SPSSX Discussion

Frequency analysis

Classic

List

Threaded

14 messages Options

Rose, Fred

Frequency analysis

Frequency analysis This is one of those questions that seems so ridiculously simple that I’m embarrassed to ask it, but I’m drawing a blank.

I’ve got nominal data (yes/no) from a sample of men and women doing a variant of the Roediger and McDermott False Memory paradigm. The yes/no responses relate to whether or not the individual indicated that they “remembered” a critical word that, in fact, was never presented (the study was more complicated than this, but I’m describing the specific analysis I’m working on now). The distribution is roughly 55% yes to 45% no. I am trying to figure out the proper test of whether this pattern deviates from chance. I’m not comparing men and women, but just over all whether this is a high rate of false recognition in the sample as a whole.

Any kind soul willing to help a guy out?

Thanks,

Fred
--
Fredric E. Rose, Ph.D.
Associate Professor of Psychology
Palomar College
760-744-1150 x2344
frose@...
====================To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD

Bruce Weaver

Re: Frequency analysis

Administrator

Rose, Fred wrote

Frequency analysis

This is one of those questions that seems so ridiculously simple that I’m embarrassed to ask it, but I’m drawing a blank.

I’ve got nominal data (yes/no) from a sample of men and women doing a variant of the Roediger and McDermott False Memory paradigm. The yes/no responses relate to whether or not the individual indicated that they “remembered” a critical word that, in fact, was never presented (the study was more complicated than this, but I’m describing the specific analysis I’m working on now). The distribution is roughly 55% yes to 45% no. I am trying to figure out the proper test of whether this pattern deviates from chance. I’m not comparing men and women, but just over all whether this is a high rate of false recognition in the sample as a whole.

Any kind soul willing to help a guy out?

Thanks,

Fred
--
Fredric E. Rose, Ph.D.
Associate Professor of Psychology
Palomar College
760-744-1150 x2344
[hidden email]

====================To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

--
Bruce Weaver
bweaver@lakeheadu.ca
http://sites.google.com/a/lakeheadu.ca/bweaver/

"When all else fails, RTFM."

PLEASE NOTE THE FOLLOWING:
1. My Hotmail account is not monitored regularly. To send me an e-mail, please use the address shown above.
2. The SPSSX Discussion forum on Nabble is no longer linked to the SPSSX-L listserv administered by UGA (https://listserv.uga.edu/).

Rose, Fred

Re: Frequency analysis

Re: Frequency analysis That is a definite issue – I’m not sure of the correct chance proportions would be. For those not familiar with the task, participants read a list of words such as “moth insect wing bird fly bug cocoon color net” and ask recall or recognition. The false memory occurs when a participant falsely remembers the critical word – in this case, “butterfly”. I don’t think it’s correct to say that 50% is the correct chance proportion but I’m not sure what a correct value would be. It is likely that a 55% false memory rate is higher than false alarm rate on a word list task where the list consists of unrelated words, but again, I’m not sure where to go with it. The bottom line is that I would like to be able to claim that the 55% rate reflects a strong tendency for false memory to occur.

On 4/18/12 1:49 PM, "Bruce Weaver" <bruce.weaver@...> wrote:

It sounds like you're looking for the binomial test. It's not clear what
proportions you expect in the two categories by chance, but if it was 50%,
the command would look like this (where "recalled" is an indicator variable
for recall, 0=No, 1=Yes):

NPAR TESTS
/BINOMIAL (0.50)=recalled
/MISSING ANALYSIS.

HTH.

Rose, Fred wrote
>
> Frequency analysis
>
>
> This is one of those questions that seems so ridiculously simple that
> I’m embarrassed to ask it, but I’m drawing a blank.
>
> I’ve got nominal data (yes/no) from a sample of men and women doing
> a variant of the Roediger and McDermott False Memory paradigm.  The
> yes/no responses relate to whether or not the individual indicated that
> they “remembered” a critical word that, in fact, was never
> presented (the study was more complicated than this, but I’m
> describing the specific analysis I’m working on now).  The
> distribution is roughly 55% yes to 45% no.  I am trying to figure out
> the proper test of whether this pattern deviates from chance.
>  I’m not comparing men and women, but just over all whether
> this is a high rate of false recognition in the sample as a whole.  
>
> Any kind soul willing to help a guy out?
>
> Thanks,
>
> Fred
> --
> Fredric E. Rose, Ph.D.
> Associate Professor of Psychology
> Palomar College
> 760-744-1150 x2344
> frose@
>
>
>
>
> ====================To manage your subscription to SPSSX-L, send a message
> to
> LISTSERV@.UGA (not to SPSSX-L), with no body text except the
> command. To leave the list, send the command
> SIGNOFF SPSSX-L
> For a list of commands to manage subscriptions, send the command
> INFO REFCARD
>

-----
--
Bruce Weaver
bweaver@...
http://sites.google.com/a/lakeheadu.ca/bweaver/

"When all else fails, RTFM."

NOTE: My Hotmail account is not monitored regularly.
To send me an e-mail, please use the address shown above.

--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Frequency-analysis-tp5650112p5650183.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

=====================
To manage your subscription to SPSSX-L, send a message to
LISTSERV@... (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

--
Fredric E. Rose, Ph.D.
Associate Professor of Psychology
Palomar College
760-744-1150 x2344
frose@...
====================To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD

Poes, Matthew Joseph

Re: Frequency analysis

I think this has become over-complicated. While the non-parametric binomial T test can be used, so can a normal one sample t-test when yes and no is dummy coded. Adjusting the cut point of the distribution proportion is not relevant without a-priori information, which seems to be missing here.

