy than the boy.

These patterns indicate contrasting evaluations of essentially the same act and show that dog versus boy behavior received markedly different interpretations. Hence, measures of causal attribution have the potential to reveal the way people view the mentality or motivation underlying an animal's act. Indeed, our aim was to extend the findings available from the Rajecki et al. (1998) attributional analysis of dog behavior. Although persuasive and instructive, that earlier article had certain limitations.

First, the method was tied to a single attribution model, that of Shaver et al. (1986). Extending the paradigm to other known models would be informative. Second, although the article assessed impressions of blameworthiness and guiltiness regarding misbehavior (biting), no measures were employed concerning the fate of the dog or boy. That is, judgments of guilt and blame are one thing, recommendations for punishment or forgiveness another. In the case of the dog, current concerns about real-life animal abuse or pet relinquishment indicate the importance of measures about the actor's fate. Third, the effects of individual differences based on respondents' general orientations toward dogs were not reported. For one dimension of orientation, ownership versus non-ownership influences perceptions and attitudes about animal issues (Fidler, Light, & Costall, 1996; Moss & Wright, 1987; Wells & Hepper, 1997). Because prior sentiments could influence ongoing evaluations regarding a dog's behavior and consequent recommendations for its fate, individual differences become an important consideration

Accordingly, the current project -- using procedures related to those of Rajecki et al. (1998) -- addressed the three points. First, in Experiment 1 we sought to broaden the understanding of the social cognition of human/companion animal relations by appealing to a different attribution model. We adopted the items used in the current study from the Causal Dimension Scale (Russell, 1982). Second, in Experiment 2 we broadened the method and range of measures to include respondents' recommendations for the fate of the offending dog. These included items about punishment and forgiveness. Third, also in Experiment 2 we approximated individual differences in general orientation toward dogs by asking whether the actor was a "good dog." Based on the polarities of their responses to the that item, individuals were divided into three post-hoc experimental groups


EXPERIMENT 1

Method

Causal Dimension Scale

The nine-item Causal Dimension Scale (Russell, 1982; Russell, McAuley, & Tarico, 1987) was originally developed to assess separately the causal dimensions of internality, controllability, and stability, each dimension represented by three questionnaire items. We chose this model because of its differences with the Shaver et al. (1986) approach and the fact that its original items could be easily reworded to fit our ends.

In the current application, our measures referred to one or the other outcome, play or bite, and actor, dog or boy. Every item was accompanied by a stem: “Is the cause of the outcome (something that is)…” and a seven-interval (1 to 7) scale with appropriate bipolar endpoints. The endpoints for the three dimensions, in the case of the canine actor called Dog B, were as follows.

Internality

1. Inside of Dog B [or] Outside of Dog B
2. Reflects an aspect of Dog B [or] Reflects an aspect of the situation
3. Something about Dog B [or] Something about other things


Controllability

1. Controllable by Dog B [or] Uncontrollable by Dog B
2. Intended by Dog B [or] Unintended by Dog B
3. Someone is responsible [or] No one is responsible


Stability

1. Permanent [or] Temporary
2. Stable over time [or] Variable over time
3. Unchangeable [or] Changeable

In our instrument -- as in the original -- some of the endpoints were reversed, and item order matched the original ( Russell, 1982). The three items per dimension were combined in analysis, and all were scored such that larger values represented estimates of more internality, controllability, and stability.

Rationale for Tests and Scenarios

We never intended to contrast the different variables within the Russell framework. Instead, following the lead of Rajecki et al. (1998), the stated point of the current research was to compare and contrast patterns of attributions for a dog's behavior with attributions concerning matched behavior on the part of a boy. To this end, we asked respondents to express their own estimates of the likelihood or applicability of a dimension drawn from the model. For example (on a 7-point scale), “Is the cause of the outcome something Intended by Dog B [or] Unintended by Dog B?”

For this project, respondents received scenarios in test booklets that gave clear pictures of physical social acts: Dog plays with boy or dog bites boy. However, motivational or dispositional and contextual or situational elements were presented to a minimal extent and their possible impact on the behavior made intentionally ambiguous. We believed that this procedure -- depending on respondents' inclinations, habits, or histories with related actors and acts would provide the latitude to attribute the behavior to internal or external determinants.

