I am conducting a research project where:
- I am surveying a random sample of military leaders (N=100).  Each
participant will complete two surveys.  The first survey measures
creativity and has three sub-scales that equate to a total score.  The
second survey measures personality type and has four sub-scales (no
total score).

My questions are:
1.  Would leadership be considered my dependent variable, if so why
and if not why not?
2.  Can and should leadership be measured, if so, how?
3.  Are the independent variables the sub-scales from the surveys? 
Would I be analyzing eight coefficients (four creativity and four
personality scales)?
4.  Would a Pearsonian correlation (r) be the best statistical
treatment to show correlation between the independent variables?
5.  Any other additional advice on data analysis for this project will
be accepted and appreciated.

Thanks for considering these questions.  Please let me know if you
need further clarification.

---

Request for Question Clarification by legolas-ga on 07 Dec 2002 09:02 PST

Can you please tell me what your H0 and H1 hypothesis are? Or, do you
want me to suggest possibilities?

Legolas-ga

---

Clarification of Question by maccreate-ga on 07 Dec 2002 10:15 PST

My hypotheses are:

Ho A significant relationship exists between innovative creative style
and intuitive and perceptive personality types.

H1 - There is a positive relationship between innovative creative
style and the extravert personality type.

---

Request for Question Clarification by legolas-ga on 07 Dec 2002 14:30 PST

Hi maccreate-ga,

Thanks for an interesting question--this takes me right back to my
Research Methods courses when I was still at University.

Before I tackle each of your questions, I'd like to first try and
define some terms to make this discussion a little easier.

Dependent Variable is the variable measured and recorded by the
experimenter.

Independent Variable is the variable manipulated by the experimenter.

In any well designed and reasonably small study, there should be only
ONE dependent variable that you are looking at. You may have more than
one independent variable however--provided that your statistics take
that into account.

Also, an important note about correlational data, whenever you have a
research design that asks if A varies with B (negatively or
positively), there is a danger of interpreting cause and effect. We
normally do not know in which direction the correlation exists: only
that there is correlation present.

Let me explain: if a fictional study looking at grades and self-esteem
came up with a strongly positive correlation between high grades and
high self-esteem, it would be impossible for us to determine if the
high grades caused the high self-esteem, or the high self-esteem
caused the high grades.

Another important point is the null hypothesis (H0) predicts that the
independent variable will have no effect on the dependent. The
scientific or alternative hypothesis (H1) is the inverse of H0: the
independent variable will have an effect on the dependent variable. In
your H0 and H1 hypothesis, you made an error in defining you
hypothesis. Both of your hypotheses predict a correlation exists. This
must first be fixed before you attempt anything further.

At this point, I'd like to help you--but, I don't believe that your
questions are appropriate to the study you wish to conduct. If you'd
like me to help you define what you are looking at and for better, I
am more than happy to help. But, at this point, there is nothing more
I can do.

Please let me know what course of action you'd like to take.

Thanks,

Legolas-ga

---

Clarification of Question by maccreate-ga on 08 Dec 2002 03:17 PST

Thank you very much for you prompt and thoughtful reply.  You are
correct that I mixed my null and directional hypotheses.  To clarify
my Ho is:  there is not a relationship between cognitive style of
creativity and personality types in military leaders.  My H1 would of
course be what I expect to find, that is there is a positive
correlation between creativity and personality type in military
leaders.

I'd like to add that this study is a replication of prior research.  I
am building on the previous studies by checking on the validity of the
research findings by using a different population - military leaders.

I hope that this clarifies what I'm trying to accomplish, and again
thanks for your inputs.

---

Request for Question Clarification by legolas-ga on 08 Dec 2002 08:39 PST

Before I can answer, I need to clarify one more point: leadership as a
measured variable. Nothing in your hypothesis mentions "leadership"
other than a cursory mention of "military leaders". In my mind you are
either still not getting your hypotheses to a desired level of
correctness or, you are trying to intruduce another variable to the
research that isn't a part of the hypothesis to begin with.

Perhaps if you can provide a link or a citation to a study that you
are replicating, that would help me better tune my answer. Also, can
you tell me if this is for a Psychology Stats class? The reason I ask
is that I can gear my answer slightly better for the level of
understanding you currently have (if I know it's an intro stats
course, I'll stay away from the really annoying minutia of stats).

