CREDI Multi-dimensional Item Factor Analysis Scoring

Jonathan Seiden 7/22/2021

Introduction

The Caregiver Reported Early Development Instruments is a measure of early development that is administered through an interview or survey completion by a child’s parent or primary caregiver. This program is designed to process the raw response data to calculate scores.

This package produces several scores based on input data, and supports both data collected on the Long Form and Short Form versions of CREDI. Most users will prefer to use the online scoring app which does not require the use of R in order to calculate scores, but this package is made publicly available to advanced users and those who wish to understand the multi-dimensional factor analysis process used to calculate Overall and Domain scores.

See the CREDI scoring manual for more details about score calculation.

Overview

Most users will only ever use the score function of the CREDI package. This is the only function in the package that is exported and converts raw item responses into an Overall, Cognitive, Motor, Language, and Social-Emotional scale scores which exhibit interval properties. In addition to these scale scores, the function produces normalized reference Z-scores and estimates of the standard error of measurement.

Standardized reference scores normalize by age and show developmental status in comparison to a reference group with advantageous home environments.

While only the score function is exported, internal functions for cleaning and calculating the posterior density functions that power the CREDI scoring procedure are well commented but and have minimal documentation. These other functions are not designed for end users are provided for reference and explanation.

Installation

Installing the CREDI scoring app is currently only available through GitHub. Easily download and install with the below lines of code:

require(devtools)
devtools::install_github("https://github.com/marcus-waldman/credi")

Example

Below is a very simple example with three simulated children, each of whom only had 5 Long Form questions. Note that one child is missing a value for LF1, resulting in no score being calculated for the child because the min_items parameter was set to 5.

library(credi)

#Create a sample dataframe
dat <- data.frame(
  ID = 1:3,
  AGE = c(3, 5, 4),
  LF1 = c(1, 0, NA),
  LF2 = c(0, 0, 0),
  LF3 = c(1, 0, 1),
  LF4 = c(1, 1, 1),
  LF5 = c(1, 0, 0)
)

#Score the dataframe
scored_dat <- credi::score(
  data = dat,
  reverse_code = FALSE,
  interactive = FALSE,
  min_items = 5
)

## 
## Scoring  2  observations:
##   |                                                                              |                                                                      |   0%  |                                                                              |===================================                                   |  50%  |                                                                              |======================================================================| 100%

#Print out domain scores:
scored_dat$scores[, c("MOT", "LANG", "SEM", "COG", "OVERALL")]

##      MOT   LANG    SEM    COG OVERALL
## 1 43.489 45.968 44.626 45.091  40.079
## 2 42.058 45.049 43.755 44.250  38.160