Multivariate multilevel model SPSS 28 (March 2022)
HTML-код
- Опубликовано: 21 авг 2024
- This video is designed to provide a demonstration of an application of multivariate multilevel modeling based on an example provided by Heck et al. (2014; www.routledge..... In this example, I provide a demonstration of multilevel analyses based on survey responses (which are treated as repeated measures within person), where person-level (i.e., Level 2) and organizational level (Level 3) predictors are included as predictors of constructs measured by those survey items.
Note: This video is not intended to be exhaustive and cover all aspects of multilevel modeling strategy nor does it address areas of controversy/debate within the multilevel modeling community.
Here are links to the data and Powerpoint files referenced in the video:
SPSS data in wide format: drive.google.c...
SPSS data in long format: drive.google.c...
Supplemental powerpoint: drive.google.c...
Hi Mike, I always find your videos so helpful and so I am just wondering if you could offer me some advice. I am very confused on levels required. I have a research question where I want to look at the impact of a specific road safety education programme in schools. I have a sample of schools who ran the programme and a sample of schools who did not. I have a self reported behaviour survey as the outcome variable. I would like to know if there is a difference in the students behaviour as a result of the education received but I also need to consider the impact of its being an urban or rural school, the gender of the student and their general safety beliefs ( collected through additional survey for all children). I am therefore wondering how I structure this in multilevel modelling? Any advice would be greatly appreciated! I have SPSS version 28. Thank you!
Based on what you are describing, students (Level 1) should be treated as being clustered within schools (i.e., Level 2 units). I'm not clear what your outcome variable is, but I gather you are predicting general safety beliefs? Given that schools represent Level 2 units and assignment to the treatment versus control condition occurs at the school level, I would treat your treatment (vs no) variable as a fixed factor at Level 2 (essentially predicting variation in school-level intercepts). Being a characteristic of each student, gender would be a level 1 predictor. Hope this helps. cheers.
HI Mike, thank you so much for your reply! I did think that education would have to be at school level so at least I was someway right! The schools have been given a road safety programme to run. I have a sample of 21 schools who did and 21 schools who did not run the programme. What I am very interested in is if there was an impact of this education on their self reported road use behaviour with a validated pedestrian scale. They also completed a road safety beliefs survey which gave an indication of how they rated their own behaviour and also how much they or others were responsible for their safety while on the road. Therefore, I would be interested to know if the education had an impact but I also need to consider the impact of their gender, whether it was an urban or rural school and also if their own beliefs also an impact on the reported behaviour. I hope that makes more sense. I really appreciate your taking the time to read this and offer advice as I suspect there will be clustering and I need to use more than just ANOVAs and linear regression. Thanks so much again! @@mikecrowson2462
This is interesting - thanks. If this is an intervention with baseline and follow up scores for behaviour, it is a repeated measures design. Does this mean that time is at level 1 and therefore student data is level 2 and school cluster is level 3? @@mikecrowson2462