What Factors Are Associated with Positive Effects of Dog Ownership in Families with Children with Autism Spectrum Disorder? The Development of the Lincoln Autism Pet Dog Impact Scale

Scientific literature exploring the value of assistance dogs to children with autism spectrum disorder (ASD) is rapidly emerging. However, there is comparably less literature reporting the effects of pet (as opposed to assistance) dogs to these children. In particular, there are no known validated scales which assess how children may alter their behaviours in the presence of the dog, to evaluate the efficacy of pet dogs to these families. Additionally, given the highly individualised nature of ASD it is likely that some children and families gain more benefits from dog ownership than others, yet no research has reported the effect of individual differences. This pilot study reports the development of a 28-item scale based on the perceived impact of a pet dog on a child with autism by parents (Lincoln Autism Pet Dog Impact Scale—LAPDIS). The scale is comprised of three mathematically derived factors: Adaptability, Social Skills and Conflict Management. We assessed how individual differences (aspects) may be associated with scores on these three factors. Family Aspects and Dog Aspects were not significantly associated with ratings on the three factors, but Child Aspects (including: contact with horses, child age, disability level and language abilities) were related to impact of the dog on all factors. Training Aspects were related to scores on Social Skills (formal training with children with ASD and dogs and attendance at PAWS workshops run by Dogs for Good). These results suggest that individual differences associated with the child and the training approach may be important considerations for a positive impact from dog ownership on families with children with ASD. Differences in family features and the dog may not be so important, but may be worthy of further investigations given the early stage of development in this field.


Introduction
conflict between parents and siblings on how to best care for the dog and division of responsibilities [20]. These latter issues are not likely to be encountered during AAI, and represent potential negative factors associated specifically with pet dog therapy, as opposed to trained dog therapy, which are important to consider.
Recent studies have highlighted how the acquisition of a pet dog can bring significant improvements to parent-carers of children with autism. Using standardised assessments of parenting stress [21] and family functioning [22] significant improvements were observed in families who acquired a pet dog, compared to families who did not acquire a pet dog, over the first year of ownership. However, these studies concentrated on the potential benefits of pet dogs to the family unit rather than the child with autism.
A problem with investigating child effects is selecting the appropriate measurement to employ. Given the often limited vocabulary, communication and cooperation skills associated with autism it is challenging to collect data from the child without using observational techniques. Observational methods are subject to bias both from the participant, in terms of altered normal behaviours with the presence of a researcher/recording device, and from the researcher in the coding of behaviours. As such researchers often rely on parental reports. Given the early stage of enquiry of this developing field, exploring the effects of pet (as opposed to assistance) dogs, there are no specific scales designed to assess the effects of dog companionship on behaviours of children with ASD. Existing studies have utilised general psychological scales to measure the effects of pet dogs on children (Social Skills Rating Scale- [18]) and their parents (Parenting Stress Index Short Form- [21]; The Brief Family Functioning General Scale- [22]). A barrier in preventing the development of specific scales to measure the dependent variables of dog ownership in children with autism has been the lack of research identifying how pet dogs may affect these children. Ethical considerations prevent mass testing on items which may not be relevant to the development of the field, therefore it is important that any scale items are informed by in depth qualitative information, administered to a sample size which would provide sufficient data to explore the scale without over recruiting, and are subject to appropriate tests to explore the psychometric properties of the test.
Although the literature on the value of trained assistance dogs is growing there is a clear need to develop our understanding of how pet dogs may benefit children with ASD and to develop suitable methods for assessing these effects. Here, we report on the development of a scale to assess the effects of pet dogs on children with autism and provide the first attempt to identify the individual circumstances in which dog ownership may prove beneficial to children with autism.
The aims of this pilot, validation study were to (i) develop items to use in a scale to assess the effects of dog companionship on behaviours of children with ASD, (ii) explore the construct of this scale by documenting it's psychometric properties, (iii) evaluate the individualised effects of pet dogs on children with ASD, to define the circumstances in which pet dog ownership may prove most effective.

