A generalized linear mixed model (GLMM) was used to look at the effects on Twitter activity on citations, while looking at possible confounds such as impact factor and time since publication. Citation rates were modeled as a random variable of a negative binomial distribution. An information theoretic framework was used to compare Twitter activity, impact factor, and time since publication on the citation rates. Specifically, the dredge function was used in R to calculate weights for GLMMs of possible combinations of the variables. Weights were summed and those with greater sums were considered to be more important.

Mixed effects models were also used to look at the effects of impact factor on Twitter activity.