I'm looking for guidance in scoring an IAT, e.g., the IAT-Race demo experiment provided by Empirisoft. Anybody that cannot provide guidance I'd greatly appreciate it!
I'm looking for guidance in scoring an IAT, e.g., the IAT-Race demo experiment provided by Empirisoft. Anybody that cannot provide guidance I'd greatly appreciate it!
Last edited by jason_reed; 12-15-2015 at 02:20 PM.
[the following reply is taken from a summary post to the SPSP mailing list by Joseph Comeau]
Hi to all of you who were interested in the DirectRT IATs and how to score them. below are the responses I got. thanks to the contributers.
Joseph Comeau
[the following edited reply is taken from a summary post to the SPSP mailing list by Joseph Comeau]
I am sure you will get a lot of replies but if you don't, call me and i will walk you through easily changing the Inquisit syntax to work with Direct RT (all you need to do is transform the variables so they are named the same as in the syntax and check a couple of procedural things in your direct rt program). My cell is *** (I am a post doc of Mahzarin Banaji's).
This all assumes that you programmed and collected data using a 7-block IAT
Dana R. Carney, Ph.D.
- Merge all of your DirectRT files with the filemerge software that Blair Jarvis (from Empirisoft) that comes included with DirectRT (save the merged file as ".csv" -- it will pull into SPSS much better than .txt).
- Pull the big huge file into SPSS.
- In DirectRT, your trial #rs must begin with #1 (and go to whatever). SO: if the range in DirectRT is 5-380, you need to transform it to range from 1-376.
- In DirectRT, your block numbers need to range from 1-7 (i recall default coding the blocks 0-6 so you will want to transform the numbers to range from 1-7)
- In DirectRT your errors vs correct responses may have been coded
differently than the syntax needs. Your errors must be coded = 0 and correct responses = 1- A critical issue is one of IAT order. the IAT syntax thinks you ran all odd subjects as order #1 (e.g., black+bad & white+good) and all even subjects as order #2 (e.g., black+good & white+bad; see syntax from the IAT SPSS syntax below).
- In the syntax file, you MUST change the order variable syntax or compute an appropriate "order variable" in your SPSS data file and then remome the part of the syntax that looks like this:
Code:*If order of the combined tasks differs in some other fashion, provide an appropriate replacement * for the following line (which works when order differs for odd and even subject numbers) . COMPUTE ORDER = mod(subj, 2) . RECODE ORDER (0=2) .- Regardless of what DirectRT named your variables, there are 5 key variables that the SPSS IAT syntax needs to read in order to
run. SO: you will want:
Code:subject variable to become --> subj block variable --> block trial variable = trial response time variable = lat correct response variable = cor- You do not need to rename anything else (looking at the syntax, you may
think you need to add additional stuff in there... you can, but it doesn't
affect how the syntax runs and is not necessary).- Attached is an SPSS IAT scoring syntax file to use.
- After running for a minute or so (you will see a bunch of SPSS files opening and closing) the output of the syntax file will be a file with a bunch of variables in it. The variables you really want to use are: D600 (in DirectRT you must add the "error penalty" in the scoring; see Greenwald et al., 2003 for an explanation). the other D variables you may wish to look at are the practice600 and the test600 (which will separate your IAT score into the practice blocks vs. the test blocks). You may also wish to look at the overall errors-- if any person has 35% or more errors, you shouldn't use their data.
- The direction of your scored IAT variable is determined by how you
programmed the experiment. Assuming you changed the ORDER variable as indicated above, the syntax scores everything to conform to blocks 3&4 minus blocks 6&7 (and, of course, smaller numbers here mean faster responding).
This should give you all you need after a good reading of Greenwald et al.'s scoring algorithm paper. Good luck!!!
Department of Psychology
Harvard University
Note: Attached is the basic IAT syntax from Tony Greenwald's site with some variable recode syntax at the top (I made some notes for the user, also). I actually used this syntax for my own data and I am pretty sure my instructions are correct (I have had to do this a few times but differently each time as I am usually helping someone else and everyone programs differently).