I would say this, if you were to find through 100’s or 1000’s of trials that completely guessing on this exact study content yielded a non-balanced yes to no ratio (for some reason), then using this a-priori in future t-tests would make sense. If you don’t know this information, or even if you had a guess but it wasn’t well replicated, I would just leave it 50/50. If you believe the response is biased due to a known factor, then you need to switch to a more complex approach modeling the probability of a Yes given a certain factor, and include this. However, the question seemed far more basic, and a t-test or NP binomial t-test will each work well (the t-test will be more powerful unless the ratio is highly biased to one side).

Matthew J Poes

Research Data Specialist

Center for Prevention Research and Development

University of Illinois

510 Devonshire Dr.

Champaign, IL 61820

Phone: 217-265-4576

email: [hidden email]

From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Rose, Fred
Sent: Wednesday, April 18, 2012 4:32 PM
To: [hidden email]
Subject: Re: Frequency analysis

That is a definite issue – I’m not sure of the correct chance proportions would be. For those not familiar with the task, participants read a list of words such as “moth insect wing bird fly bug cocoon color net” and ask recall or recognition. The false memory occurs when a participant falsely remembers the critical word – in this case, “butterfly”. I don’t think it’s correct to say that 50% is the correct chance proportion but I’m not sure what a correct value would be. It is likely that a 55% false memory rate is higher than false alarm rate on a word list task where the list consists of unrelated words, but again, I’m not sure where to go with it. The bottom line is that I would like to be able to claim that the 55% rate reflects a strong tendency for false memory to occur.

On 4/18/12 1:49 PM, "Bruce Weaver" <bruce.weaver@...> wrote:

It sounds like you're looking for the binomial test. It's not clear what
proportions you expect in the two categories by chance, but if it was 50%,
the command would look like this (where "recalled" is an indicator variable
for recall, 0=No, 1=Yes):

NPAR TESTS
/BINOMIAL (0.50)=recalled
/MISSING ANALYSIS.

HTH.

Rose, Fred wrote
>
> Frequency analysis
>
>
> This is one of those questions that seems so ridiculously simple that
> I’m embarrassed to ask it, but I’m drawing a blank.
>
> I’ve got nominal data (yes/no) from a sample of men and women doing
> a variant of the Roediger and McDermott False Memory paradigm.  The
> yes/no responses relate to whether or not the individual indicated that
> they “remembered” a critical word that, in fact, was never
> presented (the study was more complicated than this, but I’m
> describing the specific analysis I’m working on now).  The
> distribution is roughly 55% yes to 45% no.  I am trying to figure out
> the proper test of whether this pattern deviates from chance.
>  I’m not comparing men and women, but just over all whether
> this is a high rate of false recognition in the sample as a whole.  
>
> Any kind soul willing to help a guy out?
>
> Thanks,
>
> Fred
> --
> Fredric E. Rose, Ph.D.
> Associate Professor of Psychology
> Palomar College
> 760-744-1150 x2344
> frose@
>
>
>
>
> ====================To manage your subscription to SPSSX-L, send a message
> to
> [hidden email] (not to SPSSX-L), with no body text except the
> command. To leave the list, send the command
> SIGNOFF SPSSX-L
> For a list of commands to manage subscriptions, send the command
> INFO REFCARD
>

-----
--
Bruce Weaver
bweaver@...
http://sites.google.com/a/lakeheadu.ca/bweaver/

"When all else fails, RTFM."

NOTE: My Hotmail account is not monitored regularly.
To send me an e-mail, please use the address shown above.

--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Frequency-analysis-tp5650112p5650183.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

=====================
To manage your subscription to SPSSX-L, send a message to
LISTSERV@... (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

--
Fredric E. Rose, Ph.D.
Associate Professor of Psychology
Palomar College
760-744-1150 x2344
frose@...

======= To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD

Len Vir

Re: Frequency analysis

In reply to this post by Rose, Fred

Since 'butterfly' does not appear in the list, does not even a 15 percent
rate of recall constitute a 'strong tendency for false memory'?

I would have thought that to talk of 'chance proportion' the term butterfly
would have had to appear in the list.

But I am probably in the boondocks...!

Len Vir

On Wed, Apr 18, 2012 at 5:32 PM, Rose, Fred <[hidden email]> wrote:

