The Relationship between the Source of Dog Acquisition and Behavioral Issues: House Training, Coprophagia, and Stereotypic Behaviors
Lorraine Gaudio
2023-03-13
Abstract
This study aims to investigate the relationship between the source of dog acquisition and the occurrence of house training issues, coprophagia, and stereotypic behaviors. I hypothesize that dogs acquired from pet shops will exhibit a higher prevalence of these behavioral issues compared to dogs acquired from other sources. Utilizing the C-BARQ dataset, I employed chi-square tests of independence and logistic regression analyses to explore the associations between dog acquisition sources and the aforementioned behavioral problems.
The logistic regression analysis revealed that dogs acquired from pet stores had a significantly higher probability of house training issues (40.13% increase) compared to those obtained from breeders. However, in relation to coprophagia, pet store-acquired dogs demonstrated a lower probability (21.80% decrease) compared to those from breeders. For stereotypic behaviors, pet store-acquired dogs showed a higher probability (80.41%) compared to breeders (50.90%), although the relationship was not statistically significant at conventional levels.
The findings of this study highlight the importance of considering the source of dog acquisition when assessing potential behavioral issues. Dogs acquired from pet stores exhibited a higher likelihood of house training issues, while other factors beyond the source of acquisition might influence coprophagia and stereotypic behaviors. Further research is needed to establish a more robust relationship between pet store acquisition and the prevalence of stereotypic behaviors in dogs.
Introduction
In recent years, there has been growing interest in understanding the factors that contribute to the development of behavioral issues in dogs. Anecdotal evidence from professionals working in the pet industry, such as dog trainers and groomers, suggests that dogs acquired from pet stores may be more prone to behavioral issues. The belief is that during their early development, these dogs are confined to small spaces where they sleep, eat, and play, leading to a disregard for living in a space with urine and feces, an increased likelihood of coprophagia, and a predisposition to stereotypic behaviors, which are commonly observed in caged animals (Hubrecht, 2005).
Serpell and Jagoe (2005) investigated the influence of early-life events on the development of behavioral problems in domestic dogs. Their research aimed to identify events that increase the risk of behavioral issues, thereby enabling dog owners and breeders to avoid negative events and reduce the risk of problems with their pets. Scott and Fuller (1965) also conducted studies on the stages of development in canine puppies of both domestic dogs and wolves. In a case-control study by Jagoe, 451 dogs with behavioral issues were compared to 286 control dogs acquired through random door-to-door inquiries (Serpell, 2005). One notable finding of this study was that dogs acquired from pet stores were rated poorly for behavior problems, corroborating the anecdotal experiences of professionals working with dogs. However, a limitation of Jagoe’s study is the small sample of dogs from pet stores (n=5) compared to dogs from other sources.
This study aims to investigate the relationship between the source of puppy acquisition and the development of house-training issues, coprophagia, and idiosyncratic or stereotypic behaviors in dogs who were acquired during puppyhood using the C-BARQ dataset (Hsu & Serpell, 2003).
H₀: The source where people acquired their dogs has no effect on house-training issues, coprophagia, and stereotypic behaviors.
Hₐ: The source where people acquired their dogs does have an effect on house-training issues, coprophagia, and stereotypic behaviors.
By exploring the relationship between where dog owners acquired their dog as a puppy and the likelihood of house training issues, coprophagia, and idiosyncratic or stereotypic behaviors in dogs acquired during puppyhood, this study seeks to answer the question: “How does the source of a pet relate to house training issues and idiosyncratic or stereotypic behaviors?” I predict that dogs acquired from a pet shop will have more occurrences of house-training issues, coprophagia, and stereotypic behaviors than dogs acquired from any other source.
Methods
Data Selection
The study utilizes the C-BARQ dataset, collected and maintained by
James A. Serpell and the University of Pennsylvania between 2008 and
2021, which contains various factors related to dog behavior and owner
demographics. The dataset was processed to include only select variables
of interest. I created the variable house training issues
by combining the survey questions that involved behavioral issue with
urinating or defecating. I created the variable stereotypic
by combining the survey questions that involved abnormal behavioral
issues that appear as a repetitive or functionless. The variable
coprophagia is from only one survey question about the dog
eating its own fecal matter. These three variables came from survey
questions that were in Likert Scale format where 0 = never, 1 = seldom,
2 = sometimes, 3 = usually, 4 = always, and NA = not observed / not
applicable. In this study, if the survey respondent answered “never” on
all survey questions within the category, value is scored as 0. If the
survey respondent answered as seldom or more, the value is scored as 1.
If the survey respondent answered NA it remained
NA and was omitted during analysis as needed.
The data used in this study only includes survey results from dogs that were acquired by the primary care giver by or before the dog was 6 months old and if the survey was taken after the dog was at least 1 year old. Additionally, this study only includes the survey results where the primary care giver had owned a dog before the current dog in the survey. These criteria are important to control of other possible reason that a dog might have behavioral issues (i.e., still a puppy or living with an inexperienced care giver). Additionally, I am also interested in focusing in on the behavior issues in question in small dogs as it is small dogs who are rumored to be most affected. This might be because small dogs are more commonly sold by pet stores or because their size makes for less annoyance of inappropriate behaviors than in larger dogs, the behaviors are allowed to continue for a prolonged period of time. I repeat my study by investigating only the small dogs in the dataset (dogs 30 lbs or less) for some of the analyses. Finally, male dogs also have a reputation for behavior issues, especially house training issues. I repeat several analyses a third time by focusing on the male dogs.
Data Analysis Plan
The study originally planned to employ two main statistical methods for data analysis: Pearson’s Chi Square Test (Goodness of Fit) and Logistic Regression. However, due to differences in the proportion of behavioral issues across the “where acquired” categories, I determined that it would be more appropriate to use a chi-square test of independence. To further investigate the association between dogs’ acquisition sources and behavioral issues, I conducted a series of Bonferroni-adjusted pairwise tests of independence for each behavioral issue using the full dataset. Additionally, I performed the same tests for house training issues within subsets of the data consisting of small dogs and male dogs. To visualize the results, I generated bar plots that focused on the comparisons between the pet store acquisition source and all other sources, illustrating the proportion of dogs with each behavioral issue along with error bars representing the standard error and significance labels indicating the statistical significance of the differences.
