Optimization of Data Processing and Presentation in Social Surveys: From Likert-Means to 'Yes Percentage'
| Autor: | Daiva Bubeliene, Gediminas Merkys |
|---|---|
| Rok vydání: | 2019 |
| Předmět: |
Data processing
Statistical regularity Scale (ratio) Carry (arithmetic) 05 social sciences 050105 experimental psychology Likert scale 03 medical and health sciences 0302 clinical medicine Statistics Replication (statistics) 0501 psychology and cognitive sciences 030217 neurology & neurosurgery Mathematics |
| Zdroj: | Communications in Computer and Information Science ISBN: 9783030298616 |
| DOI: | 10.1007/978-3-030-29862-3_2 |
| Popis: | The article presents a replication study of a significant empirical study carried out earlier. Borg, Gabler (2002) showed that there is an extremely high correlation between the Means of Likert Scale Items and the percentage of agreement (Yes%). On this basis, they suggested to use not traditional Likert mean, but Yes % in survey reports. The latter are easier to interpret and do not have the same problem as Equidistance of Likert-type scales. It was decided, based on the big date (N ≈ 9000), to carry out the replication study in another historical time, in another culture and measuring another construct. If the statistical regularity detected by Borg, Gabler is repeated, it is universal, then it is really appropriate to move to a wider use of Yes% when preparing the survey report. The replication study showed that there is an extremely high correlation (R2 = 0.948) between the primary Likert items means and Yes% of items, approximating to the linear function. Used the classic 5-grade Likert scale. The verification is carried out only at the level of single items without passing to the level of analysis of additive indexes. It also turned out that the Likert items-mean correlation with the No% is lower (R2 = 0.865), which negates the postulate of symmetry of the scale. In addition, the Likert means correlation with “neutral category%” is even lower - R2 = 0.340. When preparing Survey studies reports, give priority to “yes %” instead of Likert items means. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|