Millisecond Forums

Calculating Internal Consistency Reliability of the IAT step-by-step process

https://forums.millisecond.com/Topic18052.aspx

By Tiffani S. Wang-Jones - 1/7/2016

Internal Consistency for IAT from Inquisit Raw Data

I’ve had a really hard time finding the specific steps in calculating internal consistency of the IAT from the point of just downloading the IAT data to actually getting a Cronbach’s alpha.  I couldn’t find it on google, youtube, and even Inquisit support couldn’t help (which they are generally great with).  It took me forever, so once I figured it out, I’ve created this guide for others to use.

Use JMP for the analysis, not SPSS.  SPSS does not have the capabilities to restructure data sets the way that JMP does.  
But you will need SPSS to convert your Inquisit file format into one that JMP can recognize and use.
You can get a free 30 day trial of JMP, or it’s a 1 year license for students for $50

Keep in mind there are different ways to calculate internal consistency of the IAT.  I got an article suggestion from Inquisit support crew.  The one I’ve chosen is to use the difference scores for every single trial of all the combined blocks (Schnabel, Asendorpf & Greenwald, 2007).  

I’m using a 2015 MacBook Pro with SPSS 20, JMP 12.1.0, and Inquisit 4 online license.
These are all the steps from the point of downloading the data to getting cronbach's alpha:

1)    Download your raw IAT data from Inquisit
2)    Convert this data into SPSS data format (you’re not going to do your analysis in SPSS because it doesn't have the capacity to rearrange the stimulus items into variables, but you need SPSS data format to open in JMP).  If you don’t know how to convert your Inquist file into SPSS format, follow these steps a-k (otherwise skip to step 3):
a.     Open SPSS
b.     Go to “File”
c.     Go to “Read text file”
d.     Pick the IAT inquisit raw data file you just downloaded
e.     Step 1:  Keep as is, and click “continue”
f.      Step 2: Click “Yes” for Are variable names included at the top of your file.  Click continue.
g.     Step3: Keep as is, and click “continue”
h.     Step4: Unclick “Space,” keep “Tab” clicked.  Click continue.
i.      Step5: Keep as is, continue
j.      Step6: Keep as is, click Done.
k.     Open this SPSS file, you will have to manually change all the expressions variables into Numeric, Scale variables because for some reason it doesn’t register correctly. And make sure there's enough decimal places available (I put in 11)!
3)    Open JMP and open the IAT raw data that you just saved into SPSS data format
4)    Hit “Command” on keyboard and multiple select: “compatibletest1” “compatibletest2” “incompatible1” “incompatibletest2” in the “blockcode” variable (these should correspond to lines 41, 61, 141, 161).
5)    Right click on the track pad or mouse, when the pulldown options show up, pick “Select Matching Cells.”  This will select all those associated blockcodes in the file for all the combined blocks.
6)    Go to JMP menu and click “Tables” and pick “Subset,” hit enter keyboard key.  This creates a subset file of the 4-category combined tasks (the test blocks of the IAT).
7)    In the subset data file, go to JMP menu “Tables” again, and now pick “Split”.
8)    Enter the variable “stimulusitem1” into the “Split By” box
9)    Enter the variable “expressions.d” into the “Split Columns” box, then click OK or hit the enter key.
10) Now you will have a new data file with each of your stimuli items as a variable column and their difference scores as the values.  If you don’t rename your file, it pops up as “untitled.”
11) At this point, you can do your internal consistency reliability analysis because you now have restructured your data set (“untitled”) so that each stimulus item is a variable of its own with difference score values which allows you to calculate their intercorrelations. To do this analysis with your "untitled" data set…
a.     Go to JMP menu, “Analysis”, “Multivariate Methods,” “Multivariate”
b.     Select all of the columns and enter them into the “Y, Columns” box, then click “OK”
c.     A correlations table will pop up showing the intercorrelations of all the difference scores for the items in your IAT.
d.     Next, click on the little upside-down red triangle next to the “Multivariate” bar just above the “Correlations” bar of the correlations table.
e.     Select “Item Reliability” and then “Cronbach’s a”
f.      Bam! Now you have your internal consistency reliability for the entire set, as well as for each item.

That’s all! Hope this helps others because it took me a long time to figure it out!
–TSWJ
By Tiffani S. Wang-Jones - 2/5/2016

Additionally, another way to calculate internal consistency reliability is to simply look at the correlations between your "a" and "b" expressions scores (2016 email communication w/ Brian Nosek).

