Hello!
I am interested in understanding change over time in a dataset that contains individual survey respondents nested in country units.
In the past, I have run MLM with individuals at level 1 and country at level 2 with no consideration for change over time. My question is, do I enter 'time' (survey rounds for the same countries) as level 3 (and individual at level 1 and country at level 3) or time as level 1 (and individual at level 2 and country at level 3)?
Any help would be greatly appreciated!
In your example time would be level 1 (with individual at level 2 and country at level 3). Repeated measures are nested within individuals and individuals are nested within countries.
This may be of use:
https://stats.oarc.ucla.edu/other/mult-pkg/seminars/mlm-longitudinal/
Also for general resources on multilevel modelling (including longitudinal) the Bristol University Centre for Multilevel Modelling has a lot of useful stuff:
https://www.bristol.ac.uk/cmm/
The key thing to remember is to set up your data in 'long format' - i.e. one row per level 1 (time) measure (so if there were three time points there would be three rows per individual with the different time related measures.
Brian Kelly Thank you so much for the response! Would this type of analysis be restricted to panels only or could I also use survey rounds in which each round of respondents was drawn from the same populations (using the same sampling methods and frame)? My research interest is primarily in between country differences (within round) and between round differences by country.
I will be sure to look into the link you kindly shared with me!
All the best
If I understand correctly your data is not repeat measures of the same individuals (longitudinal data), but rather repeat cross sectional analysis (repeated samples from the same population, but not the same individuals).
If this is the case (that you have repeated samples of the same population), then it would not be possible to do a multilevel model with time at level 1 - because you do not have measures that relate to the same person over time (the repeated measures are not nested within individuals, rather they are repeated cross sectional measures of a population). If this is the case then you cannot model individual trajectories of change, you can only model aggregate levels of change.
So, in summary: repeated cross-sectional data are created where a survey is administered to a new sample of interviewees at successive time points. Because repeated cross-sectional data take a different sample of a population over time, they are used for analysing population or group changes over time (also known as aggregate change over time). They cannot be used to look at individual change.
It may be that if you are looking at between country differences at each time, and within country differences over time, then your focus is not on individual change anyway.
I have attached a document that discusses both repeated cross sectional analysis and longitudinal data analysis which may be of use.
Thank you so much, Brian. This is very much appreciated. I will take a look at your attached materials. Thank you for taking the time, you have helped me a lot! Have a great day!
- Badges
- Science topic
- Similar topics
- Philosophy
- Esthetics
- Interpretation