Segmentation Analysis Of Participatory Behaviour In Road Race Events In Taiwan

To explore the segmentation cluster of road race events’ participants in demographic variables and participation behaviour.

To confirm the effectiveness of discriminant analysis of segmentation clusters.

To compare the differences in the demographics of participants of road race events between segmentation groups.

Data Collection

This study targeted road race event participants who have participated in road race events over the past year in Taiwan. Both purposive sampling and convenience sampling methods were adopted to collect data from those aged between 18 and 79. A total of 2,000 questionnaires were distributed to people who had participated in road race events between October 20, 2015 and February 25, 2016; a total of 1,556 valid questionnaires were collected (Table 2 ).

Respondents profile

The demographic analysis of questionnaire responses comprised 948 males (60.9%) and 608 females (39.1%), whose average age was 35.63 (SD=10.68); the highest proportion of respondents were aged between 31 and 40. The yearly rate of participation was an average of 5.96 (SD=8.52) times, and each year 2-4 times was the highest proportion (Table 3 ).

Instrumentation

The questionnaire consisted of eight sections: Participation Motivations Scale, Past Experiences Scale, Destination Image Scale, Event Image Scale, Satisfaction Scale, Intention Scale, Revisiting behaviour Scale and Demographics Scale. The development of the questionnaire was based on relevant literature regarding participation motivations (Agrusa, et al., 2011), past experiences (Kaplanidou & Gibson, 2010), destination image (Agapito, Valle, & Mendes, 2013; Papadimitriou, Apostolopoulou, & Kaplanidou, 2015), event image (Hallmann, Kaplanidou, & Breuer, 2010), event satisfaction (Goh, 2015; Papadimitriou et al., 2015), intention (Koo et al., 2014; Papadimitriou et al., 2015) and revisiting behaviour (Julaimi & Talib, 2016; Kaplanidou & Vogt, 2010). The Demographics Scale includes gender, age, education level, occupation, monthly income and participation rate in road race events over the past year. Analysis of the six scales revealed that: (1) the critical ratio values and total correlation coefficients for each item reached a significant level, showing discrimination in the construction of the scale; (2) factor analysis showed that the total cumulative reading of the six scales explained 76.04–83.59% of the variance, illustrating that all research instruments had construct validity; (3) reliability analysis showed that the Cronbach's alpha coefficients for each scale was above .90, suggesting that the internal consistency was acceptable (Chiou, 2010). The above three scales demonstrated sound discrimination, validity, and reliability, and that are suited to measure participants’ opinions regarding participation behaviours in road race events (Liu, Huang, Kuo, & Chen, 2016).

Data Analysis

SPSS version 17.0 was used to provide descriptive statistical analysis of participants’ demographics and participation behaviours. The cluster analysis was used to analyse the segregation groups to compare the effectiveness of segmentation variables and to analyse the effectiveness of the segmentation groups. The discriminant analysis and chi-square tests were used to compare the differences in the demographics of participants of road race events between different groups segments.

Analysis of the Types of Road Race Events

As shown in Table 1 , the average rate of participation in road race events per year as reported by participants is 5.96 (SD=8.52). Regarding the types of road race events (Table 2 ), the one with the highest percentage is half marathon (24.57%), followed by marathon (16.82%), 6-10 km (17.25%), 11-20 km (13.29%), 5 km or less (11.88%), ultra-marathon (8.31%) and triathlon (7.88%). The analysis shows that participants favour half marathons, 6-10 km and marathons.

Cluster Analysis of Demographic Variables and Participation Behaviour of Road Race Events Participants

In the case of unknown clusters, eleven variables, such as demographic variables (gender, age, income and participation rate) and participation behaviour (motivations, past experiences, destination image, event image, satisfaction, intention and revisiting behaviour) were used as input variables. Using a two-step cluster analysis, the distribution of dendrograms was used to roughly estimate the clusters as three clusters; subsequently, the K-mean clustering method divided the data and determined that the three clusters were suitable for differentiation. The numbers of sample distributions of the three clusters were “cluster 1” 742 (47.7%), “cluster 2” 551 (35.4%), and “cluster 3” 263 (16.9%). In accordance with the importance ranking of input forecasters in the two-step cluster analysis, the most important segmentation variables were: gender, participation intention, revisiting behaviour and satisfaction.

In this study, based on cluster analysis results, that cluster variable is the standardized score of the variable (the average is 0 and the standard deviation is 1), so the “final cluster centres” of each cluster will may be negative, and it can see the difference of the average of the three clusters. It can be used as a basis for cluster naming. From Table 6 , we can see that the higher values for “cluster 1” road race events participants are: motivations, past experiences, destination image, event image, satisfaction, intention and revisiting behaviour. “Cluster 1” participants have higher participation experience for road races events and the males account for 99.6% of participants. Therefore, they were named the “Male High Experience Group". The "cluster 2" participating experiences for road race events are also relatively high, and the female ratio is 86.8%. Therefore, they were named the "Female High Experience Group". The value of "cluster 3" for road race events is low to medium, however, the male and female ratios were 51.7% and 48.3%, respectively. Consequently, they were named the "Middle and Low Experience Group".