Initial Item Development
The survey items were developed from analysis of the responses of 20 semi-structured interviews with parents with children with ASD who were dog owners. Parents were recruited from a convenience sample of the UK population via Dogs for Good Parents Autism Workshops and Support (PAWS) network. The PAWS program involves a series of three workshops that educate parents about dog behaviour, welfare, and training, whist advising on the suitability of, and integration of pet dogs into families with children with ASD. Families also receive ongoing support following the workshops, including access to a range of resources such as fact sheets, videos and telephone support. In addition postings on websites and social networks related to Dogs for Good and the National Autistic Society (NAS), and word of mouth were used to increase the number of participants. Further details of this sample and the procedures used are published [20]). The interview questions were initially compiled following a review of the existing literature, and then circulated to the project advisory group for additional input and discussion. The project advisory group was made up of twelve members including autism professionals, psychology professionals, veterinary professionals, assistance dog professionals, academics and parents of children with ASD who own a family pet dog. Feedback from the group resulted in an interview schedule addressing specific areas associated with dog ownership in families with children with autism. Based on these procedures 28 separate items were developed for the survey (see Table 1). Eleven of the items were designed to be negatively worded to avoid set responses. A five point scale was used for responding to each item: 1 = Strongly Agree, 2 = Agree, 3 = Strongly Disagree, 4 = Strongly Disagree. Participants were also required to respond to a series of demographic questions. For purpose of the analysis the demographic questions were grouped into four sections; Family Aspects, Child Aspects, Dog Aspects and Training Aspects (see Tables 2-5).

Participants
Informed written consent was obtained from each participant, and all procedures complied with British Psychological Society "Code of Ethics and Conduct", and with the World Medical Association Helsinki Declaration as revised in October 2008. The ethical committee in the School of Life Sciences, University of Lincoln approved the study. All respondents to the survey were required to own one dog, which was not an assistance dog and have a child with ASD. Participants confirmed this at the start of the survey. A total of 347 participants responded to the online survey. Those with missing data were removed, leaving a total of 197 responses in the analysis. We chose to remove incomplete responses because we cannot rule out that where participants chose not to respond to items this was because they were presenting a bias response. For instance, families may choose to only respond to items in which the dog has a positive effect to their child. The demographic details of the sample are displayed in Tables 2-5. The categories defining the child's disability level and language ability were determined based on consultation with clinical experts and are comparable to those used in previous studies [21,22]. Parents were asked to select which category best represented their child.

Design and Procedure
The items were entered into Survey Monkey in a random order (order selected using a computerised randomisation program). The order in which each item was presented in listed in Table 1. Participants were sent a link to the online survey. Instructions were presented on the screen, asking the participants to respond to each question as honestly as possible and that there were no right or wrong answers. After confirming consent questions were presented in a list format, with 8 questions presented per page. Responses were saved automatically. The survey was online for 12 months, from June 2012 to June 2013. Data was collated and analysed once the survey had closed.

Statistical Analysis
The purpose of this pilot study was to assess whether the questions selected from the item development process where appropriate for future investigations (outside of this present study) to explore the test re-rest reliability of the scale and its predictive validity. Therefore, the first step of the analysis was to determine the factor structure of the scales. To allow any factor structure to emerge we used Exploratory Factor Analysis (EFA). The number of variables (n = 28) to participants (n = 374) ratio exceeded the 5 to 1 ratio (with a minimum number of participants, n > 150) which exceeds the standard recommendation for EFA [23,24]. Inspection of the Scree plot was used to determine the number of factors to extract, before Principal Component Analysis (PCA) was undertaken using a promax rotation, as we expected the My child is more likely to show empathy for another family member (e.g., instinctively feel sad or them, rather than being prompted or instructed by another) when he/she is with the dog 18 My child shows more use of imagination when engaging in play with other people when he/she is with the dog (e.g., will follow the lead of others and engage in the 'spirit' of the game) 19 My child is more willing to go out for a walk with other family member when he/she is with the dog 21 My child shows more independence within the home if he/she is with dog (e.g., would be more likely to off into another room away from family member if the dog is with him/her)  Prefer not to say 2 (1%) 0 (0%) 2 (1%) 0 (0%) 2 (1%) 0 (0%)

