survey_detect_qtypes: Detect question types

In DataS-DHSC/consultations: Re-usable functions for describing the responses to a public consultation or call for evidence

Description

Looks at the possible answers in a column, and determines which type of question is most likely. This is based on rules of thumb and can be tuned for different uses using the calculation criteria.

Usage

survey_detect_qtypes(
  response_col,
  unique_vals = 20,
  split_perc = 0.5,
  prop_total = 0.55,
  prop_common = 0.05
)

Arguments

response_col	column from response dataframe
unique_vals	integer
split_perc	float
prop_total	float
prop_common	float

Details

Calculation criteria can be set by the user: unique_vals: what's the maximum number of unique values a column can have, to always consider it categorical? split_perc: determines which questions are deemed multi-choice: if the answers are split by comma, how much fewer unique values should that column have than before splitting? Helps distinguish between free-text and multi-choice. prop_total: if the 5 most common values account for this proportion of the total number of answers, consider it categorical prop_common: if the most common response is at most this proportion of the total, consider it free text

Value

a string value

Examples

apply(dummy_response, 2, survey_detect_qtypes)

DataS-DHSC/consultations documentation built on Jan. 28, 2022, 1:56 a.m.