Last edited by jarvis24; 09-11-2007 at 07:17 AM.
[the following edited reply is taken from a summary post to the SPSP mailing list by Joseph Comeau]
Hello there,
Well this may be your lucky day! I have been using DirectRT to present several fully counterbalanced race-iats to kids in grades 1-5 (aka ch'iat, developed by Andrew Baron and Mazarin Banaji). I use SAS to import and analyze the data, and it works very well indeed. I can go from the Excel output files to SAS analysis in a matter of 30 minutes. If this is helpful to you let me know.
Barry Corenblum
Department of Psychology
Brandon University
Last edited by jarvis24; 09-25-2007 at 06:44 AM.
If this is the kind of help I can expect from Empirisoft and DirectRT users I am quite impressed!
I am a SAS junkie, so I'd love to know the procedure and code, if you're willing to share.
Thank You!
Caught me on a good day, I guess I've contacted the user who wrote the response about SAS. If/when I hear back, I'll post the SAS code if I can get my hands on it.
Appreciate it!
OK, here is the SAS code, courtesy of Barry Corenblum--he adds:
thanks Barry!What I did was to modify the sas procedure called proc import; it wasn't too hard to do, and it works well. I should point out that the proc import sas code does only one thing: it creates a sas data set from the directrt excel log file. After creating the sas data set you need to write other code to get the information you need from the imported file. If you don't need all the information in the log file you just read it in anyway, and then eliminate the data not needed. In this way you don't need to modify the import program for every application.
[edited by moderator]
I looked at the SAS code contained in sas.zip (atteched above). However, I don't really understand what this code is supposed to do.
For example, it says:
But I did not collect participants' personal information such as name and DOB. What I am trying to figure out is to calculate D score for each participant and start analyzing the IAT data.Code:/* this sas code reads in directrt exel log file, breaks up condition input into first name, last name birth month day and year, calculates age at testing and deletes input not necessary for subsequent analysis. */
Last edited by jarvis24; 11-12-2007 at 03:17 PM.
Thanks so much for posting the spss syntax to score the IAT using DirectRT data. My one question is about the format of the data file that works with this syntax. Does it assume that the file is organized such that each row of data represents one case (participant) or does it read the file as it is recorded within DirectRT (multiple rows per case)?
If it is the former, I could use some guidance as to how to transform the files. I've looked at the description in the DirectRT Help menu but I find it rather vague. Does anyone have an example of spss syntax you've used to transform the file and/or a data file that shows how this DirectRT IAT data is supposed to look before it is scored? I'm working with the 2006 version of DirectRT (I believe).
I don't but maybe one of our users could comment on this?
[edited from support email]
The recorded response time is that of the correct answer--as measured from the start of the trial (i.e., stimulus onset).I was wondering if you could confirm whether when running an IAT on DirectRT if it records the response time with a built in error. That is, if a participant hits the incorrect response key (makes an error) does that reaction time get recorded, or does the reaction time captured by the program continue until the correct response key is hit?
Last edited by jarvis24; 07-16-2008 at 08:51 AM.
Does anyone have SPSS syntax to score the Direct RT race IAT when only the test trials are recorded?
First off, thanks for all the work in creating, adapting, and modifying this syntax, and posting it for all to use!
That said, I have a question about the syntax you posted (the referenced post is shown below). Because I have a more complex coding scheme than even #ed Ss = order 1 (e.g., white + good first) and odd Ss = order 2, I merged all files from each separate order together rather than merging both orders together. In effect, it is as if I have not counterbalanced order (I have, but I will deal with this in another way) - and thus, in each merged file there is a single order of presentation. Does anyone out there know of an easy fix to the syntax so that SPSS will NOT assume that I have different orders for odd and even #ed Ss when computing the D statistics? I'm not sure what part I need to take out. It seems as if it should be simple, but in fact, the syntax is rather complex in places and I'm not sure what part needs to be changed.
