* This SPSS script wwas designed to analyze IAT data produced * gathered using the basic IAT Inquisit sample script, 'IAT.exp'. * It was adapted from Tony Greenwald's Generic IAT SPSS syntax. * The analysis computes D measures with the built-in error * penalty as described in: * Greenwald, A. G, Nosek, B. A., & Banaji, M. R. (2003). Understanding * and using the Implicit Association Test: I. An improved scoring algorithm. * Journal of Personality and Social Psychology, 85, 197-216. * TODO: change IAT.dat to the name of your data file. GET TRANSLATE FILE = 'pictureiat.dat' /TYPE=TAB /MAP /FIELDNAMES . VALUE LABELS blocknum 1 'Good-Bad practice' 2 'Attribute categories practice' 3 'First pairing practice' 5 'First pairing test' 6 'Reversed target categories pract' 7 'Second pairing practice' 9 'Second pairing test' . IF (MOD(subject,2) = 1) ORDER = 1 . IF (MOD(subject,2) = 1) ORDER = 1 . IF (MOD(subject,2) = 0) ORDER = 2 . IF (MOD(subject,2) = 0) ORDER = 2 . COMPUTE PAIRING = 0. IF ((ORDER=1) and (blocknum=3|blocknum=5)) PAIRING = 1. IF ((ORDER=1) and (blocknum=7|blocknum=9)) PAIRING = 2. IF ((ORDER=2) and (blocknum=7|blocknum=9)) PAIRING = 1. IF ((ORDER=2) and (blocknum=3|blocknum=5)) PAIRING = 2. COMPUTE TEST = 0. IF (blocknum=3|blocknum=7) TEST = 1. IF (blocknum=5|blocknum=9) TEST = 2. VALUE LABELS TEST 0 'single-task practice' 1 '1st combined block' 2 '2nd combined block' / correct 0 'error' 1 'correct' . VARIABLE LABELS correct "0=error, 1=correct" / blocknum "block number" / trialnum "trial number" / ORDER "order of combined tasks" / PAIRING "paired categories" / TEST "1st or 2nd combined block". ** TO DO: Replace the labels below with your targets and attributes. VALUE LABELS PAIRING 0 'single task practice' 1 'good+INSECT and bad+FLOWER' 2 'good+FLOWER and bad+INSECT' / ORDER 1 'good+INSECT and bad+FLOWER first' 2 'good+FLOWER and bad+INSECT first' . COMPUTE error = 100*(1 - correct). * Save data to re-use for computing other measures . SAVE OUTFILE = 'temp.sav' . DESCRIPTIVES ALL . SUMMARIZE TABLES latency BY TEST BY PAIRING. GET FILE = 'temp.sav' . * Use data for Blocks 3,5,7, and 9 (i.e., TEST = 1 or 2) . SELECT IF (TEST=1 OR TEST=2) . * Mark trials with latency < 300 ms (for possible subject discard) . COMPUTE FLAG_300 = 0 . IF (LATENCY < 300) FLAG_300 = 1 . * Mark trials with latency < 400 ms (to provide a count of these) . COMPUTE FLAG_400 = 0 . IF (LATENCY < 400) FLAG_400 = 1 . * Mark trials with latency > 10000 ms (to provide a count of these) . COMPUTE FLAG_10K = 0 . IF (LATENCY GT 10000) FLAG_10K = 1 . * Following line can be used to check frequencies of fast & slow responses & errors . FREQUENCIES FLAG_300 FLAG_400 FLAG_10K ERROR . * Record criteria for potential use in subject discards . * This count can later be used as the basis for subject discard . AGGREGATE OUTFILE = 'CRITERIA.SAV' / BREAK = SUBJECT ORDER / PCT_300 = PGT(FLAG_300,0) / PCT_400 = PGT(FLAG_400,0) / PCT_10K = PGT(FLAG_10K,0) / AVELTNCY ERRORPCT = MEAN (LATENCY ERROR) / NTRIALS = N . * Drop trials slower than 10000 ms for LATENCY . SELECT IF (LATENCY LE 10000) . * The following line deletes latencies less than 300. To prevent these trials from being * discarded, add an asterisk before the next line to comment it out. * SELECT IF (LATENCY GE 300) . DESCRIPTIVES ALL. * The following aggregate command computes block means and SDs needed for the D measures. AGGREGATE OUTFILE = * / BREAK = SUBJECT BLOCKNUM ORDER PAIRING TEST / MEAN_LAT MEAN_ERR = MEAN(latency error) / SD_LAT = SD(latency) / NTRIALS = N . DESCRIPTIVES ALL. *Save this as a record of this stage of analysis . SAVE OUTFILE = 'BIEP.ALL_MEANS&SDs.SAV' . IF (PAIRING = 1) N1 = NTRIALS . IF (PAIRING = 2) N2 = NTRIALS . IF (PAIRING = 1) M1 = MEAN_LAT . IF (PAIRING = 2) M2 = MEAN_LAT . IF (PAIRING = 1) ERR1 = MEAN_ERR . IF (PAIRING = 2) ERR2 = MEAN_ERR . IF (PAIRING = 1) SD1 = SD_LAT . IF (PAIRING = 2) SD2 = SD_LAT . AGGREGATE OUTFILE = * / BREAK = SUBJECT ORDER TEST / M1 M2 ERR1 ERR2 N1 N2 SD1 SD2 = FIRST(M1 M2 ERR1 ERR2 N1 