> That is a definite issue – I’m not sure of the correct chance proportions
> would be. For those not familiar with the task, participants read a list of
> words such as “moth insect wing bird fly bug cocoon color net” and ask
> recall or recognition. The false memory occurs when a participant falsely
> remembers the critical word – in this case, “butterfly”. I don’t think it’s
> correct to say that 50% is the correct chance proportion but I’m not sure
> what a correct value would be. It is likely that a 55% false memory rate is
> higher than false alarm rate on a word list task where the list consists of
> unrelated words, but again, I’m not sure where to go with it. The bottom
> line is that I would like to be able to claim that the 55% rate reflects a
> strong tendency for false memory to occur.
>
>
> On 4/18/12 1:49 PM, "Bruce Weaver" <[hidden email]> wrote:
>
> It sounds like you're looking for the binomial test. It's not clear what
> proportions you expect in the two categories by chance, but if it was 50%,
> the command would look like this (where "recalled" is an indicator variable
> for recall, 0=No, 1=Yes):
>
> NPAR TESTS
> /BINOMIAL (0.50)=recalled
> /MISSING ANALYSIS.
>
>
> HTH.
>
>
>
> Rose, Fred wrote
>>
>> Frequency analysis
>>
>>
>> This is one of those questions that seems so ridiculously simple that
>> I’m embarrassed to ask it, but I’m drawing a blank.
>>
>> I’ve got nominal data (yes/no) from a sample of men and women doing
>> a variant of the Roediger and McDermott False Memory paradigm.  The
>
>> yes/no responses relate to whether or not the individual indicated that
>> they “remembered” a critical word that, in fact, was never
>> presented (the study was more complicated than this, but I’m
>> describing the specific analysis I’m working on now).  The
>> distribution is roughly 55% yes to 45% no.  I am trying to figure out
>
>> the proper test of whether this pattern deviates from chance.
>>  I’m not comparing men and women, but just over all whether
>> this is a high rate of false recognition in the sample as a whole.  
>
>>
>> Any kind soul willing to help a guy out?
>>
>> Thanks,
>>
>> Fred
>> --
>> Fredric E. Rose, Ph.D.
>> Associate Professor of Psychology
>> Palomar College
>> 760-744-1150 x2344
>> frose@
>>
>>
>>
>>
>> ====================To manage your subscription to SPSSX-L, send a message
>> to
>> LISTSERV@.UGA (not to SPSSX-L), with no body text except the
>> command. To leave the list, send the command
>> SIGNOFF SPSSX-L
>> For a list of commands to manage subscriptions, send the command
>> INFO REFCARD
>>
>
>
> -----
> --
> Bruce Weaver
> [hidden email]
> http://sites.google.com/a/lakeheadu.ca/bweaver/
>
> "When all else fails, RTFM."
>
> NOTE: My Hotmail account is not monitored regularly.
> To send me an e-mail, please use the address shown above.
>
> --
> View this message in context:
> http://spssx-discussion.1045642.n5.nabble.com/Frequency-analysis-tp5650112p5650183.html
> Sent from the SPSSX Discussion mailing list archive at Nabble.com.
>
> =====================
> To manage your subscription to SPSSX-L, send a message to
> [hidden email] (not to SPSSX-L), with no body text except the
> command. To leave the list, send the command
> SIGNOFF SPSSX-L
> For a list of commands to manage subscriptions, send the command
> INFO REFCARD
>
>
> --
> Fredric E. Rose, Ph.D.
> Associate Professor of Psychology
> Palomar College
> 760-744-1150 x2344
> [hidden email]
> To manage your subscription to SPSSX-L, send a message to
> [hidden email] (not to SPSSX-L), with no body text except the
> command. To leave the list, send the command SIGNOFF SPSSX-L For a list of
> commands to manage subscriptions, send the command INFO REFCARD

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

Poes, Matthew Joseph

Re: Frequency analysis

Total boondocks. The nature of the situation is unimportant to this problem. Random guessing yes or no is 50:50. If response is biased to "no" because the word wasn't there then that is just fodder for interpretation of a significant result.

A change in chance here has to come from something else, an outside bias. Say the "no" button is a default response that has to be manually changed. That is akin to flipping a weighted coin.

Sent from my iPhone

On Apr 18, 2012, at 5:10 PM, "Len Vir" <[hidden email]> wrote:

> Since 'butterfly' does not appear in the list, does not even a 15 percent
> rate of recall constitute a 'strong tendency for false memory'?
>
> I would have thought that to talk of 'chance proportion' the term butterfly
> would have had to appear in the list.
>
> But I am probably in the boondocks...!
>
> Len Vir
>
>
>
>
> On Wed, Apr 18, 2012 at 5:32 PM, Rose, Fred <[hidden email]> wrote:
>> That is a definite issue – I’m not sure of the correct chance proportions
>> would be. For those not familiar with the task, participants read a list of
>> words such as “moth insect wing bird fly bug cocoon color net” and ask
>> recall or recognition. The false memory occurs when a participant falsely
>> remembers the critical word – in this case, “butterfly”. I don’t think it’s
>> correct to say that 50% is the correct chance proportion but I’m not sure
>> what a correct value would be. It is likely that a 55% false memory rate is
>> higher than false alarm rate on a word list task where the list consists of
>> unrelated words, but again, I’m not sure where to go with it. The bottom
>> line is that I would like to be able to claim that the 55% rate reflects a
>> strong tendency for false memory to occur.
>>
>>
>> On 4/18/12 1:49 PM, "Bruce Weaver" <[hidden email]> wrote:
>>
>> It sounds like you're looking for the binomial test. It's not clear what
>> proportions you expect in the two categories by chance, but if it was 50%,
>> the command would look like this (where "recalled" is an indicator variable
>> for recall, 0=No, 1=Yes):
>>
>> NPAR TESTS
>> /BINOMIAL (0.50)=recalled
>> /MISSING ANALYSIS.
>>
>>
>> HTH.
>>
>>
>>
>> Rose, Fred wrote
>>>
>>> Frequency analysis
>>>
>>>
>>> This is one of those questions that seems so ridiculously simple that
>>> I’m embarrassed to ask it, but I’m drawing a blank.
>>>
>>> I’ve got nominal data (yes/no) from a sample of men and women doing
>>> a variant of the Roediger and McDermott False Memory paradigm.  The
>>
>>> yes/no responses relate to whether or not the individual indicated that
>>> they “remembered” a critical word that, in fact, was never
>>> presented (the study was more complicated than this, but I’m
>>> describing the specific analysis I’m working on now).  The
>>> distribution is roughly 55% yes to 45% no.  I am trying to figure out
>>
>>> the proper test of whether this pattern deviates from chance.
>>>  I’m not comparing men and women, but just over all whether
>>> this is a high rate of false recognition in the sample as a whole.  
>>
>>>
>>> Any kind soul willing to help a guy out?
>>>
>>> Thanks,
>>>
>>> Fred
>>> --
>>> Fredric E. Rose, Ph.D.
>>> Associate Professor of Psychology
>>> Palomar College
>>> 760-744-1150 x2344
>>> frose@
>>>
>>>
>>>
>>>
>>> ====================To manage your subscription to SPSSX-L, send a message
>>> to
>>> LISTSERV@.UGA (not to SPSSX-L), with no body text except the
>>> command. To leave the list, send the command
>>> SIGNOFF SPSSX-L
>>> For a list of commands to manage subscriptions, send the command
>>> INFO REFCARD
>>>
>>
>>
>> -----
>> --
>> Bruce Weaver
>> [hidden email]
>> http://sites.google.com/a/lakeheadu.ca/bweaver/
>>
>> "When all else fails, RTFM."
>>
>> NOTE: My Hotmail account is not monitored regularly.
>> To send me an e-mail, please use the address shown above.
>>
>> --
>> View this message in context:
>> http://spssx-discussion.1045642.n5.nabble.com/Frequency-analysis-tp5650112p5650183.html
>> Sent from the SPSSX Discussion mailing list archive at Nabble.com.
>>
>> =====================
>> To manage your subscription to SPSSX-L, send a message to
>> [hidden email] (not to SPSSX-L), with no body text except the
>> command. To leave the list, send the command
>> SIGNOFF SPSSX-L
>> For a list of commands to manage subscriptions, send the command
>> INFO REFCARD
>>
>>
>> --
>> Fredric E. Rose, Ph.D.
>> Associate Professor of Psychology
>> Palomar College
>> 760-744-1150 x2344
>> [hidden email]
>> To manage your subscription to SPSSX-L, send a message to
>> [hidden email] (not to SPSSX-L), with no body text except the
>> command. To leave the list, send the command SIGNOFF SPSSX-L For a list of
>> commands to manage subscriptions, send the command INFO REFCARD
>
> =====================
> To manage your subscription to SPSSX-L, send a message to
> [hidden email] (not to SPSSX-L), with no body text except the
> command. To leave the list, send the command
> SIGNOFF SPSSX-L
> For a list of commands to manage subscriptions, send the command
> INFO REFCARD

Rich Ulrich

Re: Frequency analysis

In reply to this post by Rose, Fred

I Googled on <Roediger and McDermott False Memory>
and found, immediately, an article on "Factors that
determine false recall..."
http://memory.wustl.edu/Pubs/2001_Roediger.pdf

And the intro mentions rates from 0.01 to 0.65.

If you are going to start into doing research, you really
need to do a large amount of reading to prepare yourself,
both in general (when you know little about research)
and on your specific topic.

--
Rich Ulrich

Date: Wed, 18 Apr 2012 12:57:53 -0700
From: [hidden email]
Subject: Frequency analysis
To: [hidden email]

Frequency analysis This is one of those questions that seems so ridiculously simple that I’m embarrassed to ask it, but I’m drawing a blank.

I’ve got nominal data (yes/no) from a sample of men and women doing a variant of the Roediger and McDermott False Memory paradigm. The yes/no responses relate to whether or not the individual indicated that they “remembered” a critical word that, in fact, was never presented (the study was more complicated than this, but I’m describing the specific analysis I’m working on now). The distribution is roughly 55% yes to 45% no. I am trying to figure out the proper test of whether this pattern deviates from chance. I’m not comparing men and women, but just over all whether this is a high rate of false recognition in the sample as a whole.

Any kind soul willing to help a guy out?
...

Mario Giesel

Re: Frequency analysis

In reply to this post by Rose, Fred

Hi, Fred, The chance of chosing a concept were probably 50%, if no memory was involved, e.g. if respondents were to guess which concepts appeared on a hidden screen. This is not the case in your study. I would probably test against expectation of 0%. I think Error chances then are taken into consideration by the confidence interval of your significance test.
Mario

Am 18.04.2012 um 21:57 schrieb "Rose, Fred" <[hidden email]>:

This is one of those questions that seems so ridiculously simple that I’m embarrassed to ask it, but I’m drawing a blank.

I’ve got nominal data (yes/no) from a sample of men and women doing a variant of the Roediger and McDermott False Memory paradigm. The yes/no responses relate to whether or not the individual indicated that they “remembered” a critical word that, in fact, was never presented (the study was more complicated than this, but I’m describing the specific analysis I’m working on now). The distribution is roughly 55% yes to 45% no. I am trying to figure out the proper test of whether this pattern deviates from chance. I’m not comparing men and women, but just over all whether this is a high rate of false recognition in the sample as a whole.

Any kind soul willing to help a guy out?

Thanks,

Fred
--
Fredric E. Rose, Ph.D.
Associate Professor of Psychology
Palomar College
760-744-1150 x2344
[hidden email]
====================To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD

Mario Giesel
Munich, Germany

Rose, Fred

Re: Frequency analysis

In reply to this post by Rich Ulrich

Re: Frequency analysis Wow. What an incredibly condescending comment. Thank you for enriching everyone’s life for it, especially since you know little to nothing about what I do, who I am, or why I am asking the question. I spared the list the irrelevant details regarding the background of the question and focused more on a desire for some insight on statistical analysis of nonparametric data and boy am I glad you were here to school me.