Thanks!
[QUOTE=jarvis24;2239][the following edited reply is taken from a summary post to the SPSP mailing list by Joseph Comeau]
I am sure you will get a lot of replies but if you don't, call me and i will walk you through easily changing the Inquisit syntax to work with Direct RT (all you need to do is transform the variables so they are named the same as in the syntax and check a couple of procedural things in your direct rt program). My cell is *** (I am a post doc of Mahzarin Banaji's).
This all assumes that you programmed and collected data using a 7-block IAT
[LIST=1][*]Merge all of your DirectRT files with the filemerge software that Blair Jarvis (from Empirisoft) that comes included with DirectRT (save the merged file as ".csv" -- it will pull into SPSS much better than .txt).[*]Pull the big huge file into SPSS.[*]In DirectRT, your trial #rs must begin with #1 (and go to whatever). SO: if the range in DirectRT is 5-380, you need to transform it to range from 1-376.[*]In DirectRT, your block numbers need to range from 1-7 (i recall default coding the blocks 0-6 so you will want to transform the numbers to range from 1-7)[*]In DirectRT your errors vs correct responses may have been coded
differently than the syntax needs. Your errors must be coded = 0 and correct responses = 1[*]A critical issue is one of IAT order. the IAT syntax thinks you ran all odd subjects as order #1 (e.g., black+bad & white+good) and all even subjects as order #2 (e.g., black+good & white+bad; see syntax from the IAT SPSS syntax below).[*]In the syntax file, you MUST change the order variable syntax or compute an appropriate "order variable" in your SPSS data file and then remome the part of the syntax that looks like this:
[code]*If order of the combined
tasks differs in some other fashion, provide an appropriate replacement
* for the following line (which works when order differs for odd and even
subject numbers) .
COMPUTE ORDER = mod(subj, 2) .
RECODE ORDER (0=2) .
Last edited by slaurent; 02-20-2009 at 10:47 PM.
Being a total newbie to the IAT, i was wondering if anyone had a single IAT SPSS syntax to share ? I just used a single IAT in one of my experiment and i now need to look at the data and to be quite honest i don't know what to do
Hi guys, I am stuck
I am getting negative d 600 scores, is this normal? Any help will be welcome
Does anyone know a normal range for d scores? I was under the impression that they should be relatively small, and mine are pretty large. Can anyone help me or at least point me in theright direction? Thanks so much.
Hi,
is this for a single category IAT or standard IAT?
Hi,
It's been so long since I've looked at the demo that I'm not sure whether my version directly corresponds to it. (I'd check, but I no longer have DirectRT on my computer.) I've attached the kind of run file that should produce a data file that can be analyzed via the SPSS syntax file that I posted earlier. The essential elements are that blocks 3 and 5 represent the blocks of trials that are compared. For example, block 3 pairs race A with adjective 1 and race B with adjective 2. Block 5 pairs race A with adjective 2 and race B with adjective 1. Again, I'm pretty sure that this run file is either a copy of or closely corresponds to the demo file.
Just follow the instructions in the SPSS syntax file (and be sure to read the article that I mentioned earlier to understand the calculations, which are really quite basic). Good luck!
okay great! thanks for the help.
I used the syntax posted on this website to convert my IAT data into d scores (thank yo iso much for sharing). I analyzed the data several months ago and found an effect in the opposite direction from predicted. Then, today when i opened the file again and analyzed the data--the effect was gone! I re-ran the syntax front the original (raw) data and generated a new set of d-scores. those d-scores are identical to the ones that are giving me no effect except for the fact hat they have different sings (i.e., + and -). The change in the sign is not systematic and seems to vary randomly (i.e., sometimes the signs are the same and sometimes they are different from the ones giving me no effect). Even more surprisingly, the effect is now of the same magnitude as a few months ago but in the predicted direction (i.e., opposite from what I observed at first). I have re-ran the IAT syntax omn the original data several times today and I always get the same output of d-scores (but this output seems to be different from the one i got before)
Does anybody have any insight into this very peculiar and rather scary story of SPSS? Has anybody experienced the change of signs I am describing? As it stands, i cannot at all trust what I am getting form this particular study.
