data-raw/probability/README.md

In tmsalab/ecdmdata: Data Sets for Psychometric Modeling

pks Author Response to Missing Data Inquiry

We reached out to the authors of the pks package and received clarification regarding missing values in the data. We've included the response from the authors here to document why we have incorporated two distinct data sets.

From ?pks::probability:

"Of the 1127 participants eligible in the online study, 649 were excluded because they did not complete the first set of problems (p101, ..., p112) or they responded too quickly or too slowly."

The data set includes 1127 - 649 = 478 cases from the online study plus 26 cases from the lab study (where none had to be excluded). That makes 504 cases in total:

addmargins(table(probability$mode))

lab online    Sum
 26    478    504

We considered every case as valid who (1) reached Page 4 of the questionnaire (including the learning object), (2) did not have NA for all items, and (3) did not answer too quickly or too slowly. For these valid cases, remaining NA responses in the first part (p1xx) were coded as error in b1xx. For example:

addmargins(xtabs(~ is.na(p101) + b101, probability)) b101 is.na(p101) 0 1 Sum FALSE 60 409 469 TRUE 35 0 35 Sum 95 409 504

tmsalab/ecdmdata documentation built on Nov. 10, 2021, 4:45 a.m.