As to the paper you mentioned...yes, I have it and have read it. And others. They don’t address the question that I asked because those numbers all relate to norms of the DRM lists and I was not asking how to determine whether the rate of false recall in one study differed from the rate in another using the same lists. Perhaps I didn’t express it clearly, or perhaps I should be faulted for not having read every single paper on false memory (shame on me – there probably aren’t that many) but thank you for informing me that the SPSS list is not the place to ask questions of a statistical nature. Imagine my surprise, given that I’ve been a subscriber to this list for 7+ years and have read countless questions of this type, all answered by other subscribers. Apparently, things have changed.

If you don’t mind, Rich, take a look in the upper right corner of your keyboard. You’ll see a key that is probably marked “Delete”. Should I ever choose to post to this list again, daring to ask for information about the application of SPSS to a statistical problem, feel free to use that key so that you might be spared my stupidity.

To the rest of the list – I appreciate your insights and thank you for taking the time to answer a question that at least one of us feels was beneath him. I feel (somewhat naively, apparently) that it is an interesting question on probability but fear there may not be an easy answer.

On 4/18/12 6:10 PM, "Rich Ulrich" <rich-ulrich@...> wrote:

I Googled on <Roediger and McDermott False Memory>
and found, immediately, an article on "Factors that
determine false recall..."
http://memory.wustl.edu/Pubs/2001_Roediger.pdf

And the intro mentions rates from 0.01 to 0.65.

If you are going to start into doing research, you really
need to do a large amount of reading to prepare yourself,
both in general (when you know little about research)
and on your specific topic.

--
Fredric E. Rose, Ph.D.
Associate Professor of Psychology
Palomar College
(760) 744-1150 x2344
frose@...

Garry Gelade

Re: Frequency analysis

In reply to this post by Rose, Fred

Re: Frequency analysis

The “correct proportion” for the null hypothesis depends on how you define the null task. This would be clearer if you had a control condition of some sort. E.g. where the original lists were unrelated to the stimulus words or were presented aurally as opposed to visually etc.

As you don’t have a control group, you are simply measuring a false alarm rate. It is 55% which means that there is a strong tendency for false memory to occur in the task you studied. You could test whether that rate differs significantly from rates obtained by other researchers in other tasks.

Garry Gelade

Business Analytic

From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Rose, Fred
Sent: 18 April 2012 22:32
To: [hidden email]
Subject: Re: Frequency analysis

>
> Frequency analysis
>
>
> This is one of those questions that seems so ridiculously simple that
> I’m embarrassed to ask it, but I’m drawing a blank.
>
> I’ve got nominal data (yes/no) from a sample of men and women doing
> a variant of the Roediger and McDermott False Memory paradigm.  The
> yes/no responses relate to whether or not the individual indicated that
> they “remembered” a critical word that, in fact, was never
> presented (the study was more complicated than this, but I’m
> describing the specific analysis I’m working on now).  The
> distribution is roughly 55% yes to 45% no.  I am trying to figure out
> the proper test of whether this pattern deviates from chance.
>  I’m not comparing men and women, but just over all whether
> this is a high rate of false recognition in the sample as a whole.  
>
> Any kind soul willing to help a guy out?
>
> Thanks,
>
> Fred
> --
> Fredric E. Rose, Ph.D.
> Associate Professor of Psychology
> Palomar College
> 760-744-1150 x2344
> frose@
>
>
>
>
> ====================To manage your subscription to SPSSX-L, send a message
> to
> [hidden email] (not to SPSSX-L), with no body text except the
> command. To leave the list, send the command
> SIGNOFF SPSSX-L
> For a list of commands to manage subscriptions, send the command
> INFO REFCARD
>

-----
--
Bruce Weaver
bweaver@...
http://sites.google.com/a/lakeheadu.ca/bweaver/

"When all else fails, RTFM."

NOTE: My Hotmail account is not monitored regularly.
To send me an e-mail, please use the address shown above.

--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Frequency-analysis-tp5650112p5650183.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

=====================
To manage your subscription to SPSSX-L, send a message to
LISTSERV@... (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

--
Fredric E. Rose, Ph.D.
Associate Professor of Psychology
Palomar College
760-744-1150 x2344
frose@...

Poes, Matthew Joseph

Re: Frequency analysis

In reply to this post by Rose, Fred

Re: Frequency analysis

I’m still unclear what the issue is here? First, why have some suggested anything other than 50% as the false discovery rate? Aren’t random guessing of yes or no (no condition, just guessing, the null hypothesis) no different than flipping a coin, thus 50:50? In terms of the statistic, you can use a non-parametric statistic for this, but if you code it as 1’s and 0’s, then it’s a ratio with normal distribution, and thus a one sample t-test can work. Remember that it’s comparing the mean of your sample against the mean of a normal sample with a mean set to the value you set, in this case, .5 (right?). If this is wrong, I’d like to know why, as I really don’t understand.

I’m pretty certain that the nature of the condition, in the question you asked, is unimportant. You are wanting to test it against chance guessing, which would be equal to no condition at all. I believe you would only want to adjust for this if you had a priori information that chance guessing was in fact biased in some way. It appears to me that no such evidence exists, so you would keep the value at .5. It really seems simpler than people are making it out to be, but maybe I’m wrong on this.