Chi-square Test Analyses
H0: The variables ‘where dogs were acquired’ and ‘behavioral problems’ are independent (There is no association between where dogs were acquired and the occurrence of behavioral problems).
Ha: The variables ‘where dogs were acquired’ and ‘behavioral problems’ are dependent (There is an association between where dogs were acquired and the occurrence of behavioral problems).
These assumptions of the data are met: The data has two categorical
variables where one is binary and the other is nominal.The data is
independent. This is demonstrated as one unique
dogid for each survey response. The expected frequencies
are greater than 5. With large sample sizes, even small differences in
cell frequencies can lead to statistically significant results in a
chi-square test. In such cases, it’s essential to examine the effect
size, which provides information about the magnitude of the association
between the variables.For demonstration purposes, I present the
evaluation of the investigation into differences in the proportion of
the house training issues variable.
Contingency Table (Goodness of Fit)
##
## bredbyowner breeder friendorrelative other petstore shelter stray Sum
## 0 1217 13140 1887 1032 512 3583 744 22115
## 1 571 4601 1165 489 464 1401 382 9073
## Sum 1788 17741 3052 1521 976 4984 1126 31188##
## bredbyowner breeder friendorrelative other petstore shelter
## 0 2.751526 29.708343 4.266335 2.333258 1.157585 8.100837
## 1 3.146699 25.355450 6.420148 2.694809 2.557037 7.720710
## Sum 2.866487 28.442029 4.892908 2.438438 1.564704 7.990253
##
## stray Sum
## 0 1.682116 50.000000
## 1 2.105147 50.000000
## Sum 1.805181 50.000000In the table above, the row percentages represent the distribution of the “where acquired” categories within each level of house training issues (present or absent).
##
## bredbyowner breeder friendorrelative other petstore shelter stray
## 0 34.03244 37.03286 30.91415 33.92505 26.22951 35.94502 33.03730
## 1 15.96756 12.96714 19.08585 16.07495 23.77049 14.05498 16.96270
## Sum 50.00000 50.00000 50.00000 50.00000 50.00000 50.00000 50.00000
##
## Sum
## 0 35.45434
## 1 14.54566
## Sum 50.00000In the above table, the column percentages represent the distribution of the house training issues within each source of dog acquisition.
There are differences in the proportion of house training issues across the “whereacquired” categories. For example, dogs acquired from a breeder seem to have a lower proportion of house training issues (12.97%) compared to those bred by the owner (15.97%) or acquired from a friend or relative (19.09%). This suggests that the source of acquisition may have an association with house training issues. In this case, it would be more appropriate to use a chi-square test of independence rather than a chi-square goodness of fit test. The chi-square test of independence compares the observed frequencies in a contingency table with the frequencies that would be expected if the two variables were independent of each other.
House Training Contigency Tables & Chi-squared tests
Full Data
Contingency Table (Independence)
##
## bredbyowner breeder friendorrelative other petstore shelter stray Sum
## 0 1217 13140 1887 1032 512 3583 744 22115
## 1 571 4601 1165 489 464 1401 382 9073
## Sum 1788 17741 3052 1521 976 4984 1126 31188House Training Issues by Where Dogs Were Acquired
chi-square (Independence)
##
## Pearson's Chi-squared test
##
## data: contig_table
## X-squared = 397.75, df = 6, p-value < 2.2e-16## The critical value is: 12.59159The value I found (\(\chi^2\) = 397.75, df = 6) is larger than the critical value (12.59159) and the p-value is less than 0.05 (p=2.2e-16). Therefore, I have strong evidence to reject the null hypothesis in favor of the alternative that there is an association between where dogs were acquired and the occurrence of house training issues.
Small Dogs Only
Contingency Table (Independence)
##
## bredbyowner breeder friendorrelative other petstore shelter stray Sum
## 0 249 2657 403 169 269 550 131 4428
## 1 195 1969 540 182 337 432 145 3800
## Sum 444 4626 943 351 606 982 276 8228House Training Issues In Small Dogs by Where Dogs Were Acquired
chi-square (Independence)
##
## Pearson's Chi-squared test
##
## data: contig_table
## X-squared = 104.47, df = 6, p-value < 2.2e-16The value I found (\(\chi^2\) = 104.47, df = 6) is larger than the critical value (12.59159) and the p-value is less than 0.05 (p=2.2e-16). Therefore, I have strong evidence to reject the null hypothesis in favor of the alternative that where small dogs were acquired and house training issues are dependent.
Male Dogs Only
Contingency Table (Independence)
##
## bredbyowner breeder friendorrelative other petstore shelter stray Sum
## 0 429 6729 954 503 259 1798 339 11011
## 1 334 2600 620 261 275 713 205 5008
## Sum 763 9329 1574 764 534 2511 544 16019House Training Issues In Male Dogs by Where Dogs Were Acquired
chi-square (Independence)
##
## Pearson's Chi-squared test
##
## data: contig_table
## X-squared = 278.72, df = 6, p-value < 2.2e-16The value I found (\(\chi^2\) = 278.72, df = 6) is larger than the critical value (12.59159) and the p-value is less than 0.05 (p=2.2e-16). Therefore, I have strong evidence to reject the null hypothesis in favor of the alternative that there is an association between where male dogs were acquired and the occurrence of house training issues.
House Training Results
Of the 31188 dogs in my survey sample, 9073 dogs were reported to have house training issues, while 22115 did not. A chi-square test for association was conducted to test whether there was an association between where dogs are acquired and house training issues. The results showed a significant association between source of the dog and house training (χ2=397.75(6, N = 31,188=12.59 p < 2.2e-16). Further investigation of the 8,228 small dogs in my survey sample, 3,800 small dogs were reported to have house training issues, while 4,428 did not. The chi-square test for the same association was conducted and found that there was a significant association between the source of little dogs and house training (χ2=104.47(6, N=8,228=12.59 p < 5.139e-12). A third investigation of 16019 male dogs in my survey sample found that 5,008 dogs had house training issues where 11,011 dogs did not. The chi-square test for the same association was conducted and found that there was a significant association between the source of male dog and house training issues (χ2=278.72(6, N = 16,019=12.59 p < 2.2e-16). I conclude that Te variables where dogs were acquired and house training issues are dependent.