But since I am creating different IATs, I needed to know how to get the inter-item reliability so I had to go with the method I mentioned above. 
By Isa - 11/14/2016

Tiffani S. Wang-Jones - Thursday, January 07, 2016
Internal Consistency for IAT from Inquisit Raw Data

I’ve had a really hard time finding the specific steps in calculating internal consistency of the IAT from the point of just downloading the IAT data to actually getting a Cronbach’s alpha.  I couldn’t find it on google, youtube, and even Inquisit support couldn’t help (which they are generally great with).  It took me forever, so once I figured it out, I’ve created this guide for others to use.

Use JMP for the analysis, not SPSS.  SPSS does not have the capabilities to restructure data sets the way that JMP does.  
But you will need SPSS to convert your Inquisit file format into one that JMP can recognize and use.
You can get a free 30 day trial of JMP, or it’s a 1 year license for students for $50

Keep in mind there are different ways to calculate internal consistency of the IAT.  I got an article suggestion from Inquisit support crew.  The one I’ve chosen is to use the difference scores for every single trial of all the combined blocks (Schnabel, Asendorpf & Greenwald, 2007).  

I’m using a 2015 MacBook Pro with SPSS 20, JMP 12.1.0, and Inquisit 4 online license.
These are all the steps from the point of downloading the data to getting cronbach's alpha:

1)    Download your raw IAT data from Inquisit
2)    Convert this data into SPSS data format (you’re not going to do your analysis in SPSS because it doesn't have the capacity to rearrange the stimulus items into variables, but you need SPSS data format to open in JMP).  If you don’t know how to convert your Inquist file into SPSS format, follow these steps a-k (otherwise skip to step 3):
a.     Open SPSS
b.     Go to “File”
c.     Go to “Read text file”
d.     Pick the IAT inquisit raw data file you just downloaded
e.     Step 1:  Keep as is, and click “continue”
f.      Step 2: Click “Yes” for Are variable names included at the top of your file.  Click continue.
g.     Step3: Keep as is, and click “continue”
h.     Step4: Unclick “Space,” keep “Tab” clicked.  Click continue.
i.      Step5: Keep as is, continue
j.      Step6: Keep as is, click Done.
k.     Open this SPSS file, you will have to manually change all the expressions variables into Numeric, Scale variables because for some reason it doesn’t register correctly. And make sure there's enough decimal places available (I put in 11)!
3)    Open JMP and open the IAT raw data that you just saved into SPSS data format
4)    Hit “Command” on keyboard and multiple select: “compatibletest1” “compatibletest2” “incompatible1” “incompatibletest2” in the “blockcode” variable (these should correspond to lines 41, 61, 141, 161).
5)    Right click on the track pad or mouse, when the pulldown options show up, pick “Select Matching Cells.”  This will select all those associated blockcodes in the file for all the combined blocks.
6)    Go to JMP menu and click “Tables” and pick “Subset,” hit enter keyboard key.  This creates a subset file of the 4-category combined tasks (the test blocks of the IAT).
7)    In the subset data file, go to JMP menu “Tables” again, and now pick “Split”.
8)    Enter the variable “stimulusitem1” into the “Split By” box
9)    Enter the variable “expressions.d” into the “Split Columns” box, then click OK or hit the enter key.
10) Now you will have a new data file with each of your stimuli items as a variable column and their difference scores as the values.  If you don’t rename your file, it pops up as “untitled.”
11) At this point, you can do your internal consistency reliability analysis because you now have restructured your data set (“untitled”) so that each stimulus item is a variable of its own with difference score values which allows you to calculate their intercorrelations. To do this analysis with your "untitled" data set…
a.     Go to JMP menu, “Analysis”, “Multivariate Methods,” “Multivariate”
b.     Select all of the columns and enter them into the “Y, Columns” box, then click “OK”
c.     A correlations table will pop up showing the intercorrelations of all the difference scores for the items in your IAT.
d.     Next, click on the little upside-down red triangle next to the “Multivariate” bar just above the “Correlations” bar of the correlations table.
e.     Select “Item Reliability” and then “Cronbach’s a”
f.      Bam! Now you have your internal consistency reliability for the entire set, as well as for each item.

That’s all! Hope this helps others because it took me a long time to figure it out!
–TSWJ

Hi, 
I find this very interesting and useful, so thank you so much for sharing.
Can you specify the title of the article you cited (Schnabel, Asendorpf & Greenwald, 2007), please? I can not find it.