Parents in Household
Total Number of Children factors to be correlated, with delta set to 0 and specifying the number of factors in the solution, based on the results of the factor analysis. Meaningful loadings were assessed using the criteria of 0.32 = poor, 0.45 = fair, 0.55 = good, 0.63 = very good, 0.71 = excellent [25,26]. Cronbach's alpha [27] coefficients were used to assess internal reliability of factors with interpretative criteria of α 0.6, < 0.7 = 'acceptable' and α > .70 = 'good' [28,29].
To explore how four sources of individual difference (Demographics: Family Aspects, Child Aspects, Dog Aspects and Training Aspects) were related to individual factor scores we conducted a hierarchical cluster analysis. Because this is an initial stage validation study of the scale we were interested in discovering how demographic factors were related to scores on the factors in the scale, rather than controlling for the effect of demographic factors, as this may artificially remove important individual differences relating to the benefits of pet dogs. By using Wards method and squared Euclidean distance as the similarity measure we examined maximum differences between the clusters (as opposed to associations). Analyses were conducted separately for each demographic aspect across each of the factors, resulting in multiple analyses. Prior to conducting the cluster analysis we explored the optimum number of clusters Number of Children with ASD Diagnosis

Family Income
Up to £10,000 £20-40,000 60 (36%) 6 (20%) 60 (36%) 6 (20%) 60 (36%) 6 (20%)   in the data sets by visually inspecting dendrogram plots and the coefficient change scores. We specified two clusters in the analysis to split the data set into high and low responders, in order to understand which factors are important in determining successful (improved) living with a dog. To assess if the two clusters were significantly different from each other independent ttests (for continuous data) or chi-square tests (for categorical data) were conducted.

Results
Factor analysis showed a clear 3 factor solution to the data. All 28 items showed 'good' loading at equal to or above 0.55 on one of the factors, with the average loading being .73 for Factor 1, .65 for Factor 2 and .61 for Factor 3. Factor 1 was comprised of 17 positively worded questions, which were themed around 'Adaptability' and are presented in Table 1. Factor 1 (Adaptability) accounted for 34% of the variance. The second factor was comprised of 7 negatively worded relating to social interactions and therefore Factor 2 was named Social Skills, which accounted for 14% of the variance. The third factor consisted of 4 negatively worded questions relating to 'Conflict Management', this factor (Conflict Management) accounted for 5% of the variance. Cronbach's alpha for the three factor were all good: Adaptability α = .93, Social Skills α = .77 and Conflict Management α = .71. One hundred and ninety seven cases (participants) responding to the demographic questions and scale items were entered into the cluster analysis. Since analyses were conducted separately for each demographic aspect across each of the three factors, there were 12 separate analyses. Inspection of the dendrogram showed a large variation within the data. However, in all analyses two over-arching groups could be determined. The results are reported separately for each of the demographic aspects. See Tables 2-5 for the mean number of participants (and percent) in each group. Scores on Factor 1, Adaptability, were comprised solely of positively worded questions, therefore a score 1 (Strongly Agree) was a positive effect and a score of 5 (Strongly Disagree) was a negative effect. As such, lower total scores reflect greater adaptability in the presence of the dog. Conversely, Factor 2 and Factor 3 were comprised only of negatively worded questions, therefore a score of 1 (Strongly Agree) was a negative effect and a score of 5 (Strongly Disagree) was a positive effect. As such, higher total scores reflect greater social skills and improved conflict management in the presence of the dog.