Coprophagia Contigency Tables
Full Data
Contingency Table (Independence)
##
## bredbyowner breeder friendorrelative other petstore shelter stray Sum
## 0 983 9450 1695 788 598 2466 587 16567
## 1 803 8239 1345 721 379 2460 529 14476
## Sum 1786 17689 3040 1509 977 4926 1116 31043Coprophagia In All Dogs by Where Dogs Were Acquired
chi-square (Independence)
##
## Pearson's Chi-squared test
##
## data: contig_table
## X-squared = 55.837, df = 6, p-value = 3.141e-10Small Dogs Only
Contingency Table (Independence)
##
## bredbyowner breeder friendorrelative other petstore shelter stray Sum
## 0 233 2284 546 184 359 464 141 4211
## 1 208 2304 382 159 250 515 134 3952
## Sum 441 4588 928 343 609 979 275 8163Coprophagia In Small Dogs by Where They Were Acquired
chi-square (Independence)
##
## Pearson's Chi-squared test
##
## data: contig_table
## X-squared = 46.482, df = 6, p-value = 2.373e-08The value I found (\(\chi^2\) = 46.482, df = 6) is larger than the critical value (12.59159) and the p-value is less than 0.05 (p=2.373e-08). Therefore, I have strong evidence to reject the null hypothesis in favor of the alternative that where small dogs are acquired and coprophagia issues are dependent.
Male Dogs Only
Contingency Table (Independence)
##
## bredbyowner breeder friendorrelative other petstore shelter stray Sum
## 0 453 5379 896 420 348 1276 287 9059
## 1 312 3920 668 346 187 1206 251 6890
## Sum 765 9299 1564 766 535 2482 538 15949Coprophagia In Male Dogs by Where They Were Acquired
chi-square (Independence)
##
## Pearson's Chi-squared test
##
## data: contig_table
## X-squared = 54.15, df = 6, p-value = 6.882e-10The value I found (\(\chi^2\) = 54.15, df = 6) is larger than the critical value (12.59159) and the p-value is less than 0.05 (p=6.882e-10). Therefore, I have strong evidence to reject the null hypothesis in favor of the alternative that where male dogs are acquired and coprophagia issues are dependent.
Coprophagia Results
Of the 31,043 dog in this survey, 14,476 dogs were reported to eat their fecal mater while 16,567 did not. A chi-square test for association was conducted to test whether there was an association between where the dog was acquired and eating fecal mater (coprophagia issues). Results show a significant association between the source of the dog and coprophagia χ2=55.837(6, N = 31,043 =12.59 p < 3.141e-10). In a second investigation of 8,163 dogs in this survey, 3,952 small dogs had coprophagia issues while 4,211 small dogs did not. A chi-squared test for the same association found a significant association between coprophagia and the source of the small dog χ2=46.482(6, N = 8,163=12.59 p < 2.373e-08). A third investigation of 15,949 male dogs in this survey found that 6,890 male dogs had coprophagia issues while 9,059 did not. A chi-squared test for the same association found a significant association between the source of the male dog and coprophagia issues χ2=54.15(6, N = 15,949=12.59 p < 6.882e-10). I conclude that the variables where dogs were acquired and coprophagia are dependent.
Stereotypic Chi Square
Full Data
Contingency Table (Independence)
##
## bredbyowner breeder friendorrelative other petstore shelter stray Sum
## 0 1153 8600 1098 666 374 2026 447 14364
## 1 623 9021 1958 853 590 2921 667 16633
## Sum 1776 17621 3056 1519 964 4947 1114 30997Stereotypic Behaviors In All Dogs by Where Dogs Were Acquired
chi-square (Independence)
##
## Pearson's Chi-squared test
##
## data: contig_table
## X-squared = 523.74, df = 6, p-value < 2.2e-16The value I found (\(\chi^2\)= 523.74, df = 6) is larger than the critical value (12.59159) and the p-value is less than 0.05 (p=2.2e-16). Therefore, I have strong evidence to reject the null hypothesis in favor of the alternative that where dogs are acquired and stereotypic behavior are dependent.
Small Dogs Only
Contingency Table (Independence)
##
## bredbyowner breeder friendorrelative other petstore shelter stray Sum
## 0 248 2086 336 143 227 407 99 3546
## 1 188 2489 609 204 375 566 176 4607
## Sum 436 4575 945 347 602 973 275 8153Stereotypic Behaviors In All Dogs by Where Dogs Were Acquired
chi-square (Independence)
##
## Pearson's Chi-squared test
##
## data: contig_table
## X-squared = 80.564, df = 6, p-value = 2.732e-15The value I found (\(\chi^2\) = 80.564, df = 6) is larger than the critical value (12.59159) and the p-value is less than 0.05 (p=2.732e-15). Therefore, I have strong evidence to reject the null hypothesis in favor of the alternative that where small dogs are acquired and stereotypic behavior are dependent.
Male Dogs Only
Contingency Table (Independence)
##
## bredbyowner breeder friendorrelative other petstore shelter stray Sum
## 0 456 4410 560 328 194 1014 212 7174
## 1 302 4866 1014 442 331 1475 324 8754
## Sum 758 9276 1574 770 525 2489 536 15928Stereotypic Behaviors In All Dogs by Where Dogs Were Acquired
chi-square (Independence)
##
## Pearson's Chi-squared test
##
## data: contig_table
## X-squared = 191.22, df = 6, p-value < 2.2e-16The value I found (\(\chi^2\) = 191.22, df = 6) is larger than the critical value (12.59159) and the p-value is less than 0.05 (p=2.2e-16). Therefore, I have strong evidence to reject the null hypothesis in favor of the alternative that where small dogs are acquired has unequal proportions of stereotypic behavior.