N2 SD1 SD2) . *Save this as a record of this stage of analysis . SAVE OUTFILE = 'BIEP.MEANS&SDS.SAV' . IF(TEST=1) ERR1a = ERR1 . IF(TEST=1) ERR2a = ERR2 . IF(TEST=2) ERR1b = ERR1 . IF(TEST=2) ERR2b = ERR2 . * These are the numerator copmponents in millisecond units. IF(TEST=1) M1a = M1 . IF(TEST=1) M2a = M2 . IF(TEST=2) M1b = M1 . IF(TEST=2) M2b = M2 . IF(TEST=1) SD1a = SD1 . IF(TEST=1) SD2a = SD2 . IF(TEST=2) SD1b = SD1 . IF(TEST=2) SD2b = SD2 . IF(TEST=1) N1a = N1 . IF(TEST=1) N2a = N2 . IF(TEST=2) N1b = N1 . IF(TEST=2) N2b = N2 . COMPUTE D_asis_num = M2 - M1. *Use SD based on all responses (StanDevX) as denominator for D_biep . COMPUTE D_asis_denom = SQRT( ( ((N1-1) * SD1**2 + (N2-1) * SD2**2) + ((N1+N2) * ((M1-M2)**2) / 4) ) / (N1 + N2 - 1) ) . IF (TEST=1) D_BIEP_a = D_asis_num / D_asis_denom . IF (TEST=2) D_BIEP_b = D_asis_num / D_asis_denom . IF (TEST=1) Na = N1+N2 . IF (TEST=2) Nb = N1+N2 . DESCRIPTIVES ALL . AGGREGATE OUTFILE = * / BREAK = SUBJECT ORDER / D_biep_a D_biep_b M1a M2a M1b M2b ERR1a ERR2a ERR1b ERR2b Na Nb = FIRST(D_biep_a D_biep_b M1a M2a M1b M2b ERR1a ERR2a ERR1b ERR2b Na Nb). VARIABLE LABELS D_biep_a 'good+INSECT association 1st blocka' / D_biep_b 'good+INSECT association 2nd blocka' / Na '# trials, 1st combined blocks' / Nb '# trials, 2nd combined blocks' / M1a 'Mn Lat. 1st block, pairing 1' / M2a 'Mn Lat. 1st block, pairing 2' / M1b 'Mn Lat. 2nd block, pairing 1' / M2b 'Mn Lat. 2nd block, pairing 2' / ERR1a 'Error % 1st block, pairing 1' / ERR2a 'Error % 1st block, pairing 2' / ERR1b 'Error % 2nd block, pairing 1' / ERR2b 'Error % 2ns block, pairing 2' / SUBJECT 'Session ID' . COMPUTE ERR_1 = (ERR1a + ERR1b) / 2 . COMPUTE ERR_2 = (ERR2a + ERR2b) / 2 . VARIABLE LABELS ERR_1 'Error % for Pairing 1 (both combined tasks)' / ERR_2 'Error % for Pairing 2 (both combined tasks)' . *Compute unweighted combination of measures based on first and second blocks of combined tasks. * !! Do this even if there are fewer trials in the first block because the * first block has been found to have as good or better validity, even with * fewer trials (see Greenwald, Nosek, & Banaji, JPSP 2003) . COMPUTE D_biep = (D_biep_a + D_biep_b) / 2 . VARIABLE LABELS D_biep 'good+INSECT association all blocks' . ************************************************************* ** TO REVERSE THE SCORING OF THE IAT, REMOVE THE ASTERISKS ** ** FROM THE NEXT FOUR COMMAND LINES. ** ************************************************************* . * COMPUTE D_biep = 0 - D_biep . * COMPUTE D_biep_a = 0 - D_biep_a . * COMPUTE D_biep_b = 0 - D_biep_b . * VARIABLE LABELS D_biep 'good+FLOWER association all blocks' / D_biep_a 'good+FLOWER association 1st blocka' / D_biep_b 'good+FLOWER association 2nd blocka' . FORMAT D_biep D_BIEP_a D_BIEP_b D_BIEP (F6.3) SUBJECT (F7.0) Na Nb (F4.0) M1a M2a M1b M2b (F7.1) ERR1a ERR2a ERR1b ERR2b ERR_1 ERR_2 (F5.1) . SAVE OUTFILE = 'D_BIEP_temp.SAV' . DESCRIPTIVES ALL . **************************************************************. **** STEP 3: COMBINE MEASURES AND SAVE ****. **************************************************************. *Add labels to CRITERIA.SAV . GET FILE = 'CRITERIA.SAV' . VARIABLE LABELS PCT_300 '% latencies below 300 ms' / PCT_400 '% latencies below 400 ms' / PCT_10K '% latencies above 10,000 ms' / AVELTNCY 'Overall average latency (ms)' / ERRORPCT 'Overall % errors' / NTRIALS 'Total # of combined-task trials' . FORMAT AVELTNCY (F7.1) PCT_300 PCT_400 PCT_10K ERRORPCT (F6.1) SUBJECT (F7.0) . SAVE OUTFILE = 'CRITERIA.SAV' . *Combine and save . MATCH FILES FILE = 'D_BIEP_temp.SAV' / FILE = 'CRITERIA.SAV' /BY SUBJECT . SAVE OUTFILE = 'D_BIEP.SAV' . * Note: No subjects have yet been discarded. However, each record contains . * three variables that might be used for discard . * PCT_300 = overall % of responses with latencies below 300 ms . * ERRORPCT = overall error % . * AVELTNCY = overall average latency . * [Note that 'aveltncy' was computed _BEFORE_ discard of latencies above 10000 ms] . GET FILE = 'D_BIEP.SAV' . DESCRIPTIVES ALL .