Let me know,

Legolas-ga

---

Clarification of Question by maccreate-ga on 08 Dec 2002 12:33 PST

The study that it is being replicated is title "Styles of creativity:
test-score correlations between Kirton adaption-innovation inventory
and Myers-Briggs type indicator".  The authors are G.C. Carne & M.J.
Kirton, Psychological Reports, 1982, 50, 31-36. Sorry, but I don't
have a link for this report.

This study for a master thesis in human relations.  I am somewhat
rusty in stats and research methods and your help is appreciated.

You posed an interesting question concerning my attempt to
operationalize leadership.  Perhaps you are right and leadership
doesn't need to be part of the hypothesis.  Thanks again and I will
try to provide more information next time around.

---

Request for Question Clarification by legolas-ga on 09 Dec 2002 09:42 PST

I am trying to find a question that I can actually give an answer to
in all of this. You asked, in the original question 5 specific
questions. Questions 1 & 2 are moot in my opinion. There is no answer
to them. Question 3 asks about IV's being the subscales (which again,
is a question that might be very difficult to answer as your
hypotheses don't really talk about each scale--but, an overall view of
the data).

I'd be happy to give you an answer that reitterates what I've already
said, and expands on it to give you an idea of IV's and DV's based on
what you are looking at --and your H0 and H1. Also, I'm sure I can
find you some good stats that will help with the analysis--and some
advice about general best practicies in statistics.

Would this be acceptable to you as an answer?

Legolas-ga

---

Request for Question Clarification by kutsavi-ga on 10 Dec 2002 14:47 PST

Yet another researcher chiming in with a request for clarification!! 
Hoohooo!

'Scuse me.  

Anyway, here's my wonderment, and I can't seem to find that it's
already been asked or addressed:

Where does leadership come into play? In your revised hypothesis,
you're examining a relationship "between cognitive style of creativity
and personality types in military leaders". There is a construct
called "leadership", and it can be studied using various
well-validated measures, but your current hypothesis does not seem to
address leadership.

In fact, i'm not sure what your research hypothesis is intended to
examine. You state "...I expect to find...there is a positive
correlation between creativity and personality type in military
leaders." You need to be much more specific. what kind of creativity?
is the instrument you're using to measure creativity reliable and
valid?

However, your first research hypothesis ("H1 - There is a positive
relationship between innovative creative style and the extravert
personality type.") is actually more workable, as long as the H0
states that no relationship exists between these variables. In order
to use this one, you'd measure "innovative creative style" and
extraversion and use a pearson's r to determine the correlation
between the two.

One other note - in the original study that you're replicating, did
they use the myers-briggs to measure extraversion? If so, then ok, use
it. However, far more reliable and valid personality inventories exist
with which you could measure extraversion. If you have a choice, I'd
use another one.

---

Request for Question Clarification by legolas-ga on 10 Dec 2002 15:42 PST

bcguide makes an excellent point in the comments:

"...when you replicate a study you use exactly the same methods and
analysis that the original study employed...". bcguide is quite
correct that a replication should "replicate" the study. Literally,
you should be "doing it again with a different sample".

As a friendly suggestion, I'd strongly recommend heading to the
bookstore at your University and seeing if they have an intro-research
methodology book for you to read. I know it will feel silly being a
grad student looking at intro books, but, a nice easy refresher might
be just the boost you need!

BTW, I know when my wife did her MSc, she found the stats and research
methodology the most troubling aspect of her thesis. Let me know what,
if anything, I can answer for you, or, of course if you are happy with
the help that I've offered you already, let me know that I can post it
as an "answer".

Legolas-ga

---

Clarification of Question by maccreate-ga on 11 Dec 2002 11:57 PST

Mega thanks to:

legolas - for starting the ball rolling and some terrific insight
kutsavi - for clarifying some aspects of the leadership query
bcguide - for the replication elucidaion
pelican - for clearly being a 'stats guru'

having said that, I'd like to ask Rebekah for help in further
answering my question.  she seems to have intuitive insight on the
direction I'm taking this study.

Again, thanks to all for making this a wondrous learning experience. 
Each of you truly are wonderful and thanks again for your time and
effort.

Maccreate out...

Rebekah, i'm working on a further clarification and will post it on
Thursday.

