data entery questions for Factor Analysis

[Safa Gurcan]

Hi,
For some reasons I have to help a biological reserch statistical analysis. The project data were(including sea temparature, ph, heavy metals, and etc. ) collected monthly during one year at different sites. There are many variable and data set related chemical and physiological water characteristics. I have read some literatures about this area and see Factor analysis is applied. Probabliy I will use FA analysis on this project.
here is my question, some variables show seasonal variations (example; organisma and water temparature) and I have monthly data set at different sites. I think there will be loss of information taking average of monthly data for each site.

How can I enter correctly spss data entery for using FA for repeated measures?

[SR Millis-3]

I think that we need to know more about your study, eg, aims, hypotheses, what questions you wnat toanswer with the data, etc.

My initial impression is that factor analysis is NOT what you want to do with these data because you have repeated measurements and multiple sites.  FA assumes independent observations.  The nested structure of these data might point you in the direction of a hierarchical linear model (AKA linear mixed model or random coefficients model).  However, we need to know more about the purpose of your study.


~~~~~~~~~~~
Scott R Millis, PhD, ABPP (CN,CL,RP), CStat, CSci
Professor & Director of Research
Dept of Physical Medicine & Rehabilitation
Dept of Emergency Medicine
Wayne State University School of Medicine
261 Mack Blvd
Detroit, MI 48201
Email:  [hidden email]
Email:  [hidden email]
Tel: [phone]
Fax: [phone]


--- On Fri, 11/13/09, Safa Gurcan <[hidden email]> wrote:

> From: Safa Gurcan <[hidden email]>
> Subject: data entery questions for Factor Analysis
> To: [hidden email]
> Date: Friday, November 13, 2009, 6:41 AM
> Hi,
>
> For some reasons I have to help a biological reserch
> statistical
> analysis. The project data were(including sea temparature,
> ph, heavy
> metals, and etc. ) collected monthly during one year at
> different
> sites. There are many variable and data set related
> chemical and
> physiological water characteristics. I have read some
> literatures about
> this area and see Factor analysis is applied. Probabliy I
> will use FA
> analysis on this project.
>
> here is my question, some variables show seasonal
> variations (example;
> organisma and water temparature) and I have monthly data
> set at
> different sites. I think there will be loss of information
> taking
> average of monthly data for each site.
>
>
>
> How can I enter correctly spss data entery for using FA for
> repeated measures?
>

=====================
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

[J P-6]

Safa,

 

I'll try to assist, but without having a more detailed understanding of your data and the purpose of your analysis I am making assumptions that may not be correct. You have a repeated measures data set. The following is an exerpt from paper I started on using SPSS Mixed and never finished...at least not yet! Hope the tables come through.

 

************ Begin exerpt ********

 Structure of the data file for a repeated measures analysis is very important. There are two basic data structures to consider when constructing a data file for multilevel modeling: the multi-variable (MV) file structure and the mult-record (MR) file structure.

 

Conventional analysis usually requires a MV structure. The defining characteristic of multi-variable (MV) structure is that all information pertaining to a single observation is placed on one line in the dataset. For example, if there are 20 participants in a study with 3 variables recorded for each person, the resulting MV data file will contain 20 lines and 3 columns as in Table 1. When multilevel modeling is employed it is necessary to include at least one ‘nesting’ variable that identifies cluster membership. It is also common to include an individual level id number as well.

 

Table 1. Multiple Variable (MV) Data Structure

Person ID	BMI_T1	BMI_T2	BMI_T3
1	12	45	34
2	23	43	34
…	…	…	…
19	26	40	38
20	27	49	44

 

 

The defining characteristic of mult-record (MR) structure is that information pertaining to a single observation is ‘stacked’ or ordered sequentially on multiple lines in the dataset. The MR structure is required for analysis of repeated measures data using SPSS MIXED.

 

With the release of SPSS v11 the VARSTOCASES command became available and is a very useful tool for converting a MV file to a MR file. The basic syntax for converting the above MV file to MR structure is:

 

VARSTOCASES

/MAKE BMI FROM bmi_t1 TO bmi_t2

/INDEX=time1.

 

Additional information on this procedure is available in the SPSS Command Syntax Reference and online at the UCLA statistical resource web site: www.ats.ucla.edu/stat/

 

When using a MR file format for repeated measures analysis two new variables are required. First, each person is required to have a unique identifier (PersonID). With the VARSTOCASES command a variable named ‘caseid’ is created automatically and can serve this purpose. The second new variable is one that represents the timing or sequencing of measurements (TIME1). This is necessary to preserve the temporal order of each measurement. Again, using VARSTOCASES an index variable can be created that reflects the order of measurements. But be careful! SPSS simply uses the order in which variables are listed in the syntax command to create this order. Finally, a third new variable is included, TIME0. TIME0 is simply TIME1 – 1. Notice the only difference between TIME0 and TIME1 is the former begins time at 0 and the later begins time at 1. Which variable is used in the analysis? It depends. One or the other may not be used, but it is easy to compute both at this point. Note that in this context the variable TIME is a level-1 predictor and PersonID is a level-2 predictor.

 

For example if there are 20 participants in a study with 3 observations for each person on BMI, the resulting MR data file will contain 120 records and 3 variables as illustrated in

Table 2.

 

Table 2. Multiple Record (MR) Data Structure

PersonID	BMI	Time0	Time1
1	12	0	1
1	45	1	2
1	34	2	3
2	23	0	1
2	43	1	2
2	34	2	3
…	…	…	…
19	26	0	1
19	40	1	2
19	38	2	3
20	27	0	1
20	49	1	2
20	44	2	3

 

 ************** end of exerpt ***********

 

As for the factor analysis, without knowing more I suggest starting with the MV structure.

 

HTH,

John

---

From: Safa Gurcan <[hidden email]>
To: [hidden email]
Sent: Fri, November 13, 2009 6:41:06 AM
Subject: data entery questions for Factor Analysis

Hi,
For some reasons I have to help a biological reserch statistical analysis. The project data were(including sea temparature, ph, heavy metals, and etc. ) collected monthly during one year at different sites. There are many variable and data set related chemical and physiological water characteristics. I have read some literatures about this area and see Factor analysis is applied. Probabliy I will use FA analysis on this project.
here is my question, some variables show seasonal variations (example; organisma and water temparature) and I have monthly data set at different sites. I think there will be loss of information taking average of monthly data for each site.

How can I enter correctly spss data entery for using FA for repeated measures?