Please help to save my data if you can.
According to post # 3...using the syntax provided by Greenwald...
"In DirectRT, your trial #rs must begin with #1 (and go to whatever). SO: if the range in DirectRT is 5-380, you need to transform it to range from 1-376."
I believe this is because we want to drop the first two trials of each block for the original scoring method of the IAT.
When we look at the syntax file attached...the following line is used to compute the "Original IAT measure":
*Drop 1st 2 trials (this was part of the conventional procedure).
SELECT IF (TRIAL GT 2).
However, I believe there are two problems with this procedure, though I am not sure...if anyone can speak to these two potential issues, I would appreciate it:
1. This drops trials 1 and 2 from the first block (which is a single-pairing practice that isn't analyzed anyway) , but the IAT scoring procedure should drop the first two trials from EACH BLOCK.
2. The trial numbers in DirectRT do not reflect the order that these trials were presented to the subject-just the order they are programmed as-which would make this the wrong variable to reference anyway.
Also, is there any reason to actually save blocks 1,2, and 5? I don't, and besides the aforementioned problems with calculating the traditional scoring measure, this syntax appears to work just fine.
This is my first time trying to use DirectRT and I'm running into some similar issues trying to adapt this syntax so it works with the way I programmed my csv file to run in Direct RT. I've had to make several additional adjustments based on what I saved and what I didn't. I think it all could have been easier if I hadn't saved the trial blocks (as you suggested), because I've had to do some maneuvers so that the eventual data file I'm using has trials that start at 1, and I'm not even sure that matters. Anyway...
I think you're right about dropping the first two trials in each test block. Do do this, you have to add additional commands, (which will be different depending on how many trials you have) e.g. :
SELECT IF (TRIAL GE 3 AND TRIAL LE 32) OR (TRIAL GE 35).
As far as the presentation of the stimuli, "trial" does in fact tell you the order in which they were each presented. It is the "order" variable that tells you the order they were programmed. I changed "order" to "number" so I could calculate the new "order" variable. That was also kind of a pain because DirectRT didn't save any record of which of the pairings it randomly used (although maybe it could have and I just didn't click the right box), so I had to use syntax similar to the syntax above .
Hello,
Is anyone willing to share SPSS syntax for scoring the 4-block Single Category IAT? I've seen syntax on here for what appears to be a 5-block single category (or maybe single target, which is conceptually similar but slightly different in structure) IAT. Alternatively, is anyone able to help me alter the attached syntax I have to conform to the scoring guidelines outlined in Karpinski and Steinman (2006; also attached)? I would appreciate any guidance that can be given. Thanks in advance!SC-IAT Scoring Syntax.sps
I already downloaded the SPSS generic syntax for D_biep. However, when I try to score different Self-Injury IATs it gives me the same D score. Does the generix syntax D_biep needs to be edited before its use? I though it was a copy-paste thing. This is part of my doctoral thesis and honestly it desperates me to do not how to score it =/
Thank you for the help
Hello,
I'm trying to analyze my IAT Race data which I collected on DirectRT 2016, by using the run file found in their Samples folder.
Previously on this thread, contributors posted SPSS syntax for analyzing DirectRT IAT race data (attached here). Thank you for sharing this.
One important thing they mentioned is the order of the pairs (bad + African American & good + European American) that subjects saw. This syntax is specifically for where the order differs for odd and even numbered subjects. Yet, I'm not sure if the run file I found on the Sample IAT Race on DirectRT has any orders. I was wondering if anybody knows if all subjects saw the the those two pairs in the same order based on my run file (attached).
If the order is the same for all subjects, I am wondering if anybody knows how to modify the IAT Race DirectRT SPSS syntax so that I can use it to analyze data where there is no different order?
Any help is greatly appreciated!
Thank you all very much!