Matthew J Poes

Research Data Specialist

Center for Prevention Research and Development

University of Illinois

510 Devonshire Dr.

Champaign, IL 61820

Phone: 217-265-4576

email: [hidden email]

From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Fredric E. Rose, Ph.D.
Sent: Thursday, April 19, 2012 12:28 AM
To: [hidden email]
Subject: Re: Frequency analysis

Wow. What an incredibly condescending comment. Thank you for enriching everyone’s life for it, especially since you know little to nothing about what I do, who I am, or why I am asking the question. I spared the list the irrelevant details regarding the background of the question and focused more on a desire for some insight on statistical analysis of nonparametric data and boy am I glad you were here to school me.

As to the paper you mentioned...yes, I have it and have read it. And others. They don’t address the question that I asked because those numbers all relate to norms of the DRM lists and I was not asking how to determine whether the rate of false recall in one study differed from the rate in another using the same lists. Perhaps I didn’t express it clearly, or perhaps I should be faulted for not having read every single paper on false memory (shame on me – there probably aren’t that many) but thank you for informing me that the SPSS list is not the place to ask questions of a statistical nature. Imagine my surprise, given that I’ve been a subscriber to this list for 7+ years and have read countless questions of this type, all answered by other subscribers. Apparently, things have changed.

If you don’t mind, Rich, take a look in the upper right corner of your keyboard. You’ll see a key that is probably marked “Delete”. Should I ever choose to post to this list again, daring to ask for information about the application of SPSS to a statistical problem, feel free to use that key so that you might be spared my stupidity.

To the rest of the list – I appreciate your insights and thank you for taking the time to answer a question that at least one of us feels was beneath him. I feel (somewhat naively, apparently) that it is an interesting question on probability but fear there may not be an easy answer.

On 4/18/12 6:10 PM, "Rich Ulrich" <rich-ulrich@...> wrote:

--
Fredric E. Rose, Ph.D.
Associate Professor of Psychology
Palomar College
(760) 744-1150 x2344
frose@...

Rose, Fred

Re: Frequency analysis

In reply to this post by Garry Gelade

Re: Frequency analysis Yeah, I think this is the crux of the problem. This was a student project and it got me thinking about some of the claims the group wanted to make but you are right – they did not have a control task consisting of a list of unrelated words. I was thinking there still might be some, perhaps, obscure, method of testing the 55% rate against some other probability, but the possibilities are pretty endless. Matthew Poes is arguing that 50% is the correct probability since the task was a recognition task and for any given item pure guessing of old/new would have exactly that probability, and under the circumstances (at least as the question is currently framed) I would have to agree.

Again, thanks to all <ahem> for your time.

Fred

On 4/19/12 2:26 AM, "Garry Gelade" <garry@...> wrote:

The “correct proportion” for the null hypothesis depends on how you define the null task. This would be clearer if you had a control condition of some sort. E.g. where the original lists were unrelated to the stimulus words or were presented aurally as opposed to visually etc.

As you don’t have a control group, you are simply measuring a false alarm rate. It is 55% which means that there is a strong tendency for false memory to occur in the task you studied. You could test whether that rate differs significantly from rates obtained by other researchers in other tasks.

Garry Gelade
Business Analytic

From: SPSSX(r) Discussion [[hidden email]] On Behalf Of Rose, Fred
Sent: 18 April 2012 22:32
To: SPSSX-L@...
Subject: Re: Frequency analysis

That is a definite issue – I’m not sure of the correct chance proportions would be. For those not familiar with the task, participants read a list of words such as “moth insect wing bird fly bug cocoon color net” and ask recall or recognition. The false memory occurs when a participant falsely remembers the critical word – in this case, “butterfly”. I don’t think it’s correct to say that 50% is the correct chance proportion but I’m not sure what a correct value would be. It is likely that a 55% false memory rate is higher than false alarm rate on a word list task where the list consists of unrelated words, but again, I’m not sure where to go with it. The bottom line is that I would like to be able to claim that the 55% rate reflects a strong tendency for false memory to occur.

On 4/18/12 1:49 PM, "Bruce Weaver" <bruce.weaver@...> wrote:
It sounds like you're looking for the binomial test. It's not clear what
proportions you expect in the two categories by chance, but if it was 50%,
the command would look like this (where "recalled" is an indicator variable
for recall, 0=No, 1=Yes):

NPAR TESTS
  /BINOMIAL (0.50)=recalled
  /MISSING ANALYSIS.

HTH.

Rose, Fred wrote
>
> Frequency analysis
>
>
> This is one of those questions that seems so ridiculously simple that
> I’m embarrassed to ask it, but I’m drawing a blank.
>
> I’ve got nominal data (yes/no) from a sample of men and women doing
> a variant of the Roediger and McDermott False Memory paradigm.  The
> yes/no responses relate to whether or not the individual indicated that
> they “remembered” a critical word that, in fact, was never
> presented (the study was more complicated than this, but I’m
> describing the specific analysis I’m working on now).  The
> distribution is roughly 55% yes to 45% no.  I am trying to figure out
> the proper test of whether this pattern deviates from chance.
>  I’m not comparing men and women, but just over all whether
> this is a high rate of false recognition in the sample as a whole.  
>
> Any kind soul willing to help a guy out?
>
> Thanks,
>
> Fred
> --
> Fredric E. Rose, Ph.D.
> Associate Professor of Psychology
> Palomar College
> 760-744-1150 x2344
> frose@
>
>
>
>
> ====================To manage your subscription to SPSSX-L, send a message
> to
> LISTSERV@.UGA (not to SPSSX-L), with no body text except the
> command. To leave the list, send the command
> SIGNOFF SPSSX-L
> For a list of commands to manage subscriptions, send the command
> INFO REFCARD
>

-----
--
Bruce Weaver
bweaver@...
http://sites.google.com/a/lakeheadu.ca/bweaver/

"When all else fails, RTFM."

