{\rtf1\ansi\ansicpg1252\cocoartf1671\cocoasubrtf500
{\fonttbl\f0\fswiss\fcharset0 Helvetica;}
{\colortbl;\red255\green255\blue255;}
{\*\expandedcolortbl;;}
\margl1440\margr1440\vieww10800\viewh8400\viewkind0
\pard\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx6480\tx7200\tx7920\tx8640\pardirnatural\partightenfactor0
\f0\fs24 \cf0 \
Data and results from our experiment is contained in Face SDT.RData\
\
There are 4 datasets: Gender_Data, OddOneOut_Data, Expression_Data and Combined_Data\
\
For each dataset, rows are persons and columns are items.\
For the first 3 datasets, 64 item measures and 100 person measures are estimated.\
The first 50 rows correspond to the first 50 subjects in the \'93without magnification\'94 condition.\
Rows 51-100 correspond to the first 50 subjects in the \'93with magnification\'94 condition.\
For the combined dataset there are 3x64 = 192 items instead of 64.\
\
The results of our d prime analysis are stored in: Gender_dPrimes, OddOneOut_dPrimes, Expression_dPrimes, Combined_dPrimes and Combined_dPrimes_EM\
\
All except Combined_dPrimes_EM were estimated using only equations 4 and 5.\
\
Combined_dPrimes_Rasch contains estimates of the dichotomous Rasch model.\
\
\
Code used to estimate d primes are stored in dprime.R \
Code for the Rasch model is in Rasch_dichotomous.R\
\
There are 2 main functions in dprime.R: dprime_item_person_measures() which implements equations 4 and 5, and dprime_EM() which implements Expectation Maximization. These functions can be called as follows: \
\
A = dprime_item_person_measures(Correct, Trials, M_AFC) \
A = dprime_EM(Correct, Trials, M_AFC, Thresh) \
\
where \'93Correct\'94 is a matrix of correct responses, \'93Trials\'94 is a matrix of total number of trials, and \
\'93M_AFC\'94 is a vector of the number of response alternatives per item. Both \'93Trials\'94 and \'93M_AFC\'94 can be a single number if they are the same for all person-item combinations. For dprime_EM the input \'93Thresh\'94 is the threshold in d prime units that the mean absolute difference between successive iterations of item and person measures must be below before the EM algorithm stops.\
\
Examples:\
A = dprime_item_person_measures(Gender_Data, 1, 2) for the gender data\
A = dprime_item_person_measures(OddOneOut_Data, 1, 3) for the odd one out data, etc\'85\
\
or..\
\
A = dprime_EM(Gender_Data, 1, 2, 0.02) for EM with the Gender data\
\
For the combined data, there is a vector called \'93Combined_mAFC\'94 that should be used as input:\
A = dprime_item_person_measures(Combined_Data, 1, Combined_mAFC)\
\
The outputs are:\
A$item_measures\
A$item_std_errors\
A$person_measures\
A$person_std_errors\
\
Rasch_dichotomous.R contains code for the dichotomous Rasch model. \
A = Rasch_dichotomous(Gender_Data)\
}
