To ask my questions, I need to set up a hypothetical experimental design.
Let's say that I am interested in participant evaluations of art made by two different people (famous artist A and unknown artist B)...each artist has painted 10 paintings in the same 10 different colors/styles (e.g. one blue sad painting, one red angry painting, one yellow happy painting, etc). Thus, I now have 10 pairs of incredibly similar paintings, half from each artist.
Under the assumption that participants can't tell which paintings came from which artist (pretested), I am now curious as to whether or not using the name of each artist as a label will influence participant evaluations of skill on a 1-7 Likert scale. Thus, half my participants see the true labels and half see false (totally reversed) labels.
Here are my questions:
What kind of multi-level model (if any) would be most appropriate in this setting? Nesting within individual? Nesting within pairing (e.g. red pair, blue pair, green pair)?
Please correct me if I am wrong as I am very new to multi-level analysis. So far, it seems like the answer is to try both types of nesting using a random intercept and randoms slope?
One answer:
I would do 4 separate groups, each showing both similar paintings next to each other where participants had to order which they preferred.
1: no names
2: randomize the name with the painting
3: inverse the names
4: correct names
Have the same amount of people look through each and then identify who got more favorable rankings. The data could be broken up a lot then, did the name influence, did the same paintings seem to be favored, did it matter which painting was on top or bottom/ the order shown?
Mulitilevel model is A structure based on many different concentriclevels which can be used to construct many different shapes.
Multilevel models are particularly appropriate forresearch designs where data for participants are organized at more than one level (i.e., nested data). The units of analysis are usually individuals (at a lower level) who are nested within contextual/aggregate units (at a higher level).
Multilevel models recognise the existence of such data hierarchies by allowing for residual components at each level in the hierarchy. For example, a two-level model which allows for grouping of child outcomes within schools would include residuals at the child and school level. Thus the residual variance is partitioned into a between-school component (the variance of the school-level residuals) and a within-school component (the variance of the child-level residuals). The school residuals, often called ‘school effects’, represent unobserved school characteristics that affect child outcomes. It is these unobserved variables which lead to correlation between outcomes for children from the same school.