NOTE: My Hotmail account is not monitored regularly.
To send me an e-mail, please use the address shown above.

--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Frequency-analysis-tp5650112p5650183.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

=====================
To manage your subscription to SPSSX-L, send a message to
LISTSERV@... (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

--
Fredric E. Rose, Ph.D.
Associate Professor of Psychology
Palomar College
(760) 744-1150 x2344
frose@...

Jim Marks

Re: Frequency analysis

You might ask some people to rate the concordance of the false memory word to the list.

Give the raters the total list including the false memory word and ask them to rate how much each word belongs in the list. That would give you a probability for the false memory word to be selected without biasing the raters towards the extra word.

The more the word fits the category, the more likely you will get a false memory.

This would give a value for each word that could be used in the binomial test.

Since the students didn't devise a control task, assign this as an extra requirement for a valid project :~). Or ask the whole class to provide the ratings (along with some other more random lists), and teach the distribution of responses as an inter-coder reliability task

Jim Marks
Sr Market Research Manager
National Market Research
Kaiser Foundation Health Plan of the Mid-Atlantic States, Inc.
2101 E. Jefferson St.
Rockville, MD 20852
Phone: (301) 816-6822
Cell Phone: (301) 456-6164

NOTICE TO RECIPIENT: If you are not the intended recipient of this e-mail, you are prohibited from sharing, copying, or otherwise using or disclosing its contents. If you have received this e-mail in error, please notify the sender immediately by reply e-mail and permanently delete this e-mail and any attachments without reading, forwarding or saving them. Thank you.

From: "Fredric E. Rose, Ph.D." <[hidden email]>
To: [hidden email]
Date: 04/19/2012 10:38 AM
Subject: Re: Frequency analysis
Sent by: "SPSSX(r) Discussion" <[hidden email]>

Yeah, I think this is the crux of the problem. This was a student project and it got me thinking about some of the claims the group wanted to make but you are right – they did not have a control task consisting of a list of unrelated words. I was thinking there still might be some, perhaps, obscure, method of testing the 55% rate against some other probability, but the possibilities are pretty endless. Matthew Poes is arguing that 50% is the correct probability since the task was a recognition task and for any given item pure guessing of old/new would have exactly that probability, and under the circumstances (at least as the question is currently framed) I would have to agree.

Again, thanks to all <ahem> for your time.

Fred

On 4/19/12 2:26 AM, "Garry Gelade" <garry@...> wrote:

The “correct proportion” for the null hypothesis depends on how you define the null task. This would be clearer if you had a control condition of some sort. E.g. where the original lists were unrelated to the stimulus words or were presented aurally as opposed to visually etc.

As you don’t have a control group, you are simply measuring a false alarm rate. It is 55% which means that there is a strong tendency for false memory to occur in the task you studied. You could test whether that rate differs significantly from rates obtained by other researchers in other tasks.

Garry Gelade
Business Analytic

From: SPSSX(r) Discussion [[hidden email]] On Behalf Of Rose, Fred
Sent: 18 April 2012 22:32
To: SPSSX-L@...
Subject: Re: Frequency analysis

That is a definite issue – I’m not sure of the correct chance proportions would be. For those not familiar with the task, participants read a list of words such as “moth insect wing bird fly bug cocoon color net” and ask recall or recognition. The false memory occurs when a participant falsely remembers the critical word – in this case, “butterfly”. I don’t think it’s correct to say that 50% is the correct chance proportion but I’m not sure what a correct value would be. It is likely that a 55% false memory rate is higher than false alarm rate on a word list task where the list consists of unrelated words, but again, I’m not sure where to go with it. The bottom line is that I would like to be able to claim that the 55% rate reflects a strong tendency for false memory to occur.

On 4/18/12 1:49 PM, "Bruce Weaver" <bruce.weaver@...> wrote:
It sounds like you're looking for the binomial test. It's not clear what
proportions you expect in the two categories by chance, but if it was 50%,
the command would look like this (where "recalled" is an indicator variable
for recall, 0=No, 1=Yes):

NPAR TESTS
/BINOMIAL (0.50)=recalled
/MISSING ANALYSIS.

HTH.

Rose, Fred wrote
>
> Frequency analysis
>
>
> This is one of those questions that seems so ridiculously simple that
> I’m embarrassed to ask it, but I’m drawing a blank.
>
> I’ve got nominal data (yes/no) from a sample of men and women doing
> a variant of the Roediger and McDermott False Memory paradigm.  The
> yes/no responses relate to whether or not the individual indicated that
> they “remembered” a critical word that, in fact, was never
> presented (the study was more complicated than this, but I’m
> describing the specific analysis I’m working on now).  The
> distribution is roughly 55% yes to 45% no.  I am trying to figure out
> the proper test of whether this pattern deviates from chance.
>  I’m not comparing men and women, but just over all whether
> this is a high rate of false recognition in the sample as a whole.  
>
> Any kind soul willing to help a guy out?
>
> Thanks,
>
> Fred
> --
> Fredric E. Rose, Ph.D.
> Associate Professor of Psychology
> Palomar College
> 760-744-1150 x2344
> frose@
>
>
>
>
> ====================To manage your subscription to SPSSX-L, send a message
> to
> LISTSERV@.UGA (not to SPSSX-L), with no body text except the
> command. To leave the list, send the command
> SIGNOFF SPSSX-L
> For a list of commands to manage subscriptions, send the command
> INFO REFCARD
>

-----
--
Bruce Weaver
bweaver@...
http://sites.google.com/a/lakeheadu.ca/bweaver/

"When all else fails, RTFM."

