A review of multilevel model usage in Journal Memory

[lawsofthought]

This article is aimed at the readership of the journal Memory. We will argue that multilevel models are a valuable tool for these researchers. We are doing this on the assumption that multilevel models are both relevant and underused in the type of research described in Memory. Presumably then it would be a good idea to review Memory to see if multilevel models are used, and more importantly, how many analyses do not use multilevel models when they clearly should. We could go through every published article in a year, for example, and consider all the analyses done in all the experiments/studies done in each article. At minimum, we could classify each analysis as multilevel or not, and whether it clearly could/should have been multilevel (i,e, a multilevel analysis would have been more appropriate).

[lvjustice]

Great idea!

[lawsofthought]

Great.The journal Memory has 10 issues per year, each with around 10 or so research articles. It seems all 10 for 2017 are already out. So I think just taking all the articles published in these 2017 issues would be seen by most people as constituting a representative sample of current research interests of the Memory community. If we are going to review the analyses in each of these, it would be probably ultimately better to review more than just specifically whether they use multilevel models or not. We could also keep tally of what methods are used generally. This would be useful for us when it comes of other sections of the paper, because again we could establish that certain specific methods are underused. This information could also prove valuable for introductory material in the paper. For example, we will hopefully be able to provide hard evidence that the repertoire of statistical methods used by Memory researchers is very small (my guess is that the majority will be anovas).