assess_range_of_bias(), which allows users to assess the range of potential bias based on assumptions about how nonrespondents differ from respondents.The regression analysis types (predict_response_status_via_glm() and predict_outcome_via_glm()) now invisibly return an attribute named reference_levels. It can be retrieved using attr(x, 'reference_levels'). This is a data frame which lists, for each categorical predictor, the category used as a reference level in the regression. It is mainly useful for supporting the idcnrba package.
Minor updates:
The chisq_test_ind_response() function now returns an informative error if you do not supply any variables to the aux_vars argument. Previously it returned a data frame with zero rows and columns.
Minor code updates required to keep up with new dplyr deprecations and new warnings in *_join() functions.
Added a variable CONTACT_ATTEMPTS to the example datasets, for use in level of effort analysis.
Made the involvement survey data examples more realistic by setting values of the outcome (WHETHER_PARENT_AGREES) equal to NA for all cases whose response status isn’t equal to "Respondent".
Simplified the involvement_survey_str2s dataset by removing columns of jackknife and bootstrap replicate weights.