Confirming the Effectiveness of Discriminant Analysis of Segmentation Cluster of Road Race Events Participants

To confirm that it is appropriate to segment participants of the road race events into 3 clusters, this study further used discriminant analysis to confirm the correctness of the cluster results. The analysis results in Table 5 show that there are actually 742 people in the “Male and High Experience Group”; as 728 people were correctly predicted to belong to this cluster, the correct prediction rate was 98.1%. In the “Female and High Experience Group”, the total number of people are 551; as 469 people were predicted to belong to this cluster, the correct prediction rate was 85.1%. In the “Middle and Low Experience Group”, there are 263 people; as 223 people were predicted to belong to this cluster, the correct prediction rate was 84.8%. The percentage of overall correct results of above discriminant analysis was 91.3% ((728+469+223)/1556)). Since the chi-square value reached a significant level of .05 (χ2=2349.5>5.99), the result of this discriminant analysis was significantly higher than that of random classification results.

Comparing the Differences in Demographic Statistics and Segmentation Analysis of Road Race Events Participants

In order to understand the differences between the three clusters and demographic variables of the “Male and High Experience Group”, “Female and High Experience Group”, and “Middle and Low Experience Group” of road race participants, this study conducted a cross-analysis and a chi-square test on gender, age, educational level, occupation, income and participation rate. The results are presented in Table 7 . From the above analysis results, it can be seen that demographic variables of road race events participants are significantly different from those of the segmentation clusters (χ2=25.8, P- value<.001〜χ2=1001.6, P-value<.000). Therefore, participants showing different demographic segmentation of road races have significant differences in the cluster segmentation (Wu, 2008).

The above results are similar to those of previous studies that indicated gender, age, education, occupation, and income are the main demographic variables influencing participation behaviour of road race events or sports tourism; moreover, they confirm that men are more likely to participate in such events than women (Liu, 2016; Liu & Lee, 2011). Second, the participation behaviour of road race events or sports tourism are influenced by satisfaction (Hallmann et al., 2010; Koo et al., 2014; Oshimi & Harada, 2016), participation intention (Julaimi & Talib, 2016; Basaran, 2016; Jin et al., 2013; Papadimitriou et al., 2015) and revisiting behaviour (Agapito et al., 2013; Basaran, 2016; Julaimi & Talib, 2016; Kaplanidou & Gibson, 2010).

However, most studies that include a participants’ segmentation analysis in sport activities are based on demographic variables and sports participation motivation, behaviour or attitude (Australian Sports Commission, 2012, 2013; Chao, Wu, Lin, & Wang, 2013; Chen & Huang, 2014). In the study of cluster segmentation on road race (marathon) participants, most of the studies were analysed using demographic variables or participation motivation or training characteristics or individual and social aspect. Ogles and Masters (2003) conducted a cluster analysis based on a motivational profile and demographic and training characteristics of 1519 marathon runners, which yielded 5 definable subgroups: running enthusiasts, lifestyle managers, personal goal achievers, personal accomplishers, and competitive achievers. Vos et al. (2008) conducted a study where ‘runner identity’, ‘health/wellness’, ‘social aspect of running’, and ‘ease of individual participation’ served as input for the cluster analysis; consequently, five groups could be distinguished: ‘traditional/generic runners’(35.2%), ‘social competitive runners’(14.5%), ‘individual fitness runners’(12.2%), ‘individual competitive runner’ (24.9%), and ‘companionship runners’ (13.2%).

Based on the above analysis, it is evident that past cluster analyses of road race events (marathon) participants have taken participants’ motivation, training characteristics, or individual and social aspects as input variables. Therefore, personal motivations and attitudes have become the main key factors; personal motives were endorsed most often across all groups. This study mainly used demographic variables (such as gender, age, income or participation rate) and participation process factors (motivation, past experiences, event imagery, destination imagery, event satisfaction, participation intention and revisiting behaviour) as input variables; consequently, this is the main difference between this study and past research.

Road Race Participants Cluster Segmentation

From the results of the research and analysis, it can be found that the demographic variables (gender, age, income, and participant rate) and participant behaviours (participation motivation, past experiences, destination image, event image, event satisfaction, participation intention, and revisiting behaviour) of road race events in Taiwan could be the basis for cluster segmentation. Based on these dimensions, three clusters could be identified: "Male and High Experience Group", "Female and High Experience Group", and "Middle and Low Experience Group". This study further used discriminant analysis to confirm the correctness of the cluster results; the percentage of overall correct results of above discriminant analysis was 91.3% and the chi-square value reached a significant level. The result of this discriminant analysis was significantly higher than that of random classification results. In addition, this study conducted a cross-analysis and a chi-square test on demographic variables among three clusters; the results are significantly different from those of the segmentation clusters. Therefore, demographic variables and participation behaviour can be applied as the basis for market segmentation with regard to participants of road race events in Taiwan.

Recommendations

The following recommendations have been devised to assist the hosting city, organizers, sports associations and running clubs to provide products that meet needs of road race participants, as well as increase their participation and loyalty. First , the event and destination imagery of road race events are key factors affecting satisfaction. Consequently, it is necessary to enhance the quality of the event, including the atmosphere and characteristics, to improve event satisfaction. Second, past experiences, destination image, and participation motivation are key factors affecting the participation intention and revisiting behaviour of the road race events. The organizers and the host city must consider allowing participants to accumulate emotional awareness, strengthen destination characteristics, meet participants’ needs, and arouse participants’ goal orientation and drive of motivation; such measures will have a positive impact on intention and revisiting behaviour of road race events. It is important for host cities to consider promoting local accommodation, transportation, restaurants, and attractions. Offering well-trained staff or volunteers to assist participants with all aspects of their stay in the host city will help to foster a positive destination image, improve event satisfaction and participation experience, and improve revisiting behaviour (Goh, 2015; Hallmann et al., 2010; Jin et al., 2013).