We chose the acts of biting and playing for two reasons: The literature offers precedents for the use of these scenarios (Rasmussen & Rajecki, 1995; Rajecki et al., 1998), and both acts are common to dogs and boys. Concerning canine aggression, dog bites occur frequently enough to be viewed as a national health problem (Boenning, Fleisher, & Campos, 1983; Griego, Rosen, Orengo, & Wolf, 1995). And children do bite children. According to one review, "[a]fter the toddler period, the next peak of occurrence is between 11 and 16 years of age" (Leung & Robson, 1992, p. 255).

Test Booklet Construction

The first two pages of test booklets stated that participation was voluntary and anonymity guaranteed. They provided space to indicate gender and instructions on how to use the rating scales.

Scenario and outcomes. The booklet’s third page contained a scenario involving a certain situation and one of two outcomes, as adapted from Rasmussen et al. (1995, p. 121). A certain Dog B or a school-aged Boy B. played with or bit a child, Boy A. The only stated provocation for the play episode or the bite attack was that Boy A had picked up a toy belonging to Dog or Boy B. Next, under the heading of outcome, the booklets providing the play condition offered the following statement: “Dog or Boy B ran up to Boy A and pushed him gently as an invitation to play. Boy A became delighted, dropped the toy, and ran laughing back into his own yard.”

The following statement was provided for the bite condition: “Dog or Boy B flung himself on Boy A and bit him hard, causing a serious tear in the skin. Boy A became terrified, dropped the toy, and ran screaming back into his own yard.”

Items. At the bottom of the scenario, an item read, "I believe the outcome associated with Dog [Boy] B's behavior was . . . ." Responses were made on a 7-point, bipolar scale with endpoints of POSITIVE (GOOD) and NEGATIVE (BAD). These were scored as positive = 7 and negative = 1. Finally, the attribution items were listed beginning on the fourth page.

Participants. Undergraduate students recruited from an urban public university in the Midwestern United States participated. Age, ethnicity, and other demographic characteristics not recorded were considered random variables. Prior to a testing session, dog and boy (play or bite) instruments were alternated in a single stack and handed out in sequence. Thus, respondents in convenience samples were assigned to one or another actor (and act) conditions on an effectively random basis. The test group included about an equal number of men and women. A preliminary data analysis indicated that the gender of respondents did not have a strong influence on ratings. For example, regardless of the actor, women (M = 5.46) tended to rate play less positively than men (M = 5.58), a pattern that also emerged for bite (Ms = 2.08 and 2.46, respectively), but these differences were not statistically reliable. Accordingly, gender was dropped from consideration as a variable. The final experimental design involved four independent cells based on the two actors and the two acts. Sample sizes were 24 or 25 per condition.

Results

Outcome Measure

Table 1 shows the average outcome measures for actors in reference to acts.

--------------------------------------------------------------------------------
Table 1. Dog and boy means for outcome and attribution measures under conditions of play and bite in Experiment 1.

Note: Larger mean values indicate more positivity of the outcome or higher attributional estimates on a scale of 1 to 7. Probability statements reflect simple effects tests for particular pairs of dog versus boy means.

Play Bite
Measure Dog Boy p Dog Boy p
Outcome 6.12 4.92 .01 2.84 1.63 .01
Attribution
Internality 4.29 4.09 ns 3.85 4.94 .02
Controllability 4.39 4.59 ns 4.05 5.42 .01
Stability 2.64 2.49 ns 3.35 3.20 ns
(n) (25) (25) (25) (24)

Note: Larger mean values indicate more positivity of the outcome or higher attributional estimates on a scale of 1 to 7. Probability statements reflect simple effects tests for particular pairs of dog versus boy means.
--------------------------------------------------------------------------------

As a manipulation check, respondents were obviously paying attention to the scenarios and the scales. The general pattern of means is very much in line with the prosocial and antisocial incidents so designed. A two-way analysis of variance showed strong main effects between the play versus bite behaviors, F(1, 95) = 143.15, p < .01, and the dog versus boy actors, F(1, 95) = 19.31, p < .01. The table indicates by parallel simple effects the obvious absence of an interaction between the actors and the acts on this measure, F(1, 95) = 0.00.is Thus the play outcome was seen as reliably more positive for the dog and the bite outcome as less negative for the dog.