Stereotypic Results
Of the 30,997 dog in this survey, 16,633 dogs were reported to have abnormal behavior that appear as a repetitive and apparently functionless behavior (stereotypic behavior) while 14,364 did not. A chi-square test for association was conducted to test whether there was an association between stereotypic issues and where the dog was acquired. Results show a significant association between the source of the dog and stereotypic χ2=523.74(6, N = 30,997 =12.59 p < 2.2e-16). In a second investigation of 8,153 small dogs in this survey, 4,607 small dogs had stereotypic issues while 3,546 small dogs did not. A chi-squared test for the same association found a significant association between stereotypic and the source of the small dog χ2=80.564(6, N = 8,153=12.59 p < 4.946e-08). A third investigation of 15,928 male dogs in this survey found that 8,754 male dogs had stereotypic issues while 7,174 did not. A chi-squared test for the same association found a significant association between the source of the male dog and stereotypic issues χ2=191.22(6, N = 15,928=12.59 p < 2.2e-16). I conclude that there is an association between where dogs were acquired and the occurrence of stereotypic behavioral problems.
Note: Odds Ratio can only compare four cells. Since it can’t handle larger contingency tables, I do not calculate the odds ratio.
Chi-Square Conclusion
In conclusion, the results show dependence association between where dogs are acquired and house training issues, coprophagia, and stereotypic behavior for all dogs, small dogs, and male dogs. This suggests that where a dog is acquired may have an impact on the likelihood of these behavioral issues. Further research is needed to better understand the reasons behind these associations and to develop interventions or strategies to address these issues.
Post Hoc Tests
House Training Issues
All Dogs
## Comparison p.Chisq p.adj.Chisq
## 1 bredbyowner : breeder 4.95e-08 1.04e-06
## 2 bredbyowner : friendorrelative 1.46e-05 3.07e-04
## 3 bredbyowner : other 9.25e-01 1.00e+00
## 4 bredbyowner : petstore 7.57e-16 1.59e-14
## 5 bredbyowner : shelter 2.50e-03 5.25e-02
## 6 bredbyowner : stray 2.83e-01 1.00e+00
## 7 breeder : friendorrelative 4.31e-44 9.05e-43
## 8 breeder : other 1.56e-07 3.28e-06
## 9 breeder : petstore 2.85e-49 5.98e-48
## 10 breeder : shelter 2.21e-03 4.64e-02
## 11 breeder : stray 4.54e-09 9.53e-08
## 12 friendorrelative : other 7.48e-05 1.57e-03
## 13 friendorrelative : petstore 2.55e-07 5.36e-06
## 14 friendorrelative : shelter 7.60e-21 1.60e-19
## 15 friendorrelative : stray 1.29e-02 2.71e-01
## 16 other : petstore 1.56e-14 3.28e-13
## 17 other : shelter 2.65e-03 5.56e-02
## 18 other : stray 3.58e-01 1.00e+00
## 19 petstore : shelter 7.85e-33 1.65e-31
## 20 petstore : stray 2.90e-10 6.09e-09
## 21 shelter : stray 1.23e-04 2.58e-03The pairwise_results show the comparisons between different dog acquisition sources and their association with house training issues. Each row represents a pairwise comparison, and the columns show the p-values before and after applying the Bonferroni correction for multiple comparisons.
Since my hypothesis is that dogs acquired from a pet shop will have more occurrences of house-training issues, coprophagia, and stereotypic behaviors than dogs acquired from any other source, I will focus on the comparisons involving pet stores. For all these comparisons, the adjusted p-values are very low (<< 0.05), which indicates that there is a significant association between the acquisition source and house training issues, specifically when comparing pet stores to other sources.
Small Dogs
## Comparison p.Chisq p.adj.Chisq
## 1 bredbyowner : breeder 6.16e-01 1.00e+00
## 2 bredbyowner : friendorrelative 4.48e-06 9.41e-05
## 3 bredbyowner : other 3.13e-02 6.57e-01
## 4 bredbyowner : petstore 2.32e-04 4.87e-03
## 5 bredbyowner : shelter 1.00e+00 1.00e+00
## 6 bredbyowner : stray 2.96e-02 6.22e-01
## 7 breeder : friendorrelative 1.82e-16 3.82e-15
## 8 breeder : other 8.66e-04 1.82e-02
## 9 breeder : petstore 1.55e-09 3.26e-08
## 10 breeder : shelter 4.32e-01 1.00e+00
## 11 breeder : stray 1.44e-03 3.02e-02
## 12 friendorrelative : other 9.30e-02 1.00e+00
## 13 friendorrelative : petstore 5.56e-01 1.00e+00
## 14 friendorrelative : shelter 7.62e-09 1.60e-07
## 15 friendorrelative : stray 1.86e-01 1.00e+00
## 16 other : petstore 2.90e-01 1.00e+00
## 17 other : shelter 1.34e-02 2.81e-01
## 18 other : stray 9.28e-01 1.00e+00
## 19 petstore : shelter 8.63e-06 1.81e-04
## 20 petstore : stray 4.37e-01 1.00e+00
## 21 shelter : stray 1.43e-02 3.00e-01The analysis of house training issues in small dogs revealed significant differences between various acquisition sources. When comparing pet stores to other sources, the adjusted p-values are consistently low (<< 0.05), indicating a strong association between acquiring small dogs from pet stores and increased house training issues.
Further examination of the pairwise results for the acquisition sources of small dogs reveals that the most significant comparisons involving pet stores are with dogs acquired from bred by owner, friend or relative, and shelter sources. The p-values for these comparisons are very low after the Bonferroni correction, signifying a significant difference in the prevalence of house training issues between small dogs acquired from pet stores and those from the mentioned sources.
The comparison between pet stores and shelters has the lowest adjusted p-value of all comparisons (p.adj.Chisq = 1.81e-04), indicating the strongest association between acquiring small dogs from pet stores and house training issues. In contrast, the comparison between breeder and shelter has an adjusted p-value above 0.05 (p.adj.Chisq = 1.00e+00), suggesting that acquiring small dogs from these sources may not have a significant difference in the occurrence of house training issues.