---

Clarification of Question by maccreate-ga on 14 Dec 2002 08:35 PST

I thought that I posted this on Dec. 11, however it doesn't seem to be
on the board.  here is my latest thoughts on this project.  many
thanks go out to:

legolas - for patience and some great questions
kutsavi - for clarifing aspects on the leadership aspect
bcguide - for the insight on replication
pelican - for clearly being a 'stats guru'

having said this, i would especially wish to thank Rebekah.  she  has
hit most of the major points of what i'm trying to accomplish.  
Rebekah - would you be interested in helping me further along?  I have
a couple of questions to pose based on your points and I think you
could realy be a big help.  thanks to all.

Maccreate...

---

Clarification of Question by maccreate-ga on 17 Dec 2002 12:15 PST

OK Rebekah, here goes!
On the post you made 10 Dec 2002 22:48 PST you discuss the hypotheses.
 You state "if your research question/hypothesis says, "there is a
positive relationship", then maybe you are trying to do a pearson r...
just get data of 100 men (not random) and run a correlation.  This is
in fact what I'm doing, since the study I'm replicating (surveying
military leaders) uses a pearson r for the correlation between
creativity and personality.

Here are my questions:
1.  Why not a random sample?  How can I assure that these results wil
not generalize beyond what is my sample vs. random selection?

2.  You mentioned subject variable - unfortunately I can't add another
survey in this study.  Having said that, can leadership still be
considered a subject variable without measuring it?  Also, can one
conduct a  One Way - ANOVA test without using the SV.

That's all for now, hoping that you'll see this post and be able to
make sense out of my questions.

Thanks a bunch!

---

Clarification of Question by maccreate-ga on 19 Dec 2002 11:50 PST

Rebekah:

I posted a couple of questions two days ago - have you seen them?  I
checked and they are on listed after my 10 and 11 Dec posts, about 2/3
the way down the page.  Hope you have a chance to answer them.

Thanks:

Maccreate

---

Request for Question Clarification by legolas-ga on 19 Dec 2002 13:04 PST

You should know that starrebekah-ga is not a Google Answers Researcher
to the best of my knowledge. If she does answer your questions, it
will be out of the goodness of her heart--not because she's being
paid. Commenters are not screened by Google Answers prior to being
allowed to post comments on questions: and their answers may or may
not be appropriate or correct.

Legolas-ga

---

Clarification of Question by maccreate-ga on 28 Dec 2002 23:36 PST

Rebekah:

Hello, and how are you?  My main concern is not so muchin having my
question answered, but in hoping that your mother makes a speedy and
full recovery.  IMHO, family ALWAYS comes first and to have an
accident occur around the holidays, well ...

So, thanks for your attention to my question.  As I have said in the
past, you are right on target.  I also respect your self-confidence
and I hope the powers that be in Google soon certify you.

My previous post contains the brunt of my question.  I'm going to run
a Pearson's on my creativity and personality variables, and also test
the means and stand. dev.  The creativity instrumetn measures three
attributes and the personalitly instrument measures eight attributes.

I expect to find a positive correlation between innovative style of
creativity and intuitive and perceptive personality types.  I have
access to both SAS and SPSS, which do you recommend?

Basically, this study was originally set out to determine the nature
and extent of the association between creativity and personality type.
My twist is that I'm using a population (sample?) of military leaders
instead of private sector leaders.

Thanks for your hlep.  Take care and hope that Mom gets well real
soon.

Maccreate out...

Hi,
If I understand you correctly, you are replicating a study. Normally,
when you replicate a study you use exactly the same methods and
analysis that the original study employed to verify that they got the
results that are claimed and that the results are able to be
replicated.

I don't understand why you don't just use the same analysis that the
original study reported.

If you do anything original, you can't make a clean comparison to the
original study.

Regards,
bcguide-ga

If the method used in the 1982 study was flawed, what would be the
sense of repeating the same mistake?  You all know the old saying,
"there are 3 kinds of lies: small lies, big lies, and statistics."

Valid statistical inference requires collecting data by random
sampling from a homogeneous (stable distribution) population under
controlled conditions, so that the Central Limit Theorem applies. 
Then, if the "independent" variables are really independent of each
other -- and only then -- you can infer how the dependent variable
will change in response to changes in the independent variables.

I am reluctant to recommend statistical methods to crunch numbers, and
treat the results as statistical inferences, unless it is clear that
the fundamental assumptions of mathematical statistics are valid.  In
the case at hand, it is not obvious to me that style and personality
are statistically independent or even that they are random variables
with stable distributions.  Therefore, how they can be used to infer
leadership ability, is beyond me.