Attribution Measures

Internality. A two-way analysis of variance (act X actor) did not yield main effects for either the acts (play/bite) or the actors (dog/boy). Nevertheless, as shown in Table 1, an interaction between act and actor emerged, F(1, 95) = 4.93, p < .03. The source of this interaction is evident from simple effects tests. There was no difference between actors in the case of play but a sharp difference between them under the bite condition. For the biting behavior, the boy received relatively higher internality scores compared with those assigned in the play condition. In contrast, when biting was involved, the dog received relatively lower internality scores compared with those he received for play.

Controllability. Results for controllability ratings were in line with the internality pattern. Although there were no main effects for acts or actors, a significant interaction emerged across those variables, F(1, 95) = 6.23, p < .02. As shown in Table 1, the actors were not rated differently on controllability in the play condition but did differ based on the bite behavior. For biting, the boy received relatively higher controllability scores compared with those assigned in the play condition. In contrast, when biting was involved, the dog received relatively lower controllability scores compared with those received for play.

Stability. A two-way analysis of variance revealed a main effect for the acts only, with generally higher stability ratings under the antisocial compared with the prosocial condition, F(1, 95) = 8.89, p < .01. Table 1 indicates the absence of an interaction between the actors and the acts on this measure by parallel simple non-effects.


Discussion

We concluded that, in comparison with a boy actor, a dog's good behavior was about equally credited to dispositional influences, a dog's bad behavior more likely credited to situational influences. Aspects of our findings, however, warrant clarification.

Internality and Controllability

The finding that, under the bite condition, the cause of the dog's outcome was rated as less internal and controllable might indicate that respondents felt the dog simply acted instinctively. However, an appeal to the instincts of Dog B does not account for the lack of difference in ratings by other respondents of the internality and controllability of the play outcome for both the dog and the boy. Where a prosocial act was involved, people were willing to assign responsibility to the companion animal: ("something about Dog B"; "intended by Dog B") for the outcome. The very ratings of the actors' play outcomes (Dog B = 6.12; Boy B = 4.92) seem to indicate that the dog should be given substantial credit for its act.

Two interpretations might be placed on these stability results. First, the scale endpoints in question ("permanent/temporary"; "stable over time/variable over time"; "unchangeable/changeable") may have proved inapplicable or unclear in connection with the current task. It is known that respondents sometimes feel uncertain about using such scales to express their impressions (Miller, Smith, & Uleman, 1981). Lack of clarity, however, would not account for the direction or reliability of the difference found across the play and bite conditions.

We favor a second possibility. Given that the cause of the boy's bite was generally rated as internal and controllable, the relatively high stability estimate (M = 3.20) indicates that respondents judged dispositional factors would continue to govern the boy's behavior in other times. Further, given that the cause of the dog's bite was generally rated as comparatively external, not under control, the relatively high stability estimate (M = 3.35) indicates that respondents judged situational factors would continue to govern the dog's behavior in other times. In any event, the ratings of the actors' bite outcomes (Dog B = 2.84; Boy B = 1.63) seem to indicate that the dog should be held less accountable for the act.

Dog-Positivity Bias?

Previously, Rajecki et al. (1998) used the behavior of a boy as a standard against which to gauge reactions to the behavior of a dog living in a home with humans. In comparison, they found considerable leniency in humans' judgments of blameworthiness of the bite behavior of the dog and generosity in judgments of the praiseworthiness of the dog's play behavior. That research identified links between judgments of blame and praise for behavior, on the one hand, and causal attribution processes on the other. Results of Experiment 1 support and expand this line of findings.

Using a different attribution model, we documented related forms of positive judgments in human social cognitions concerning the behavior of a dog. Compared with a boy actor and in terms of the dimensions of internality and controllability, people seemed willing to credit a dog for good behavior but unwilling to hold the dog accountable for bad behavior. Taken together, the findings of Rajecki et al. (1998) and Experiment 1 suggest a general inclination to cast the behavior of Dog B in a good light or to give him the benefit of the doubt. This is reminiscent of the phenomenon identified by social psychologists as "person-positivity bias" (Sears, 1983): all else being equal, human perception tends toward positive rather than negative evaluations.