NOTE: My Hotmail account is not monitored regularly.
To send me an e-mail, please use the address shown above.

--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Frequency-analysis-tp5650112p5650183.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

=====================
To manage your subscription to SPSSX-L, send a message to
LISTSERV@... (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

--
Fredric E. Rose, Ph.D.
Associate Professor of Psychology
Palomar College
(760) 744-1150 x2344
frose@...

Art Kendall

Re: Frequency analysis

In reply to this post by Poes, Matthew Joseph

Of course I would also like to hear from those who propose other criteria in case my guess is off the mark.
These kinds of studies might be focused on the heuristics/models people use to store information.

In one consideration there is a single response -- yes no it was on the list. In the other consideration, one might wonder whether the response is to something more like "is this word similar to the the ones on the list".

Whereas each flip of the coin is independent of previous flips, the question is whether a response is or is not independent of the other responses.

A list of words like red, orange, yellow, green, blue, indigo, violet is presented and removed. After a while words are presented either one at a time or in a check all that apply format. It is possible that "lilac" would evoke a response of "in the list" because it is also a color word. Or "indigo" might be said not to have been presented because it is not a common a word as the others.

Likewise, suppose a list of 7 planets were presented one at a time omitting "Mars". Then the respondent might say that "Mars" was on the list because it is similar to the other words. Maybe even "moon" might yield a response of "on the list" because of its similarity to the words that were on the list.

Art Kendall
Social Research Consultants

On 4/19/2012 9:21 AM, Poes, Matthew Joseph wrote:

Re: Frequency analysis

I’m still unclear what the issue is here? First, why have some suggested anything other than 50% as the false discovery rate? Aren’t random guessing of yes or no (no condition, just guessing, the null hypothesis) no different than flipping a coin, thus 50:50? In terms of the statistic, you can use a non-parametric statistic for this, but if you code it as 1’s and 0’s, then it’s a ratio with normal distribution, and thus a one sample t-test can work. Remember that it’s comparing the mean of your sample against the mean of a normal sample with a mean set to the value you set, in this case, .5 (right?). If this is wrong, I’d like to know why, as I really don’t understand.

I’m pretty certain that the nature of the condition, in the question you asked, is unimportant. You are wanting to test it against chance guessing, which would be equal to no condition at all. I believe you would only want to adjust for this if you had a priori information that chance guessing was in fact biased in some way. It appears to me that no such evidence exists, so you would keep the value at .5. It really seems simpler than people are making it out to be, but maybe I’m wrong on this.

Matthew J Poes

Research Data Specialist

Center for Prevention Research and Development

University of Illinois

510 Devonshire Dr.

Champaign, IL 61820

Phone: 217-265-4576

email: [hidden email]

From: SPSSX(r) Discussion [[hidden email]] On Behalf Of Fredric E. Rose, Ph.D.
Sent: Thursday, April 19, 2012 12:28 AM
To: [hidden email]
Subject: Re: Frequency analysis

Wow. What an incredibly condescending comment. Thank you for enriching everyone’s life for it, especially since you know little to nothing about what I do, who I am, or why I am asking the question. I spared the list the irrelevant details regarding the background of the question and focused more on a desire for some insight on statistical analysis of nonparametric data and boy am I glad you were here to school me.

As to the paper you mentioned...yes, I have it and have read it. And others. They don’t address the question that I asked because those numbers all relate to norms of the DRM lists and I was not asking how to determine whether the rate of false recall in one study differed from the rate in another using the same lists.   Perhaps I didn’t express it clearly, or perhaps I should be faulted for not having read every single paper on false memory (shame on me – there probably aren’t that many) but thank you for informing me that the SPSS list is not the place to ask questions of a statistical nature. Imagine my surprise, given that I’ve been a subscriber to this list for 7+ years and have read countless questions of this type, all answered by other subscribers. Apparently, things have changed.

If you don’t mind, Rich, take a look in the upper right corner of your keyboard. You’ll see a key that is probably marked “Delete”.   Should I ever choose to post to this list again, daring to ask for information about the application of SPSS to a statistical problem, feel free to use that key so that you might be spared my stupidity.

To the rest of the list – I appreciate your insights and thank you for taking the time to answer a question that at least one of us feels was beneath him. I feel (somewhat naively, apparently) that it is an interesting question on probability but fear there may not be an easy answer.

On 4/18/12 6:10 PM, "Rich Ulrich" <rich-ulrich@...> wrote:

I Googled on <Roediger and McDermott False Memory>
and found, immediately, an article on "Factors that
determine false recall..."
  http://memory.wustl.edu/Pubs/2001_Roediger.pdf

And the intro mentions rates from 0.01 to 0.65.

If you are going to start into doing research, you really
need to do a large amount of reading to prepare yourself,
both in general (when you know little about research)
and on your specific topic.

--
Fredric E. Rose, Ph.D.
Associate Professor of Psychology
Palomar College
(760) 744-1150 x2344
frose@...

===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD

Art Kendall
Social Research Consultants