Male Dogs
## Comparison p.Chisq p.adj.Chisq
## 1 bredbyowner : breeder 2.04e-20 4.28e-19
## 2 bredbyowner : friendorrelative 4.80e-02 1.00e+00
## 3 bredbyowner : other 1.45e-04 3.04e-03
## 4 bredbyowner : petstore 7.22e-03 1.52e-01
## 5 bredbyowner : shelter 2.15e-15 4.51e-14
## 6 bredbyowner : stray 3.17e-02 6.66e-01
## 7 breeder : friendorrelative 2.53e-20 5.31e-19
## 8 breeder : other 2.44e-04 5.12e-03
## 9 breeder : petstore 2.71e-31 5.69e-30
## 10 breeder : shelter 6.21e-01 1.00e+00
## 11 breeder : stray 1.04e-06 2.18e-05
## 12 friendorrelative : other 1.63e-02 3.42e-01
## 13 friendorrelative : petstore 1.29e-06 2.71e-05
## 14 friendorrelative : shelter 3.87e-13 8.13e-12
## 15 friendorrelative : stray 5.14e-01 1.00e+00
## 16 other : petstore 6.21e-10 1.30e-08
## 17 other : shelter 2.63e-03 5.52e-02
## 18 other : stray 2.10e-01 1.00e+00
## 19 petstore : shelter 6.73e-25 1.41e-23
## 20 petstore : stray 6.75e-06 1.42e-04
## 21 shelter : stray 2.31e-05 4.85e-04In the analysis of house training issues in male dogs, pairwise comparisons were utilized to examine the relationship between the acquisition source and the prevalence of house training issues. The results, as presented in the pairwise results table, showed that most comparisons involving pet stores and other sources have low adjusted p-values (<<0.05), indicating a significant association between the acquisition source and house training issues in male dogs.
Notably, the pairwise comparison between pet stores and breeders (p.adj.Chisq = 5.69e-30) and between pet stores and shelters (p.adj.Chisq = 1.41e-23) demonstrated very low p-values. This suggests that acquiring male dogs from pet stores is significantly associated with a higher occurrence of house training issues compared to acquiring them from breeders or shelters.
Furthermore, the pairwise comparison between other sources and pet stores (p.adj.Chisq = 1.30e-08) also yielded a very low p-value, reinforcing the significant association between the acquisition source and house training issues in male dogs. However, it is important to note that the comparison between bred by owner and pet stores (p.adj.Chisq = 1.52e-01) had a higher adjusted p-value, implying that there may not be a significant association between these two acquisition sources in terms of house training issues for male dogs.
HTI Bar Plots With Pairwise Comparisons
The significance labels are set to “****” to represent p-values << 0.05.
Coprophagia Post Hoc
Following the investigation of house training issues, I proceeded to examine the coprophagia behavior issue in the context of dogs’ acquisition sources.
## Comparison p.Chisq p.adj.Chisq
## 1 bredbyowner : breeder 2.00e-01 1.00e+00
## 2 bredbyowner : friendorrelative 6.50e-01 1.00e+00
## 3 bredbyowner : other 1.14e-01 1.00e+00
## 4 bredbyowner : petstore 1.98e-03 4.16e-02
## 5 bredbyowner : shelter 3.46e-04 7.27e-03
## 6 bredbyowner : stray 2.13e-01 1.00e+00
## 7 breeder : friendorrelative 1.81e-02 3.80e-01
## 8 breeder : other 3.83e-01 1.00e+00
## 9 breeder : petstore 2.38e-06 5.00e-05
## 10 breeder : shelter 3.13e-05 6.57e-04
## 11 breeder : stray 6.14e-01 1.00e+00
## 12 friendorrelative : other 2.61e-02 5.48e-01
## 13 friendorrelative : petstore 3.10e-03 6.51e-02
## 14 friendorrelative : shelter 8.62e-07 1.81e-05
## 15 friendorrelative : stray 7.54e-02 1.00e+00
## 16 other : petstore 1.27e-05 2.67e-04
## 17 other : shelter 1.50e-01 1.00e+00
## 18 other : stray 8.79e-01 1.00e+00
## 19 petstore : shelter 2.37e-10 4.98e-09
## 20 petstore : stray 8.83e-05 1.85e-03
## 21 shelter : stray 1.34e-01 1.00e+00The analysis of the coprophagia behavior issue provided several interesting findings from the Bonferroni-adjusted pairwise tests of independence. For instance, dogs acquired from pet stores exhibited a significantly lower prevalence of coprophagia compared to those from shelters (p.adj.Chisq = 4.98e-09) and strays (p.adj.Chisq = 1.85e-03). Additionally, there was a significant difference between dogs from breeders and pet stores (p.adj.Chisq = 5.00e-05) as well as between dogs from friends or relatives and pet stores (p.adj.Chisq = 6.51e-02), although the latter comparison only marginally met the threshold for statistical significance.
On the other hand, several comparisons did not yield significant associations. For example, the comparisons between dogs from pet stores and those bred by owners (p.adj.Chisq = 1.00e+00), dogs from pet stores and other sources (p.adj.Chisq = 1.00e+00), and dogs from pet stores and strays (p.adj.Chisq = 1.00e+00) did not exhibit statistically significant differences in the prevalence of coprophagia.
Coprophagia Bar Plot
These findings suggest that the acquisition source of dogs may be associated with differences in coprophagia behavior, with pet store acquisitions showing a lower prevalence compared to some other sources.