By extension, the pattern emerging in the current literature might well be called a dog-positivity bias. People may be predisposed to assume that any given dog, including Dog B, is a good dog. But even if there is a dog-positivity bias, it is certainly not shared equally by everyone. In an examination of the dog bite data from Experiment 1, for example, we can point to variability in the outcome measure. The mean for this measure in that condition was 2.84 (see Table 1), but note that the standard deviation was 1.15. Obtained scale scores ranged from 6 (positive), through 4 (neutral), to 1 (negative). These values may reflect the individual differences in orientations to dogs.

To learn more about the influences of such individual differences and to extend further the attributional approach to the perception of dog misbehavior, we turn to Experiment 2. Experiment 2 also had provisions for assessing people's recommendations for the fate -- forgiveness or punishment -- of a biting dog.

EXPERIMENT 2

Experiment 2, with merged methodological features from Experiment 1 and from Rajecki et al. (1998), focussed exclusively on Dog B. The point of the experiment was to gain additional information by combining the bite scenario with new scale items such as whether Dog B was likable and a good dog. Other new items included recommendations for forgiveness and punishment. These measures and others could capture attribution processes through an intercorrelation analysis. Further, the measure of good dog had the potential to identify individuals with different orientations toward the dog. Polarities of responses to this specific item were used to identify post-hoc groupings of individuals and to examine differences in their reactions to the scenario.

Method

We again employed the booklets. For this study, however, we presented only the single scenario about Dog B's biting. As the booklets also served as a pilot test of Sanders' (1990) propositions concerning owners' excusing tactics for pet impropriety, we modified the bite scenario. In some booklets the outcome section reads as follows:


Dog B flung himself at Boy A and bit him hard, resulting in a serious tear in the skin. Boy A became terrified, dropped the toy, and ran screaming back into his own yard. Right after the contact involving Dog B and Boy A, two Adults, Mr. B and Mr. A, met at the open gate between yards. Mr. B, the owner of Dog B, said, "My Dog B really loves his new toy. I think he felt that your Boy A was going to take it from him." He walked back to his house.

However, the Dog B bite scenario was apparently an inappropriate vehicle for tests of excusing tactics. Analyses indicated no effects for excuse tactic type, and the presence of owners in the scenario did not appreciably affect the respondents' judgments.

Scale Items

Following the standard outcome measure (Experiment 1), respondents encountered a series of 13 questions, each of which began with the stem, “Is it reasonable to say that…” and concluded as follows:


Dog B was the cause of Boy A's terror and screaming?
Dog B should have been judged guilty of bringing about terror and screaming in Boy A?
Dog B would feel sorry about what happened?
Dog B intended to bring about terror and screaming in Boy A?
Dog B had an excuse for biting Boy A?
Dog B should be punished for what happened?
Dog B had justification for biting Boy A?
Dog B is likable?
Dog B meant to bring about terror and screaming in Boy A?
Dog B should be forgiven for what happened?
Dog B should have been blamed for bringing about terror and screaming in Boy A?
Dog B is a "good dog"?
Boy A's terror and screaming were due to Dog B?”.

Responses were made on 7-point, bipolar scales with endpoints of REASONABLE and UNREASONABLE (reversed for some items). These were scored as reasonable = 7 and unreasonable = 1.

Participants

A total sample of 91 undergraduate participants, again recruited from the Midwestern university (n = 51) and also from a public state university in the northeast (n = 40), included 29 men. Gender, however, was ignored. Demographic characteristics were considered random variables

Results

Outcome Measure

Unlike Experiment 1, the scenario used in Experiment 2 contained characters other than Boy A and Dog B. It is therefore essential to determine if the additional actors influenced responses. The most direct test compared the means for the dog bite outcome rating in the two experiments. For Experiment 1, that mean was 2.84 (see Table 1); for Experiment 2, the mean was 3.18. These two means are not reliably different, t(114) = 0.96, ns, indicating that the presence of Mr. A. and Mr. B. did not have a strong effect on respondents' replies.

Strength of Association

--------------------------------------------------------------------------------

Table 2. Intercorrelations between measures in Experiment 2.

Note: For this table, df = 89
*p < .05
**p < .01
--------------------------------------------------------------------------------

The eight relevant rows in the table are characterized individually.

