Conduct F test of null hypothesis LB = 0 using output from betta() or betta_random()

This function performs an F-test of a null hypothesis LB = 0 where B is a vector of p fixed effects returned by betta() or betta_random() and L is an m x p matrix with linearly independent rows.

test_submodel(
  fitted_betta,
  submodel_formula,
  method = "bootstrap",
  nboot = 1000
)

Arguments

fitted_betta	A fitted betta object -- i.e., the output of either betta() or betta_random() -- containing fixed effect estimates of interest.
submodel_formula	A formula defining which submodel to treat as the null. It is not necessary to include random effects in this formula (they will be ignored if included -- the submodel will be fit with the same random effect structure as the full model regardless of input.)
method	A character variable indicating which method should be used to estimate the distribution of the test statistic under the null.
nboot	Number of bootstrap samples to use if method = "bootstrap". Ignored if method = "asymptotic".

Value

A list containing

pval

The p-value

F_stat

The calculated F statistic

boot_F

A vector of bootstrapped F statistics if bootstrap has been used. Otherwise NULL.

References

Willis, A., Bunge, J., and Whitman, T. (2015). Inference for changes in biodiversity. arXiv preprint.

See also

betta;

betta_random;

betta;

F_test

Author

David Clausen

Examples

# generate example data
df <- data.frame(chats = c(2000, 3000, 4000, 3000,
2000, 3000, 4000, 3000), ses = c(100, 200, 150, 180,
100, 200, 150, 180),
                 Cont_var = c(100, 150, 100, 50,
                 100, 150, 100, 50),
                 Cont_var_2 = c(50,200,25,125,
                 50,200,25,125))

# fit betta()
example_fit <- betta(formula = chats ~ Cont_var + Cont_var_2, ses = ses, data = df)


# construct L for hypothesis that B_cont_var = B_cont_var_2 = 0
L <- rbind(c(0,1,0),
           c(0,0,1))

F_test_results <- F_test(example_fit, L)