Stereotypic Post Hoc
## Comparison p.Chisq p.adj.Chisq
## 1 bredbyowner : breeder 3.39e-38 7.12e-37
## 2 bredbyowner : friendorrelative 2.99e-84 6.28e-83
## 3 bredbyowner : other 1.16e-33 2.44e-32
## 4 bredbyowner : petstore 3.01e-39 6.32e-38
## 5 bredbyowner : shelter 2.99e-67 6.28e-66
## 6 bredbyowner : stray 1.04e-38 2.18e-37
## 7 breeder : friendorrelative 1.75e-39 3.68e-38
## 8 breeder : other 2.28e-04 4.79e-03
## 9 breeder : petstore 1.72e-09 3.61e-08
## 10 breeder : shelter 1.68e-22 3.53e-21
## 11 breeder : stray 2.25e-08 4.72e-07
## 12 friendorrelative : other 2.63e-07 5.52e-06
## 13 friendorrelative : petstore 1.16e-01 1.00e+00
## 14 friendorrelative : shelter 8.46e-06 1.78e-04
## 15 friendorrelative : stray 1.44e-02 3.02e-01
## 16 other : petstore 1.46e-02 3.07e-01
## 17 other : shelter 4.89e-02 1.00e+00
## 18 other : stray 6.17e-02 1.00e+00
## 19 petstore : shelter 2.26e-01 1.00e+00
## 20 petstore : stray 5.67e-01 1.00e+00
## 21 shelter : stray 6.35e-01 1.00e+00In the analysis of stereotypic behavior issues in all dogs, a post hoc pairwise comparison was conducted to assess the relationship between the acquisition source and the prevalence of these issues. The results, as presented in the pairwise results table, showed varying levels of significance between the different acquisition sources.
Focusing on the comparisons involving pet stores, the pairwise comparison between pet stores and bred by owner (p.adj.Chisq = 6.32e-38), and between pet stores and breeder (p.adj.Chisq = 3.61e-08) revealed interesting findings. The comparison between pet stores and these breeders demonstrated a very low adjusted p-values, indicating a significant association between acquiring dogs from pet stores and the occurrence of stereotypic behavior issues compared to acquiring them from breeders.
Conversely, the comparison between pet stores and shelters had an adjusted p-value above 0.05 (p.adj.Chisq = 1.00e+00), suggesting that there may not be a significant difference in the occurrence of stereotypic behavior issues between dogs acquired from pet stores and those acquired from shelters. Both dogs in pet stores and animal shelter spend a significant amount of time confined. This may by the reason for this lack of significant difference.
Stereotypic Bar Plot
Discussion
The results of the chi-square tests indicate a significant association between where dogs were acquired and the occurrence of house training issues, coprophagia, and stereotypic behavior for all dogs, small dogs, and male dogs. These findings suggest that the source from which a dog is acquired may have an impact on the likelihood of these behavioral issues.
In the case of house training issues, the results show significant differences between pet stores and other sources, especially when it comes to small dogs and male dogs. The most significant comparisons involving pet stores for small dogs are with dogs acquired from bred by owner, friend or relative, and shelter sources. For male dogs, the most significant comparisons are between pet stores and breeders, and pet stores and shelters.
When analyzing coprophagia, dogs acquired from pet stores exhibited a significantly lower prevalence of this behavior compared to those from shelters. However, comparisons between dogs from pet stores and those bred by owners, other sources, and strays did not exhibit statistically significant differences in the prevalence of coprophagia. Pet store acquired dogs do not have an increase in coprophagia compared to other sources. It is quite the opposite which I found surprising. More research is needed to determine the source of this result. I expect there are several other variables involved that were not investigated here. Perhaps there are a large proportion of cage setups that allow fecal mater to fall out of reach of the young puppy. This might prevent the behavior from establishing.
Conclusion
In conclusion, this study highlights the impact of the acquisition source on behavioral issues in dogs and emphasizes the need for further investigation to better understand the reasons behind these associations. This knowledge can help develop interventions or strategies to address these issues and promote the well-being of dogs, regardless of their acquisition source. These findings emphasize the importance of considering the acquisition source when acquiring a dog, as it may be associated with differences in house training issues, coprophagia, and stereotypic behavior. Further investigation the factors contributing to these differences, such as early life experiences, socialization, or rearing conditions may shed light on why the differences exist. This understanding can inform targeted interventions and educational efforts to address these issues and improve the well-being of dogs, particularly those acquired from pet stores.
Logistic Regression Analyses
A logistic regression is also an appropriate method of analyzing the CBARQ data because the outcome variable is binary. I will determine the probability of behavior problems based on the predictor variables (where dogs are acquired). Although logistic regression is typically used for continuous independent variables, I can use it for nominal/categorical variables by converting them into dummy variables (binary variables that represent each category).
Dummies
Multicollinearity issues arises when using dummy variables and one solution to address this issue is to exclude one of the dummy variables, typically the most frequent or least frequent category, as a reference. I selected to exclude the most frequent.
## The most frequent category of where a dog is acquired is: breederRegressions
House Training Regression
## # A tibble: 8 × 5
## term estimate std.error statistic p.value
## <chr> <dbl> <dbl> <dbl> <dbl>
## 1 (Intercept) -0.667 0.0629 -10.6 3.29e-26
## 2 relevel.whereacquired..ref...most_freq… -0.383 0.0652 -5.87 4.42e- 9
## 3 relevel.whereacquired..ref...most_freq… -0.0901 0.0808 -1.11 2.65e- 1
## 4 relevel.whereacquired..ref...most_freq… 0.184 0.0731 2.52 1.17e- 2
## 5 relevel.whereacquired..ref...most_freq… -0.0803 0.0835 -0.961 3.37e- 1
## 6 relevel.whereacquired..ref...most_freq… 0.568 0.0898 6.32 2.54e-10
## 7 relevel.whereacquired..ref...most_freq… -0.272 0.0704 -3.87 1.09e- 4
## 8 relevel.whereacquired..ref...most_freq… NA NA NA NA## Category Estimate Percent_Diff
## 1 Intercept -0.67 -38.51
## 2 breeder -0.38 -23.56
## 3 bredbyowner -0.09 -5.87
## 4 friendorrelative 0.18 12.52
## 5 petstore 0.57 40.13
## 6 shelter -0.27 -17.14
## 7 stray NA NA
## 8 other -0.08 -5.23The logistic regression analysis for house training issues provides insights into the role of different sources of acquiring puppies. In this analysis, the reference category is dogs acquired from a breeder. It is important to interpret the results in the context of the p-values, which indicate the statistical significance of the coefficients. The overall probability of house training issues in dogs acquired from a breeder (the reference category) is 38.51%. When comparing other sources of acquiring puppies to the reference category, I observe that dogs bred by the owner have a lower probability of house training issues by 5.87 percentage points compared to dogs from a breeder. However, this difference is not statistically significant (p = 0.2649), suggesting that the observed difference might not be reliable. Dogs acquired from a friend or relative have a significantly higher probability of house training issues by 12.52 percentage points (p = 0.0117) compared to the reference category. Dogs obtained from a pet store exhibit a significantly higher probability of house training issues (40.13%) compared to dogs from a breeder (p < 0.001). Dogs adopted from a shelter display a significantly lower probability of house training issues by 17.14 percentage points compared to the reference category (p < 0.001). And finally, for dogs acquired from the “other” category, the probability of house training issues is lower by 5.23 percentage points compared to the reference category, but this difference is not statistically significant (p = 0.3365). The categories, friend or relative, pet store, and shelter categories have statistically significant differences compared to the reference category, while the bred by owner and other categories do not.