Outcome. Coefficients in this row indicate that respondents who thought the outcome was relatively positive also tended to rate Dog B as high on excuse/justify and as good/likable. Further, people who rated the outcome more positively tended to give relatively lower ratings of guilt/blame and lower recommendations to punish as well as higher recommendations to forgive.

Cause/due to. Relatively high ratings of Dog B on the cause/due to measure predicted higher ratings of guilt/blame, and recommendations to punish.

Intent/meant. Ratings of intent/meant correlated with none of the other variables.

Excuse/justify. People who judged that Dog B was high on excuse/justify tended to rate him more highly on good/likable, relatively lower on guilt/blame, and gave lower recommendations to punish.

Good/likable. Respondents who saw Dog B as relatively good/likable, estimating that the dog would feel sorry, gave relatively lower ratings on the guilt/blame and punish scales. Also, people who rated him high on good/likable gave high recommendations to forgive.

Sorry. As noted above, ratings of sorry correlated positively with the variable of good/likable only.

Guilt/blame. Ratings of guilt/blame were positively related to recommendations to punish and negatively related to forgive.

Punish. Recommendations to punish and to forgive were negatively related. Indeed, the strength and extent of associations between certain variables shown in Table 2 suggest an alternative analysis. Another way to describe the relationship of variables in this attributional approach is to frame the data in terms of the respondents' individual differences.

"Good Dog?": Individual Differences Analysis

We suggested earlier that even if there were a dog-positivity bias on the part of many people, not all individuals would share that viewpoint. Using the obtained scale scores from item 12, "Is Dog B a good dog?" in the preceding list of scales is one way to look for the effects of this proposed distinction. Based on the polarity of their responses to this item, individuals were assigned to one of three post-hoc groups. Those who chose the unreasonable-to-say scale intervals (1, 2, and 3) were assigned to the first group (n = 15). Persons who selected the scale mathematical midpoint (4) were assigned to the second group (n = 46), and persons who chose the reasonable-to-say scale intervals (5, 6, and 7) were assigned to the third group (n = 30). Average ratings for these groups on selected other variables are shown in Table 3.

--------------------------------------------------------------------------------

Table 3. Means for outcome, attribution, and judgment measures under the post. Hoc Conditions Based on the "Good Dog" Item in Experiment 2

Note: Larger mean values indicate more positivity of the outcome, or higher attributional and judgmental estimates on a scale of 1_7. Probability statements reflect simple effects tests for particular sets (rows) of means.

*The scale item read: "Is it reasonable to say that Dog B is a 'good dog?' " Scale scores of 1, 2, and 3 represented the "unreasonable" pole; 4 was the mathematical midpoint; and 5, 6, and 7 represented the "reasonable" pole.
--------------------------------------------------------------------------------

The table reveals several strong effects in terms of the following types of measures:

Outcome. There was a positive relationship between "good dog" scores and ratings of the positivity of the outcome, F(2, 88) = 4.66, p < .05.

Attribution. People scoring lowest on the good dog scale were also the lowest on the excuse/justify variable; people scoring highest on the scale assigned the dog the highest average excuse/justify ratings, F(2, 88) = 4.39, p < .05.

Judgment. All of the measures in the judgment category were strongly related to the good dog index. Persons scoring highest on that predictor variable assigned the animal the lowest levels of guilt/blame, F(2, 88) = 3.54, p < .05; the lowest levels of punish, F(2, 88) = 4.83, p < .05; and the highest levels of forgive, F(2, 88) = 8.16, p < .01. The means for the sorry measure (2.00, 2.80, 3.47) were in line with certain other patterns in Table 3 but only approached significance, F(2, 88) = 3.03, p < .06.

Two other measures in Table 3 did not show reliable differences across groups. Even so, these measures are informative. The similar and high cause/due to averages (6.00, 5.62, and 5.45) indicate that most respondents viewed Dog B as the causative agent in Boy A's terror and screaming. Of even more interest, the similar and low intent/meant averages (2.07, 2.38, and 2.23) indicate that most respondents thought it unreasonable to say that Dog B intended to or meant to bring about terror and screaming in Boy A.

