Statistician good with factor analysis needed for publication

#1
Hi all,

I'm reposting a message from the Psychology section as I think this location makes more sense.

I am working on a new manuscript with a Columbia University professor in the realm of stigma and psychosis research. We conducted a broad, yet intensive quantitative study with a large clinical interviewing component. We are seeking to create a scale, or rather create one for existing instruments used in the study, to address a unique, emergent psychosocial issue, and publish results of the analysis in a peer-reviewed journal.

I am looking for a statistician to be second author on the paper. S/he would guide the entire analysis and perform all tests and write up the methods. The analysis will be very heavy on factor analysis, so very strong, fluid skills in that area are ideal. Need someone who can work consistently within a 2-3 week time frame. Data is pretty clean and sample is relatively small, so this shouldn't be an issue. We will provide the general conceptual framework. Uncompensated, but manuscript has strong publication acceptance probability, and there will be opportunities for future publications on this particularly fascinating cohort.

If you are associated with an academic organization, that would probably be best for avoiding any conflicts of interest.

Please send me a PM with a few details on your background and availability in coming weeks.

Thanks!
 
#2
Hi again - just re-upping this and adding on the detail that this should require 2 to 3 hours a week max and that you don't have to have a specialty in psychiatric or psychology research. Please let me know if you have any specific questions. Really looking forward to getting someone to collaborate with! If you have a friend or associate who may be interested, that works too!
 

spunky

Doesn't actually exist
#3
sample is relatively small, so this shouldn't be an issue.
this is an issue in factor analysis (at least most of the time it is). how many people are we talking about here?

and on to the original point of my post.... speaking from the vantage point of being a psychometrician... you do realize that when developing a scale probably the least of your concerns is the statistical aspect of it, correct (i mean with computers nowadays crunching the numbers and making sense of them takes pro'lly an afternoon)? i'd be a little bit more concerned about the whole validity issue, especially if this is going to be used for clinical purposes
 
#4
this is an issue in factor analysis (at least most of the time it is). how many people are we talking about here?

and on to the original point of my post.... speaking from the vantage point of being a psychometrician... you do realize that when developing a scale probably the least of your concerns is the statistical aspect of it, correct (i mean with computers nowadays crunching the numbers and making sense of them takes pro'lly an afternoon)? i'd be a little bit more concerned about the whole validity issue, especially if this is going to be used for clinical purposes
Thanks for the reply. The sample is n=65. Definitely not a large sample. Validity is certainly a concern, but there are many angles to explore within the dataset, although as you point out even the sample size cannot be overlooked. The design of the study was not epidemiologic or cross sectional in nature. We conducted a robust suite of clinical and psychometric evaluations with a selected cohort of individuals with the goal of getting at more interpersonal aspects of the mental condition.

The hope is to find someone who can help psychometrically explore some of these contours and determine how strongly some of the items are loading. We would like introduce a new measure for a very specific population within the SMI population. I think replication in future, larger studies would absolutely be warranted, particularly if introducing a clinical measure. At this stage, this is exploratory and not translational. Still haven't found a partner! This analysis is probably best suited for someone who is interested in provisional analyses and wants a publication. Please PM if interested. Thanks again for the reply.
 

spunky

Doesn't actually exist
#5
The hope is to find someone who can help psychometrically explore some of these contours and determine how strongly some of the items are loading.
so... besides doing a standard factor analysis and some rotations here and there. what else are you looking for? do you want a full item analysis of the scale in question or just the factor analysis is fine?

a very specific population within the SMI population.
what's SMI? please keep in mind that most of us don't work on the health/medical field. when you use terms specific to your area it makes it difficult for us, the contributors, to help you.

Still haven't found a partner!
something that kinda surprises me given the fact that knowing factor analysis inside out is sort of expected from people who work on developing scales.

i guess i'm wondering just how complicated is the analysis you're trying to obtain. i'm not saying i have time nor that i'm interested in publishing anything, but i do think that if it is straightforward enough we could maybe all chip in here through the board as a collaborative effort. there are at least two other people i can think of here who understand latent variable modelling and who can probably help you out as well if i'm not able to (and those 2 people know who they are so stop pretending and come forward :p)
 
#6
so... besides doing a standard factor analysis and some rotations here and there. what else are you looking for? do you want a full item analysis of the scale in question or just the factor analysis is fine?



what's SMI? please keep in mind that most of us don't work on the health/medical field. when you use terms specific to your area it makes it difficult for us, the contributors, to help you.



something that kinda surprises me given the fact that knowing factor analysis inside out is sort of expected from people who work on developing scales.

i guess i'm wondering just how complicated is the analysis you're trying to obtain. i'm not saying i have time nor that i'm interested in publishing anything, but i do think that if it is straightforward enough we could maybe all chip in here through the board as a collaborative effort. there are at least two other people i can think of here who understand latent variable modelling and who can probably help you out as well if i'm not able to (and those 2 people know who they are so stop pretending and come forward :p)
"SMI" stands for "serious mental illness." I guess I just kinda tossed that out there. I don't owe any statistical software myself, but was able to some some unadjusted analyses and I think there is enough "meat" there - that said, it would probably be more accurate to say that we're looking for statistical consultation. Through my provisional work, I don't have a sense of how meaningful the connections are (or even conceptually logical). There could be a more logical route to take (or more interesting at least) that does not involve factor analysis. The scale is [supposed to be] 8 items. I wouldn't want to set up a priori expectations for the analyses needed since the scales were conceptualized after the point. I realize some of you are in this for fun/gaining more knowledge. This opportunity might be of most interest to those who want to publish and are looking for a long-term partnership. This study has a lot of data, however as it is currently unfunded we don't have resources to fund a statistician. So it could be fun to explore over several years with publications wherever meaningful/viable.
 

CB

Super Moderator
#8
I am looking for a statistician to be second author on the paper. S/he would guide the entire analysis and perform all tests and write up the methods. The analysis will be very heavy on factor analysis, so very strong, fluid skills in that area are ideal.
I'm not sure that a statistician is really what you're after. Factor analysis is something that originated in psychometrics and tends mainly to be of interest to social scientists, especially psychologists. The folks on TS who are interested in factor analysis tend to be psychologists or psychometricians, not statisticians per se.