House Training Plot
The results show that dogs acquired from pet stores have a 65% higher predicted probability of having house training issues compared to dogs acquired from breeders. Dogs acquired from friends or relatives have a predicted probability of having house training issues that is more than 35% higher than dogs from breeders. On the other hand, dogs acquired from shelters show only a slightly higher predicted probability of house training issues, just above 5% compared to dogs from breeders. This information suggests that dogs acquired from breeders tend to have fewer house training issues compared to dogs from pet stores and friends or relatives, while dogs from shelters show a relatively small difference in the probability of house training issues.
Coprophagia Regression
## # A tibble: 8 × 5
## term estimate std.error statistic p.value
## <chr> <dbl> <dbl> <dbl> <dbl>
## 1 (Intercept) -0.104 0.0599 -1.74 8.27e-2
## 2 relevel.whereacquired..ref...most_frequ… -0.0331 0.0618 -0.535 5.92e-1
## 3 relevel.whereacquired..ref...most_frequ… -0.0982 0.0765 -1.28 1.99e-1
## 4 relevel.whereacquired..ref...most_frequ… -0.127 0.0702 -1.81 6.99e-2
## 5 relevel.whereacquired..ref...most_frequ… 0.0152 0.0791 0.192 8.48e-1
## 6 relevel.whereacquired..ref...most_frequ… -0.352 0.0889 -3.96 7.51e-5
## 7 relevel.whereacquired..ref...most_frequ… 0.102 0.0664 1.53 1.26e-1
## 8 relevel.whereacquired..ref...most_frequ… NA NA NA NA## Category Estimate Percent_Diff
## 1 Intercept -0.10 32.10
## 2 breeder -0.03 37.29
## 3 bredbyowner -0.10 32.53
## 4 friendorrelative -0.13 30.41
## 5 petstore -0.35 14.35
## 6 shelter 0.10 47.20
## 7 stray NA NA
## 8 other 0.02 40.84Based on the logistic regression analysis for the variable Coprophagia, the overall probability of coprophagia in dogs acquired from a breeder (the reference category) is 37.29%. Compared to dogs acquired from a breeder, the probability of coprophagia is decreased in dogs acquired from a friend or relative by 6.88%, and in puppies from a pet store by 22.94%. On the other hand, the probability of coprophagia in dogs from a shelter is increased by 9.91%. If one were to breed their own dog and keep a puppy, the probability of coprophagia in that puppy as an adult dog would be decreased by 4.76% compared to buying a puppy from a breeder. Finally, for dogs acquired from the “other” category, the probability of coprophagia is increased by 3.55%. These percentage differences represent a meaningful difference in the probability of coprophagia between the different categories of where the dogs were acquired.
Coprophagia Plot
The results show that dogs acquired from pet stores have a predicted probability of coprophagia that is more than 20% lower (approximately 23% lower) than dogs acquired from breeders. Dogs acquired from friends or relatives and bred by owner also have lower predicted probabilities of coprophagia compared to dogs from breeders, with values just more negative than -5% and -5%, respectively. On the other hand, dogs acquired from shelters have a 10% higher predicted probability of coprophagia compared to dogs from breeders. This information suggests that dogs acquired from breeders are more likely to exhibit coprophagia than dogs from pet stores, friends or relatives, and bred by owner, while dogs from shelters are more likely to exhibit coprophagia than dogs from breeders.
Stereotypic Regression
## # A tibble: 8 × 5
## term estimate std.error statistic p.value
## <chr> <dbl> <dbl> <dbl> <dbl>
## 1 (Intercept) 0.400 0.0611 6.55 5.85e-11
## 2 relevel.whereacquired..ref...most_freq… -0.352 0.0630 -5.60 2.17e- 8
## 3 relevel.whereacquired..ref...most_freq… -1.02 0.0788 -12.9 5.02e-38
## 4 relevel.whereacquired..ref...most_freq… 0.178 0.0718 2.48 1.31e- 2
## 5 relevel.whereacquired..ref...most_freq… -0.153 0.0801 -1.91 5.64e- 2
## 6 relevel.whereacquired..ref...most_freq… 0.0556 0.0900 0.618 5.37e- 1
## 7 relevel.whereacquired..ref...most_freq… -0.0344 0.0676 -0.508 6.11e- 1
## 8 relevel.whereacquired..ref...most_freq… NA NA NA NA## Category Estimate Percent_Diff
## 1 Intercept 0.40 103.41
## 2 breeder -0.35 50.90
## 3 bredbyowner -1.02 3.40
## 4 friendorrelative 0.18 88.86
## 5 petstore 0.06 80.41
## 6 shelter -0.03 74.05
## 7 stray NA NA
## 8 other -0.15 65.53The logistic regression analysis for Stereotypic behaviors reveals several interesting insights into the role of different sources of acquiring puppies. As with the other analyses, the reference category is dogs acquired from a breeder. The overall probability of stereotypic behavior in dogs acquired from a breeder (the reference category) is 50.90%. When comparing other sources of acquiring puppies to the reference category, dogs bred by the owner exhibit a significantly lower probability of stereotypic behavior, with a decrease of 47.5 percentage points (3.40% probability) compared to dogs from a breeder.Dogs acquired from a friend or relative have an increased probability of stereotypic behavior by 37.96 percentage points (88.86% probability) compared to the reference category. Dogs obtained from a pet store exhibit a probability of stereotypic behavior at 80.41%, which is an increase of 29.51 percentage points compared to dogs from a breeder. Dogs adopted from a shelter display a probability of stereotypic behavior at 74.05%, which is an increase of 23.15 percentage points compared to the reference category. And, lastly, for dogs acquired from the “other” category, the probability of stereotypic behavior is 65.53%, which represents an increase of 14.63 percentage points compared to the reference category.