Discussion

Experiment 2 proved successful in extending the findings of Experiment 1. From the perspective of attribution processes, Table 2 indicates that the various measures concerning the bases and ramifications of Dog B's behavior showed fairly high and interpretable consistency. Coefficients were large for associations between ratings on certain pairs of variables central to attribution theory:

cause/due to with guilt/blame (r = .47)
excuse/justify with punish (r = -.48), and
good/likable with forgive (r = .45).

As to individual differences in orientation to dogs, the post-hoc groupings shown in Table 3 indicate that response polarity to the good dog item was clearly predictive of attribution elements and consequences (Rajecki et al. 1998 ; Shaver et al., 1986). To put it plainly, people who thought Dog B was a good dog were more lenient and generous toward the offending animal. Of course, the item was actually the 13th scale encountered by respondents as they worked through booklets.

Chronologically earlier evaluations of the outcome as good/positive or bad/negative may simply have generalized to the matter of Dog B's goodness. We think it far more plausible, however, that our respondents had differing feelings about dogs before they received the instrument and that those prior sentiments, among others, influenced ratings of outcome. Future research can evaluate this assumption.

Finally, the overall average rating of intent/meant from Experiment 2 (M = 2.28) deserves attention. In an earlier finding, Rajecki et al. (1998) also reported a similar low rating of intent/meant under a Dog B bite condition (M = 2.51). These low ratings are in contrast to higher overall excuse/justify ratings from the 2 sources: Ms = 4.71 and 4.77, respectively. Apparently, when Dog B bites Boy A, many people are reluctant to hold the animal personally accountable for its dark deed.

Implications

Based on methods for the study of causal attribution, our experiments add to the literature showing that there is something like a "dog-positivity bias" on the part of many human observers (Experiment 1). Further, the degree of this positivity across individuals predicts levels of leniency and generosity toward dog misbehavior (Experiment 2). We believe there are three areas of human/companion animal relations where this positivity dimension can have a real-life impact: under-reported dog bites, love for metaphorical children, and pet relinquishment decisions.

Underreported Dog Bites

A previous article by Rajecki et al. (1998) pointed to the phenomenon of underreported bites of owned dogs. They reasoned, based on findings similar to those from the current experiments, that the leniency and generosity shown by people toward the misbehavior of a companion animal could lead to an unstated understanding that bites by a companion animal need not or should not be reported. Thus, positivity toward the familiar animal would spare it disagreeable consequences, at least for mild or first bites.

Love for Metaphorical Children

In a recent essay, Tenner (1998) argued that not only do people seek reflections of their own personal qualities in their pets but that dogs can also represent the virtues of entire nations. One national symbol is the "mascot." Mascot animals, not held to high performance standards, are said to be natural greeters and goodwill ambassadors: the English bulldog, French poodle, and German dachshund.. In effect, “. . . mascots are metaphorical children, loved as much for their foibles and mild misbehavior as for the positive side of their character" (Tenner, 1998, p. 74).

Tenner's insight seems echoed in today's direct marketing world. The Celebration Fantastic catalog recently featured a cover photo of a colorful papier mache statue of a dog lifting its hind leg as in urination, touted as a "naughty" beagle. Additionally, a recent Miles Kimball catalog offered two iron doorstops: one that resembled a cat appearing to sharpen its claws on a vertical surface, and the other a dog lifting its hind leg as in the act of urination. According to the catalog blurb, "These two pesky pet doorstops are humorous --but realistic --examples of forgotten lessons in behavior."

Apparently, these "naughty" objects are marketable because they are amusing or touching to some consumers. They illustrate Tenner's point about metaphorical children (dogs and cats) who are loved (by some) as much for their foibles and mild misbehavior as for their positive
characteristics. In the terminology of the current research and theory, a urinating dog icon available from a catalog--even if it had really soiled valued object--can continue to be loved because some people (if asked) would rate him low on intent, and high on excuse.

Pet Relinquishment Decisions

In extension of the matter of soiling, behavior problems are the basis for a fair percentage of pet relinquishments. One study reported that 32% of previous owners voluntarily identified such problems and, during in-depth interviews, suggested that the rate might have been even higher (DiGiacomo, Arluke, & Patronek, 1998).

We propose that the pet-positivity dimension, in the context of attributional processes, could figure into decisions about problematic animal behavior. Given an act by a companion animal that might be seen as a problem, as with the dog in Figure 1, a decision to employ therapy or training versus a decision to surrender the animal to a shelter could well depend on whether an owner views the actor as a "good dog."