Family Aspects
See Table 2 for detailed information on group membership relating Family Aspects.
Factor 1: Adaptability. There was no significant difference between the two groups in adaptability (t(195) = .740, p = .460). The two groups differed significantly based on country of residence and family income. Group 1 was comprised of more parents from the UK, Australia, Belgium and Canada. Group 2 was comprised of more parents living elsewhere in the European Union, Philippines and the United States (χ² = 197.00, df = 8, p = .000). More parents in Group 1 selected their income group (χ² = 39.45, df = 5, p = .000). The groups were not significantly different based on any other family aspects.
Factor 2: Social Skills. There was no significant difference between the two groups in social skills (t(195) = 1.49, p = .137). The two groups differed significantly on country of residence and family income, in the same way as described for Factor 1 (country of residence χ² = 197.00, df = 8, p = .000, income group χ² = 39.45, df = 5, p = .000). The groups were not significantly different based on any other family aspects.
Factor 3: Conflict Management. There was no significant difference between the two groups based on social skills (t(195) = 0.69, p = .489). As with Factor 1 and Factor 2, the two groups were only significantly different from each other based on country of residence and family income (tests of statistical inference were identical to that reported in Factor and Factor 2).
The groups that emerged from the cluster analysis were not significantly different based on their scores of Adaptability, Social Skills and Conflict Management with the dog, and showed the same grouping based on family aspects. This suggests that family aspects do not significantly affect parent's perceptions of behaviour impact of the dog.

Child Aspects
See Table 3 for detailed information on group membership relating to Child Aspects.
Factor 1: Adaptability. Two groups emerged which significantly differed in their scores of adaptability (t(195) = 17.86, p = .000). The smaller group (Group 1: n = 74), showed less adaptability in the presence of the dog compared to the larger group (Group 2: n = 123). Significantly more children in Group 2 had contact with horses than children in Group 1 (χ² = 10.03, df = 1, p = .002), implying increased adaptability in the presence of the dog was associated with children who had more regular contact with horses.
Factor 2: Social Skills. The two groups were significantly different based on their scores of social skills (t(195) = 14.25, p = .000). Group 1 (n = 58) showed fewer social skills in the presence of the dog in comparison to Group 2 (n = 139). Group 2 was comprised of significantly older children in comparison to Group 1 (t(195) = 2.36, p = .01), implying increased social skills in the presence of the dog, was associated with having an older child.
Factor 3: Conflict Management. The two groups were significantly different based on their scores of conflict management (t(195) = 3.56, p = .000). Group 1 (n = 128), showed reduced conflict management in the presence of the dog in comparison to Group 2 (n = 69). The two groups were significantly different from each other based on child age, child diagnosis and child language abilities. Children in Group 2 were significantly older than children in Group 1 (t(195) = 18.58, p = .000), had less severe disability on the autism spectrum (χ² = 8.12, df = 3, p = .04), and had better language abilities (χ² = 13.05, df = 3, p = .005).
The results suggest that better conflict management from the presence of the dog is associated with children who are older, who are on the higher functioning scale of autism spectrum disorder and who have better abilities.

Dog Aspects
See Table 4 for detailed information on group membership relating to Dog Aspects.
Factor 1: Adaptability. The two groups were significantly different in their scores of adaptability (t(195) = 15.24, p = .000). Group 1 (n = 25), demonstrated lower adaptability in the presence of the dog compared to Group 2 (n = 172). The two groups did not significantly differ from each other based on individual differences on the dog aspects. However, there was a near statistically significant effect of dog sex (χ² = 3.40, df = 1, p = .06), with people in Group 2 owning more female dogs than male dogs. The results suggest that increased adaptability in the presence of the dog might be associated with owning a female dog, but this effect of dog gender, if real, is small.
Factor 2: Social Skills. The two groups scored significantly differently on social skills (t (195) = 12.05, p = .000). Group 1 (n = 163), showed fewer social skills in the presence of a dog in comparison to Group 2 (n = 34). The groups did not significantly differ based on individual differences in dog aspects, indicating that children's social skills in the presence of a dog are not significantly affected by dog aspects.
Factor 3: Conflict Management. The two groups were not significantly different based on their scores of conflict management, although the p value approached significance (t(195) = 1.82, p = .07), with Group 1 demonstrating reduced conflict management in the presence of the dog in comparison to Group 2, but we do not consider this effect reliable given the multiple testing undertaken. The groups were significantly different from each other based on dog age, how long the dog was owned for, whether the dog was neutered or not and where the dog came from. Families in Group 2 owned older dogs than families in Group 1 (χ² = 121.57, df = 5, p = .000). More families in Group 2 had owned the dog for a longer period of time (χ² = 164.77, df = 7, p = .000), and had a dog that was neutered (χ² = 7.94, df = 1, p = .005). The groups also differed based on where the dog was obtained from (χ² = 19.35, df = 5, p = .02).

