1: \begin{abstract}

2: Linear mixed models with large imbalanced crossed random effects structures

3: pose severe computational problems for maximum likelihood estimation

4: and for Bayesian analysis. The costs can grow as fast as $N^{3/2}$ when there are

5: $N$ observations.  Such problems arise in any setting where the underlying

6: factors satisfy a many to many relationship (instead of a nested one)

7: and in electronic commerce applications, the $N$ can be quite large.

8: Methods that do not account for the correlation structure can greatly underestimate uncertainty.

9: We propose a method of moments approach that takes account of the correlation

10: structure and that can be computed at $O(N)$ cost.

11: The method of moments is very amenable to parallel computation

12: and it does not require parametric distributional assumptions, tuning parameters

13: or convergence diagnostics.

14: For the regression coefficients, we give conditions for consistency and

15: asymptotic normality as well as a consistent variance estimate.

16: For the variance components, we give conditions for consistency and

17: we use consistent estimates of a mildly conservative variance estimate.

18: All of these computations can be done in $O(N)$ work.

19: We illustrate the algorithm with some data from Stitch Fix where the crossed random effects

20: correspond to clients and items. 

21: \end{abstract}

22: