I have used New ecological paradigm scale in a study to measure the endorsement of respondents about new ecological paradigm. New ecological paradigm scale is consist of 15 questions including both anthropocentric and ecocentric questions. respondents have to indicate their rate of agreement from strongly agree to strongly disagree. In other words, I have used likert scale to know their views.
The questionnaire was also consistof several items relating to the participants’ sociodemographic background, such as age, gender, place of residence, year of study at the university, name of faculty, and family’s socioeconomic status as perceived by the participant.
Which statistical test to be used to know the difference in the views of respondents based on their sociodemographic background?
Student t-test, if related to SEM analysis can perform multi-group analysis by socio-demographic trait etc.
Because there have been so many questions here about using Likert-scored items and converting them of interval-level scales, I have compiled a set of resources on this topic:
https://www.researchgate.net/post/How_can_one_help_people_create_simple_scales_from_Likert-scored_items
If you create an interval level scale as your dependent variable, then you can use t-Tests and correlations to assess bi-variate relationships, and multiple regression to build a multi-variate model.
David L Morgan the dependent variable of my study is in Likert scale (Categorical variable), so how to transform the data from ordinal to interval?
You have Likert-scored items, and the ResearchGate page I referred to earlier contains several resources that will help you determine whether you can combine those items into a scale.
Abdul Hadi David's suggestions are indeed correct. Moreover, the discussion he has shared is immensely helpful. I have learnt so many things for analysis from the same thread.
Coming to your problem, if not solved yet. There is a whole lot of debate on the dimensionality of New ecological paradigm (NEP) scale. Originally it was proposed to have one dimension, however, some researcher using factor analysis has proven otherwise. Now the important point is your stance: Should NEP scale be treated as a whole or you would like to run factor analysis to see if there are more than one dimensions? Once you decide on that, you can compute variable in SPSS and run t-tests and/or ANOVA depending upon the type of the independent variables and examine the differences. Effect size and post-hoc analysis will further enhance the analysis.
Should you have further query, feel free to ask (even using private message).
- Badges
- Science topic