Training Aspects
See Table 5 for detailed information on group membership based on Training Aspects.
Factor 1: Adaptability. There was a significant difference between the two groups based on adaptability (t(195) = 18.21, p = .000). Group 1 (n = 58) showed lower adaptability in the presence of the dog in comparison to Group 2 (n = 139). The groups did not significantly differ on any of the training aspects, indicating that high and low responses were not associated with individual differences in training aspects.
Factor 2: Social Skill. There was a significant difference between the two groups based on social skills (t(195) = 15.93, p = .000). Group 1 (n = 82) showed fewer social skills in the presence of the dog in comparison to Group 2 (n = 115). The two groups were significantly different from each other based on whether or not they had attended formal dog training with children with ASD and whether they had attended PAWS workshops. More parents in Group 2 had attended formal training (χ² = 3.99, df = 1, p = .04), and PAWS workshops (χ² = 5.01, df = 1, p = .02). The results suggest that children who show better social skills in the presence of the dogs are those whose parents have attended formal training with dogs and children with ASD and PAWS workshops.
Factor 3: Conflict Management. There was a significant difference between the two groups based on conflict management (t(195) = 16.70, p = .000). Group 1 (n = 109) was made up of more members who showed reduced conflict management in the presence of the dog in comparison to Group 2 (n = 88). The groups did not significantly differ on any of the training aspects, suggesting that high and low responses in conflict management were not associated with individual differences in training.

Discussion
With the aim of piloting the development of an assessment scale for evaluating the impact of pet dog ownership on children with ASD we designed a 28 item survey. The items were developed based on discussions with our advisory group and the results of semi-structured interview with parents [20]. With the aim of documenting the psychometric properties of this scale we conducted PCA and internal reliability analysis (alpha coefficients). The results showed that the scale was comprised of 3 factors, which were themed around Adaptability, Social Skills and Conflict Management. All the factors demonstrated good internal reliability, and were reported by parents on the advisory group as having good comprehensibility. In order to identify which types of individual difference influence impact of the dog on the child, we identified significant differences between high and low responders to the clustered items based on Family, Child, Dog, and Training Aspects. The results suggest that Family Aspects are not significantly associated with perceived impact of the dog on any of the three factors identified. Child Aspects did affect experience with dog, on all the factors (adaptability, social skills and conflict management), highlighting important individual child differences in the therapeutic effect of dog ownership in these families. Dog Aspects had little impact, and this might be due to preselection of the dog by the family to determine the type of dog that appears to fit best with the family, indicating that the effect is a general "dog" effect, assuming successful integration into the family. Training Aspects were related to social skills, in particular attendance at a workshop aimed at setting realistic expectations and maximising the potential of the family dog. We discuss the results in relation to each of the three child related factors in order to collate the potential effects for practitioners and parents who may be interested in exploring strategies to improve child functioning in these areas.