Stereotypic Plot
The bar plot presents the percentage difference in the predicted probability of stereotypic behavior for various sources of acquiring dogs, with “breeder” as the reference category. The “bred by owner” category exhibits the most significant difference with approximately -47% compared to dogs acquired from breeders. This indicates that dogs bred by the owner are less likely to display stereotypic behaviors than dogs from breeders. The highest positive value is the “friend or relative” category, which has an estimated 37% difference in the predicted probability of stereotypic behavior compared to dogs from breeders. This suggests that dogs acquired from friends or relatives are more likely to exhibit such behaviors.
Interestingly, the “pet store” category shows a value of approximately 30%, making it one of the categories with a higher predicted probability of stereotypic behavior when compared to breeders. In the context of the p-values, the bred by owner category is highly significant (p < 0.001), while the pet store category has a higher p-value (p = 0.5366), suggesting that the pet store differences observed are not as reliable.
Discussions
My logistic regression analysis focused on the relationship between the source of dog acquisition and various behavioral issues, particularly in the context of pet stores. The results provide valuable insights into the potential risks associated with purchasing dogs from different sources, emphasizing the importance of considering the source when acquiring a dog.
Specifically focusing on pet shop sources, my analysis suggests that puppies acquired from pet stores have a significantly higher probability (40.13% increase) of house training issues compared to those obtained from a breeder. This finding is notable, as it emphasizes the potential risks associated with purchasing dogs from pet stores, where the conditions in which the animals are raised and the resultant early learning they receive may be suboptimal for house training.
When comparing the role of pet stores as a source of puppies in relation to coprophagia, it is evident that pet stores are associated with a significantly lower probability of coprophagia compared to acquiring puppies from breeders. The probability of coprophagia in puppies from pet stores is decreased by 21.80% when compared to puppies from breeders. This suggests that dogs acquired from pet stores are less likely to exhibit coprophagia behavior compared to those from breeders. However, it is important to consider other factors beyond the source of acquisition that might contribute to the development of coprophagia, such as genetics, environmental factors, and the individual dog’s experiences.
Regarding the role of pet stores in stereotypic behavior compared to other sources of puppies, my analysis shows that dogs acquired from pet stores exhibit a higher probability of stereotypic behavior (80.41%) compared to those obtained from breeders (50.90%). This finding suggests that pet stores might be associated with a higher likelihood of dogs exhibiting stereotypic behaviors. However, it is essential to consider that the p-value for the pet store category is not statistically significant. Therefore, the relationship between pet stores as a source and the occurrence of stereotypic behaviors is interpreted with caution. Further research might be necessary to establish a more robust relationship between pet store acquisition and the prevalence of stereotypic behaviors in dogs.
It is important to acknowledge that I used dummy variables in my logistic regression analysis to handle the categorical variables. Multicollinearity issues can arise when using dummy variables, and one solution to address this issue is to exclude one of the dummy variables, typically the most frequent or least frequent category, as a reference. In my analysis, I chose to exclude the most frequent category (breeders) as the reference. This approach allows me to make meaningful comparisons between the various sources of dog acquisition while mitigating the potential issues associated with multicollinearity.
In conclusion, my logistic regression analysis provides valuable insights into the relationship between the source of dog acquisition and the probability of house training issues, coprophagia, and stereotypic behaviors. Notably, dogs acquired from pet stores exhibit a significantly higher likelihood of house training issues compared to those from breeders, highlighting the importance of considering the source when acquiring a dog. Additionally, my analysis demonstrates that using a reference category in my regression model, in this case breeders, allows for meaningful and reliable comparisons between different dog acquisition sources while addressing potential multicollinearity issues.
Conclusion
In conclusion, the results of both the chi-square tests and logistic regression analysis reveal significant associations between the source of dog acquisition and the occurrence of specific behavioral issues, such as house training problems, coprophagia, and stereotypic behaviors. These findings emphasize the importance of considering the source when acquiring a dog, as it may have a considerable impact on the likelihood of these behavioral issues manifesting. Additionally, the study highlights the need for further research into the factors contributing to these differences, such as early life experiences, socialization, and rearing conditions. This understanding can inform targeted interventions and educational efforts aimed at addressing these issues and improving the well-being of dogs, particularly those acquired from pet stores.
Future research should focus on understanding the underlying reasons for these behavioral differences, expanding the scope of investigation to encompass other behavioral concerns not addressed in this paper, as well as considering the actual occurrences of these behavior problems in society. Ultimately, these insights can help dog owners, breeders, shelters, and pet stores make more informed decisions when acquiring and raising dogs, leading to better welfare for dogs and stronger bonds between dogs and their human companions.
References
Scott, John Paul, and John L. Fuller. 1965. Dog Behavior: The Genetic Basis. Chicago: The University of Chicago Press.
Serpell, James, and J. A. Jagoe. 2005. “Early Experience and the Development of Behaviour.” In The Domestic Dog: Its Evolution, Behavior, and Interactions with People , edited by James Serpell, 79–102. New York: Cambridge University Press.
Hubrecht, R.C. 2005. “The Welfare of Dogs in Human Care.” In The Domestic Dog: Its Evolution, Behavior, and Interactions with People , edited by James Serpell, 179–198. New York: Cambridge University Press.
Hsu, Y., & Serpell, J. A. (2003). Development and validation of a questionnaire for measuring behavior and temperament traits in pet dogs. Journal of the American Veterinary Medical Association, 223(9), 1293-1300.