Research On Applicability And Mode Effects Of Social Media Surveys

With the gradual mature of Web 2.0 ideas and technologies, social media has aroused widespread concern from all sectors of community in 2008. After rapid development in recent years, all kinds of social media has been riddled with every aspect of people’s lives and showed tremendous values in several areas.In the fields of survey methodology, mixed mode surveys are rapidly gaining popularity. Meanwhile, emergence and rapid development of social media provides new channels and platforms for designing of mixed mode web surveys. Researchers should investigate fully relative data obtained by this survey method when implementing web surveys based on social media. Given the new concepts of social media, little researches and studies has presently addressed the social media surveys in academically domestic and foreign fields. However, data quality is the critical issue in the field of surveys. If researchers need to incorporate social media surveys into mixed mode designs, it is extremely important that they detect data gathered by social media surveys, which determining the adaptability and application prospects of social media as a new survey platform.Given the above consideration, the focus of this thesis is the suitability and mode effects of data gathered by social media surveys. Main contents and relevant conclusions involving in this thesis are as follows.Firstly, it considered the measurement equivalence between traditional mode surveys and social media surveys.By using the CFA, this chapter had studied the measurement equivalence among the pen-and-pencil survey, web survey, BBS survey, micro blog survey and SNS survey from the perspective of scales in order to incorporate social media surveys into mixed mode designs to make up for the problem of low sample coverage in a traditionally single survey mode. The results showed that these five survey methods had the same basic structure and factor loading matrix. Except for pen-and-pencil survey, the rest four survey methods met the equivalence of factor covariance matrix. Furthermore, the micro blog survey and SNS survey had equivalence in the intercepts.The findings indicate that scale meets a certain degree of equivalence in both traditional mode surveys and social media surveys. Therefore, investigators can incorporate social media surveys into traditionally mixed mode designs.Secondly, it considered the differences between the data qualities of traditional mode surveys and social media surveys.According to analytical results of former chapter, this chapter studied the differences among the data qualities of the pen-and-pencil survey, web survey, BBS survey, micro blog survey and Instant Messenger survey from the perspectives of response quality and social desirability response. The findings showed that the scales of different sensitivity had different effects on the data quality of five survey methods. Using the same scale, the five survey methods had significant differences in the response quality. Furthermore, the extents of the scales’ sensitivity influenced the five survey platforms with different degrees of social desirability bias.The results indicate that Investigators should consider the impact on data qualities produced by survey modes when mixing this method with traditional mode surveys in order to avoid relevant survey bias due to direct mixture. However, data gathered by different social media surveys has higher consistency. Researchers can analyze statistically these types of data directly without relative pretreatment. In short, we can learn that social media has some applicability as a new survey platform based on this research and former research.Thirdly, it assessed the measurement effects between anonymous survey and non-anonymous survey based on social media.Considering the various types and application characteristics of social media, by using the MG-SEM, this chapter had studied the similarities and differences between the measurement effects of the anonymous survey and non-anonymous survey based on social media in order to explore the feasibility of social media as an independently mixed survey platform. The results showed that both mode surveys had the same response style effect, but this effect was different from any traditional effects. The two surveys had the same cognitive effect and they can lead to the same main effect. The interaction effects between survey mode and gender, age respectively were not significant differences across the modes. However, the interaction effect of education on response style was the significant difference across the modes.The results indicate that two types of social media surveys with different response situations can produce the consistent measurement effects. Therefore, investigators can implement mixed mode designs based on this platform. However, the differences between the compositions of respondents will potentially lead to a confounding of selection effects and measurement effects in mixed mode surveys.Lastly, it evaluated the mode effects between anonymous survey and non-anonymous survey based on social media.According to the former findings, by using the extended mixed-mode comparison, this chapter evaluated the selection effects and measurement effects of the anonymous survey and non-anonymous survey based on social media. The results showed that the above two mode effects had differently significant differences among the different types of variables in mixed mode surveys. Moreover, it was also different that the power conditions of mode effects among the different types of variables with regard to the same types and degrees of mode effects.The results demonstrate that the differences of selection effects and measurement effects produced by mixed social media surveys will lead to the essential impact on obtained data when the total sample size is sufficiently large. In a word, we can learn that social media has better application prospects as one kind of independent survey platforms in mixed mode designs based on this chapter and former chapter.