Adaptability
Rigidity in behaviours and thought patterns is a hallmark symptom associated with autism [30], as well as impacting on the behaviours of the individual with autism these symptoms can impact on the entire family unit, reducing their ability to partake in, and enjoy, a range of different activities. Indeed, it has also been suggested that first degree relatives may also show problems with [31], although it is arguably difficult to isolate whether this is a genetic or learned trait. Evidence suggests that adaptability is important in determining effectiveness of family based autism interventions [32]. Therefore, in order to develop the potential of dog ownership as an effective tool for improving adaptability it is important that we examine under what environment (i.e., individual differences) dogs are most effective at improving adaptability.
The results indicated that adaptability (questions related to both children with autism and the family), improved when the dog was present if the child had more regular contact with horses. There was some evidence to suggest that increased adaptability was associated with owning a female as opposed to male dog, but this effect if real is small, and probably not of clinical relevance-it is perhaps more important that the family get the dog who they feel is right for them. Adaptability was not significantly associated with individual differences in family factors or training on the integration of the dog into the family. These findings suggest that perhaps, in general, exposure to social non-human species may be beneficial to children (or the family unit) who have a particular difficulty with adaptability, and that the effects of exposure to different species may be synergistic. The benefits of equine therapy to children with autism are becoming of increasing interest (e.g., the horse boy method). There is some evidence to suggest that activities with horses can benefit children with ASD, with observed improvements in children's social behaviours and functioning [33,34], sensory processing and seeking [33,35] and attention and self-regulation behaviours [33,36]. However, these conclusions are based on small scale studies, which often lack a suitable control sample/condition. Owing to a number of differences in design and analysis it is difficult to relate studies highlighting the effects of horses to children with autism to studies reporting the effects of dogs. Nonetheless, our results suggest that in combination with dog ownership, contact with horses may help improve the adaptability of children with autism. Owing to the broad definition used in our study (regular contact with horses), it is not possible to define which aspect of contact with a horse may be important, and how this may relate to having a dog present. Furthermore, it is not possible to decipher whether contact with horses requires the child to be actively involved in equine therapy, or if it is the petting or presence of a horse that is important. Possible speculations include that these animals help to reduce children's repetitive behaviours by providing a stimulating distraction and breaking obsessive cycles. It is also possible that these animals help to calm and reassure the child which increases their confidence in completing new tasks [36]. Although the anxiety reducing effects of horses are not well studied, the anxiety reducing effects of the presence of a dog is of growing scientific interest, and deserves further investigation. Many reports indicate that dogs can bring a calming effect to human arousal [37][38][39]13], but some evidence is contrary to this indicating that dogs have an excitatory effect [40], or no effect [41]. A further possibility is that the presence of an animal may help to improve adaptability by increasing family freedom and ensuring child safety. Research shows that trained assistance dogs help families to safely engage in new activities and public outings by providing an anchor point and a calming influence [14]. Therefore, it is possible that the presence of a dog supports the child to engage with horses.

Social Skills
There is a well-documented history of problems in social skills and functioning in individual with autism (e.g., [42][43][44][45]. Despite the prevalence of social skill deficits in autism spectrum disorder, few children have access to effective social skills programming [46]. Indeed, results from systematic reviews and meta-analysis indicate little support for current interventions [47,48]. Social skills have a wide reaching impact on outcomes including, academic performance, social failure, anxiety, depression, suicide and substance abuse [49][50][51][52]. Therefore, it is important that we define the individual circumstances in which dog ownership may be most effective in improving social skills. The results highlight that children who show a greater improvement in social skills in the presence of the dog are older children (child aspects), and those who have had a parent attend PAWS workshops and formal training with dogs and children with ASD. There were no significant differences between high and low responders based on family aspects or dog aspects. These findings suggest that if children have a particular problem with social skills functioning parent's may want to consider acquiring a dog once their child gets older (approximately 8 years seems a reasonable recommendation, given the current data). However, the effect of age on social skills is unlikely to be unique to being in the presence of the dog. Given that social skills show a gradual maturation process it is quite possible that improved social skills with age could be found in relation to any intervention/condition. It may prove important to note that those who attended formal training, through PAWS workshops (organised by Dogs for Good) and other training showed better social skills. This could suggest that the improved social skills associated with acquiring a trained assistance dog [14] may not be due to the incidental nature of the dog, but instead may reflect the importance of training both the dog and the handler (parent) to maximise the benefits of presence of the dog. This is in contrast to observations reported by [16], who stated an increase in pro-social behaviours in children with autism upon acquisition of a pet. However, [16] explored the arrival of a pet, which may bring about novelty effects, as opposed to longer term living with a pet. Additionally, [16] examined the effects of acquiring a 'pet' as opposed to specifically a pet dog. Whilst the effect of training highlights the importance of organisations such as PAWS it should also be pointed out that it could be that parents who have attended PAWS workshops are more alert to detecting changes in their child's social behaviours, which may have contributed to the effect observed here. Future studies could investigate the impact of educating parents on detecting small changes in their child's behaviour in a group of parents who have attended PAWS workshops and in a matched control group.