In fact, it seems to me that that style, personality, and leadership
constitute a closed-loop feedback system, i.e., personality influences
style, both in turn influence leadership ability and, as leadership
ability grows, both style and personality traits are affected by the
experience of being a leader.  This would not be a statistical
problem, and would require a completely different method of analysis.

:-) Sorry for the soap opera,

pelican-ga

Oops, a correction to my own soap opera.  In the second sentence of
the second paragraph, after ".... then -- you can infer", please
insert ",with some level of statistical significance," ...

:-) Sorry, 

pelican-ga

Hi Maccreate!
    I'm actually in several advanced research design and statistics
classes now to further my career as a psychology student - and I think
I have to disagree with the researchers/commenters.

    I admit I am not familiar with the study that you are trying to
"replicate" - but from reading the description of your study, this is
what it seems to me:

    Well, let me explain a few things first.  Let's pick a very simple
study.  Let's say that 100 college students (random sampled) had to
complete a survey on feelings of romance.  They then watched a film
with passionate scenes, and filled out another (post) survey about
romance.   The DV of course, would be the scores measured from their
post-test (as compared to the pre)  and the IV would be the movie. 
Hopefully after watching the romantic movie - the feelings of romance
on the test went up.

     Your study is not like this - because there is no IV, nothing is
manipulated.  (If I understand the study correctly).  So what's the
point of having these men take these surveys?  Because you want to see
what kind of scores they get?  No, the answer is you want to see what
kind of scores THOSE PARTICULAR army men get.  THat is where your
"leadership" variable comes in.

      One of your questions was if your leadership variable was a DV,
no it is not.  However - it is a subject variable, or SV.   Subject
variables are used to describe, strangely enough, the subject.   Let's
say I wanted to see if there was a difference between men & women in
relation to how much ice cream they ate on a hot day, versus a cold
day.   DV would be # of ice cream cones consumed - IV would be
temperature (hot/cold) and - alas, SV would be male or female (because
you've got to put it in there somewhere!  You cannot manipulate
whether the person is male/female (so not IV) and you cannot measure
maleness or femaleness (well maybe, but that's a completely different
issue :))


You asked if you should use a Pearson r.  Absolutely not.  A pearson r
does not analyze cause/effect - and there's no point in randomly
sampling these men - if you're just going to do a simple correlation. 
 What the pearson r does is this:  here's another example.   Pick 100
college students, look at their GPA, and their amount of time watching
tv.   Run the pearson r - it will tell you if as GPA goes up, their
time watching tv goes down ---  or vice versa -- or no relation at
all.   There's no IV, there is nothing manipulated - and it doesn't
tell you cause and effect (you don't know whether the high GPA CAUSED
the low shoe size, or vice versa -  they could have a high GPA b/c
they watch academic TV, or have a low TV # b/c the parents grounded
them because of low GPA - you don't know!)

What you want to run here - especially if you're putting in that
leadership SV is an ANOVA, or Analysis of Variance test.   IT would be
a One Way - ANOVA,  with SV X SV X DV X DV.   The problem here though
is that - to do this correctly, you need 100 men who are non-leaders. 
  So, the test would look at  non-leaders & test on first score, and
non-leaders & test on second score  versus leaders & test on first
score, and leaders & test on second score.


I'd be more than happy to answer any of your questions or to clear up
any misconceptions you (or anyone else) may have!!

Good Luck with the study!!

-Rebekah

Sorry for not posting this before - but I just read your Hypotheses:

*****************
My hypotheses are: 
 
Ho A significant relationship exists between innovative creative style
and intuitive and perceptive personality types.
 
H1 - There is a positive relationship between innovative creative
style and the extravert personality type.
******************

Huge problem with this.  Did you get them from the original study - or
are these your own?  I have a few questions about using these:

1) Where is the leadership supposed to go in on this?  If you're just
trying to look at different personality types and creative styles then
what is the point of "drafting" the army leaders?   You may want to
consider rewriting your hypothesis to say: "There is a positive
relationship between innovative creative style and the extravert
personality type among military leaders" (or something to that effect.
2) If your research question/hypothesis says "There is a positive
RELATIONSHIP" - then maybe you are trying to do a pearson r, or a
correlation - but again, there is no need to random sample if you're
doing this.  Just get data of 100 men (not random) and run a
correlation.


Another point - I now understand that you are not EXACTLY replicating
previous studies, but you have done a literature review (looked at
previous studies) and are adding on to them by changing the population
-- now military leaders.  If this is the case then you DEFINITELY want
to use that subject variable - and make sure that the military leader
part is included in that study.

