Prof, Can i do one way Anova on individual items measured on a 5 point scale, which constitute a formative construct (not reflective)?. Most of the reviews say that i cant do one way Anova on individual likert items as they are ordinal in nature. Please clarify & give appropriate reference for the same. Thanks
Hello, this debate feels as old as Likert scales themselves. Wu, H., & Leung, S. O. (2017). Can Likert Scales be Treated as Interval Scales?-A Simulation Study. Journal of Social Service Research, 43(4), 527-532. doi.org/10.1080/01488376.2017.1329775 sum it up pretty nicely: "Results show that more Likert scale points will result in a closer approach to the underlying distribution, and hence normality and interval scales." So, to give a clear answer: it depends. ;-) If in doubt, simply go with the Kruskal-Wallis-Test - if you samples is large enough, there is no difference in power anyways. When using formative measuements, it is worth looking at whether construct validity was achieved or not: Edwards, J. R. (2011). The Fallacy of Formative Measurement. Organizational Research Methods, 14(2), 370-388. doi.org/10.1177/1094428110378369 Cheers, Björn.
Understood++
Prof, Can i do one way Anova on individual items measured on a 5 point scale, which constitute a formative construct (not reflective)?. Most of the reviews say that i cant do one way Anova on individual likert items as they are ordinal in nature. Please clarify & give appropriate reference for the same. Thanks
Hello, this debate feels as old as Likert scales themselves.
Wu, H., & Leung, S. O. (2017). Can Likert Scales be Treated as Interval Scales?-A Simulation Study. Journal of Social Service Research, 43(4), 527-532. doi.org/10.1080/01488376.2017.1329775 sum it up pretty nicely: "Results show that more Likert scale points will result in a closer approach to the underlying distribution, and hence normality and interval scales." So, to give a clear answer: it depends. ;-) If in doubt, simply go with the Kruskal-Wallis-Test - if you samples is large enough, there is no difference in power anyways.
When using formative measuements, it is worth looking at whether construct validity was achieved or not: Edwards, J. R. (2011). The Fallacy of Formative Measurement. Organizational Research Methods, 14(2), 370-388. doi.org/10.1177/1094428110378369
Cheers, Björn.