I made survey to rank factors(eg. material price fluctuation) affect on construction performance. I collect data by using Likert-scales to know impact of the factors ( 1.very low, 2.low, 3.medium, 4.high, 5.very high). I also ask the frequency of factors(eg. how often does material price fluctuation have you experienced?). I used similar Likert scales on the frequency ( 1. never, 2.rarely, 3.sometimes, 4.often, 5. always).
I want to know the significant opinion between two groups ( private and public) on the impact on construction performance.
Can I use ANOVA test or what would be the most appropriate test?
I will suggest to change the nature of factors, split each factor into two or three categories , High, medium and low then go for ANOVA.
In your case your data on construction performance and factors both are continuous in nature which doesn't meet the requirements.
To meet the requirements split your continuous data and convert it into categories.
I hope this will help.
For further clarification read ANOVA assumptions.
Dear Sung ko ko Myint,
Before deciding on which test to use, it is essential to know the level of measurement of your DV & IVs. In your case, you measured your DV at ordinal level ( 5 point Likert scale). Basically, ANOVA is used to compare means of more than two groups whose DV is measured at scale level( interval or ratio). When we want to compare means of two groups, we apply t- test provided that the DV is measured at scale level. Both ANOVA & T-test are parametric tests which are applied with data that meet the assumptions of ordinary least square (OLS). However, your DV is measured at ordinal level which does not satisfy the OLS assumptions. Hence, you should choose non- parametric tests. Mann Whitney U test ( the non- parametric version of t- test) is appropriate for your study since it is used to compare two groups. I hope this can help you a lot.
- Badges
- Science topic
- Similar topics
- Engineering
- Control Theory
- Control Systems
- Control Systems Engineering