Also, legolas had a good point - that maybe it is not exactly geared
to LEADERSHIP.  But, I have an answer for you!  Use your military men,
and the ANOVA I described above - however.. add in ANOTHER survey - a
survey on Leadership (I'm sure I can find a scale that has been used
in the past and is well known) and do a TWO way anova.  That way, you
can look also at military leaders who have high versus those who have
low leadership scores on this well known (and accepted) test.


Best of luck -- and PLEASE don't hesitate to ask for my help!  I'm
enjoying helping on this project!

-Rebekah

I found the abstract of the original article on PSYC INFO - if anyone
is interested, here it is:  (Sorry it's not the full article!)

"AN: 1982-27059-001
DT: Journal-Article
MT: Print-Paper
AU: Carne,-G-C; Kirton,-M-J
AF: Carne,-G-C: Chandler and Macleod Pty, Melbourne, Australia
TI: Styles of creativity: Test-score correlations between Kirton
Adaption-Innovation Inventory and Myers-Briggs Type Indicator.
PY: 1982
SO: Psychological-Reports. 1982 Feb; Vol 50(1): 31-36.
PB: US: Psychological Reports. 
IS: 0033-2941
LA: English
AB: Findings from the administration of the Kirton Adaption-Innovation
Inventory and the Myers-Briggs Type Indicator to 109 21-55 yr old
management students indicate statistically significant moderate
correlations between the Kirton and 2 of the 4 Myers-Briggs
dimensions. To the extent that the Myers-Briggs Type Indicator
measures creativity, it is style rather than level that it measures.
(9 ref) (PsycINFO Database Record (c) 2002 APA, all rights reserved)
KC: Kirton Adaption-Innovation Inventory vs Myers-Briggs Type
Indicator, measurement of styles of creativity, 21-55 yr olds
MJ: *Creativity-; *Creativity-Measurement; *Test-Scores
CC: 2223-Personality-Scales-and-Inventories;
3120-Personality-Traits-and-Processes; 2223; 3120; 22; 31
PO: Human
SF: References; Peer-Reviewed
UD: 19821001
"

Maccrete,

Thank you for your kind words!  I would absolutely be interested in
helping you with further aspects of the study.  I was looking forward
to your re-post on Thursday, but am afraid that I didn't find it.  Are
you going to post something new soon?  I look forward to hearing your
new clarification! :)

-Rebekah

Hello Maccrete!

    First of all, I'd like to apologize for taking so long to get back
to you.  I normally check Google Answers every night, but due to a
tragic freak accident - I've spent the last few weeks with my mother
in the ICU.  She was finally released from the hospital yesterday, and
even though I'm still caring for her - I finally have time/ access to
the computer.  Sorry for any inconvenience that caused you, and I
still am very eager to help you with your study.
    Secondly, Legolas is correct, I am NOT a Google Researcher.  I
have put applications in time & time again, and they will not accept
me - NOT because I am not overly qualified, but because there are too
many applications coming in at this time.  I have already e-mailed the
answers people and they say it's perfectly fine if I answer your
question out of the goodness of my heart, which -- good news, means
you'll get your answer for FREE! :)
    I hope you'll find that my answers have been appropriate and
correct, and I'll give you some of my background so that you can
decide for yourself:  I am currently an undergraduate Senior
psychology major, studying specifially Forensic Psychology - but my
degree will be a BA in Psych.  I will graduate this May.  I have taken
over 35 hours in Psychology, including: Introduction to Psych, Social
Psych, Abnormal Psych, Developmental Psych, Introduction to Research
Methods, Introduction to Statistics, Research Methods II,
Environmental Psych, Forensic Psych, History and Systems of Psych, and
most recently, Advanced Experimental Statistic and Design (which is an
honor course with 8 students - and that's why I'm specifically
interested in helping you - it was helping me remember and utilize my
matieral! :)  I have also been an assistant researcher in two
full-fledged studies, and am participating in two mmore (which are
both soon to be published), one at Loyola Univ, and another at the
University of New Orleans,  and have designed and run three studies of
my own, with another on the way (soon to be published).  I hope this
is enough for Legolas to stand assured that I AM qualified!  And if
Legolas (or anyone else) feels that they could do a better job, I urge
them to go ahead.

But for now, I will review what you've posted -- and give you the best
answer that I possibly can!   I hope you enjoy your holidays, and I'll
get back to you as soon as possible! :)

-Rebekah