Conflict Management
Autism spectrum disorder is associated with frequent child tantrums [53] and there is a relationship between family conflict and severity of ASD symptomology [54]. It is thought that children with ASD may be particularly sensitive to conflict due to their problems with sensory sensitivity and perspective [54]. Therefore, an important aspect in developing a family therapy for ASD would be to consider under what circumstances conflict management is improved. Our analysis indicates that conflict management is improved in the presence of the dog if the child is older, has better language abilities, has less severe disabilities and has owned a dog before. Dog factors seem to have little effect, but, given our results, might deserve further investigation in future. No significant effects were observed with family and or training aspects. The child aspects surrounding age, language and disability level are not surprising, and as with the previous results discussed, are unlikely to be unique to owning a dog. It would be expected that as children get older conflict management is improved, possibly through parents learning the most effective way to avoid, or solve any conflicts as they arise. Furthermore, previous research has identified a relationship between family conflict and ASD symptomology [54], supporting the result here that families whose children are higher functioning and show better languages abilities show better conflict management. The possible effect of owning a dog for longer and owning an older dog, might indicate that experience with being a dog owner may be important in effectively resolving conflict behaviours in the presence of the dog. In any case helping owners to gain the necessary expertise associated with successfully owning a dog for a longer time, should always be encouraged.
Overall, these results indicate that positive outcomes surrounding adaptability, social skills and conflict management are associated with a range of individual related factors. Whilst acquiring a pet dog may bring a range of benefits to many families living with a child with ASD, not all families are likely to experience the same improvements and to the same degree. Our results provide two particularly important contributions. Firstly, we have developed a useful scale (Lincoln Autism Pet Dog Impact Scale-LAPDIS) for assessing the impact of the presence of a pet dog in children with ASD. This is the first scale of its kind, targeted on the specific effects reported by a large sample of parents internationally. Secondly, we describe the initial disentangling of the individualised circumstances which might influence the efficacy of dog ownership within families with a child with ASD, identify important controls for more mechanistically focussed research in this field in future. It is clear that the dog is not a simple intervention, and it is likely to have multiple effects that vary between families [55]. Understanding these factors is important if families are to receive the best quality advice from their clinical advisors. We should perhaps point out, that these data clearly relate to families who want a dog and so a dog should not be prescribed to those who are not willing to make the additional commitment required for its successful integration into the family. Additionally, whilst the data presented here shows that pet dogs can be a positive addition to family life and highlights the need for further research to explore the impact of pet dogs as well as trained assistance dogs, we do not claim that pet dogs can replace the important role played by trained assistance dogs [56]. Indeed, whereas assistance dogs are chosen specifically for their temperament (e.g. calmness) and characteristics (e.g. stable enough to prevent a child from bolting when on a duallead) to support children with ASD pet dogs come in many shapes and sizes. Therefore future research should further explore how dog characteristics may affect the perceived benefit of dog ownership and compare child functioning in families who own a trained assistance dog with those who own a pet dog.
However, we stress that further research is needed to test the hypotheses generated by this study. Further validation of the LAPDIS scale is required to assess test re-test reliability and predictive validity. Additionally, it is clear that further research is needed to assess a wider range of individual differences and to conduct controlled comparisons exploring child and family responses to situations based on the individual differences identified here. Furthermore, due to survey design it was not practical to make independent assessments of the child's ASD symptoms. Therefore, we relied on parent's confirmation that their child had received a clinically confirmed diagnosis of ASD, as with our earlier studies in this area [21,22]. Given that parents had to give up 40-45 minutes of their time, with no tangible reward, there was little incentive for parents to be dis-honest in reporting their child's diagnosis. Nonetheless, future studies could ensure that parent's completed standardised checklists of symptoms as a control to improve upon this present design.
This study documents how sores on the three LAPDIS scale factors (Adaptability, Social Skills, and Conflict Management) are associated with individual differences in the child, and the training undertaken and possibly the dog, but not the family. The results further highlight important individual differences which may affect the value of pet dogs to families with children with autism which should be considered when deciding whether to acquire a dog and what might be reasonably expected given the families' circumstances.