Akademik Veri Yönetim Sistemi
5th International VETistanbul Group Congress, Ohrid, Makedonya, 23 - 27 Eylül 2018, ss.1, (Özet Bildiri)
The aim of this study was to create a scale which determines the breed perceptions of the dog
owners during pre-purchasing and/or adaptation processes. For this purpose Principal
Component Analysis was conducted. In order to assess the reliability of the scales, the internal
consistency of the 20 Likert-type items was determined by calculating Cronbach’s alpha
coefficient. The overall reliability coefficient of the 20 items was 0.832. The Kaiser-Meyer-
Olkin (KMO) test and Bartlett's test of sphericity were also applied to evaluate the suitability of
the respondents’ data for factor analysis. The KMO value greater than 0.7, and the significant
value, P < 0.05 for Bartlett's test of sphericity were accepted as the threshold levels. The actual
KMO value was 0.827; therefore, the questionnaire’s data was accepted as suitable for factor
analysis. Moreover, the actual result of Bartlett's test of sphericity (P < 0.001; chi-square =
3087.05, and df = 190) indicated that the data matrix had sufficient correlation. After these
analyses, factor analysis was conducted to assess the dimensionality of the 20 Likert-type items.
Eigenvalues greater than one were used as the criterion in the determination of the factors. An
extraction method using PCA and a rotation method using varimax with Kaiser normalization
were utilized. Four components were extracted. According to the result, the highest loadings for
Q. 8 and Q. 17 were 0.392 and 0.320, respectively. These items were excluded from the
analyses. Moreover, the loadings for Q. 15 for Component 1 and Component 3 were 0.423 and
0.419, respectively. Therefore, this item was considered a double-loaded factor and excluded
from further analyses. The remaining 17 items were subjected to reanalysis using the methods
employed for the initial dataset. Cronbach's alpha value calculated for the 17-item scale was
0.811, which indicated a high degree of internal consistency for the global scale. More than one
item provided an improvement in the global internal consistency when removed. From the
results of the PCA, four components with eigenvalues greater than one were identified. These
components explained 54.8% of the total variance.