References

Boenning, D. A., Fleisher, G. R., & Campos, J. M. (1983). Dog bites in children: Epidemiology, microbiology, and penicillin prophylactic therapy. American Journal of Emergency Medicine, 1, 17-21.

DiGiacomo, N., Arluke, A., & Patronek, G. (1998). Surrendering pets to shelters: The relinquisher's perspective. Anthrozoos, 11, 41-51.

Fidler, M., Light, P., & Costall, A. (1996). Describing dog behavior psychologically: Pet owners versus non_owners. Anthrozoos, 9, 196-200.

Griego, R. D., Rosen, T., Orengo, I. F., & Wolf, J. E. (1995). Dog, cat, and human bites: A review. Journal of the American Academy of Dermatology, 33, 1019-1029.

Hewstone, M. (1989). Causal attribution. Cambridge, MA: Basil Blackwell.

Leung, A. K. C. & Robson, W. L. M. (1992). Human bites in children. Pediatric Emergency Care, 8, 255-257.

Miller, F. D., Smith, E. R., & Uleman, J. (1981). Measurement and interpretation of situational and dispositional attributions. Journal of Experimental Social Psychology, 17, 80_95.

Moss, S. P. & Wright, J. C. (1987). The effects of dog ownership on judgments of dog bite likelihood. Anthrozoos, 1, 95-99.

Rajecki, D. W., Rasmussen, J. L., Modlin, S. J., & Holder, A. M. (1998). Dog bites boy: Judgments of blame and shame. Anthrozoos, 11, 66-73.

Rasmussen, J. L. & Rajecki, D. W. (1995). Differences and similarities in humans' perceptions of the thinking and feeling of a dog and a boy. Society & Animals, 3, 117-137.

Ross, M. & Fletcher, G. J. O. (1985). Attribution and social perception. In G. Lindzey and E. Aronson (Eds.), Handbook of social psychology: Vol. 2 (3rd ed., 73-122). New York: Random House.

Russell, D. (1982). The causal dimension scale: A measure of how individuals perceive causes. Journal of Personality and Social Psychology, 42, 1137-1145.

Russell, D. W., McAuley, E., & Tarico, V. (1987). Measuring causal attributions for success and failure: A comparison of methodologies for assessing causal dimensions. Journal of Personality and Social Psychology, 52, 1248-1257.

Sanders, C. R. (1990). Excusing tactics: Social responses to the public misbehavior of companion animals. Anthrozoos, 4, 82-90.

Sears, D. O. (1983). The person_positivity bias. Journal of Personality and Social Psychology, 44, 233-250.

Shaver, K. G. & Drown, D. (1986). On causality, responsibility, and self_blame: A theoretical note. Journal of Personality and Social Psychology, 50, 697-702.

Tenner, E. (1998). Citizen canine. Wilson Quarterly, 22, 71-79.

Weiner, B. (1995). Judgments of responsibility. New York: Guilford Press.

Wells, D. L. & Hepper, P. G. (1997). Pet ownership and adults' views on the use of animals. Society & Animals, 5, 45-63.


Notes

1. Correspondence should be sent to D. W. Rajecki, Department of Psychology, LD124, IUPUI, 402 North Blackford Street, Indianapolis, IN 46202-3275. Phone (317) 274-6766; Fax: (317) 274-6756: E-mail: drajecki@iupui.edu. Copies of the questionnaires employed in this project are available on request.

2. The word "causing" was inadvertently employed in the bite outcome of Experiment 1. The concern is that in drawing respondents' attention to any causal relationship, attributional distinctions about what characterized the cause of the bite outcome (internality, controllability, stability) may have been blurred. However, if the bite results were truly contaminated by increased error variance, one would expect parallel shifts for both actors relative to play ratings and small differences between the actors. Instead, opposite shifts and strong differences emerged for biting. In any event, the word "causing" was replaced in Experiment 2.

 

For Abstracts of all issues, including the most current, click Article Abstracts

To order Society & Animals Journal, go to our secure online ordering page

You can Search the online issues of Society & Animals, as well as the entire Society & Animals Forum (formerly PSYETA) website,
for topics and keywords of your interest: