Graduate student needing latin square help please

[lamphearss]

I am a graduate student who is STRUGGLING to figure out how to run my collected data using a latin square. I did a mouse study that tested six mice, influenced by six scents, for six nights, over a six month period (six new mice were studied for each month). I simply do not know how to structure my data so that it will run. The response variable is th GUD's or difference in food intake. I could probably include time as a response variable but at this point I'm not willing to consider that as an additional factor. Can any one help???

data RCB;
 
input night scent $ food @@;
 
datalines;
 
1 v  1.60 1 W  2.16 1 C  1.70 1 RA 1.08 1 B  1.17 1 RF 0.69
 
2 B  0.94 2 RF 1.58 2 V  1.33 2 W  0.51 2 C  0.07 2 RA 1.41

3 C  0.08 3 RA 1.53 3 B  1.78 3 RF 0.05 3 V  1.38 3 W  0.44
 
4 W  3.46 4 C  0.08 4 RA 1.85 4 B  1.43 4 RF 0.57 4 V  0.49
 
5 RF 1.04 5 V  1.69 5 W  2.67 5 C  2.03 5 RA 0.33 5 B  1.02
 
6 B  0.71 6 B  0.68 6 RF 2.56 6 V  0.76 6 W  0.61 6 C  0.57
 
;
 
proc anova data=RCB;
 
class night scent;
 
model food = night scent;
 
run;


Leaving out the month of the year, I was able to run my data but I must figure out how to incorporate the months in too.  The letters correspond with the scents, the single digit numbers correspond with the mouse/night and the three digit number is the GUD. Thank you

[Rich Ulrich]

1) You have provided this SPSS list with SAS syntax.  Your
SPSS equivalent is something like List Free, but I would not
use it.  I would read in by line  (scent1 GUD1 scent2 GUD2 ... )
and then write out using VarsToCases -- in order to give each
Scent its proper "day", 1-6, in the new file.

2) This is not a Latin Square as I understand it.  For whatever that
is worth.

Your orders are (1-6), (2-6,1), (3-6, 1,2), (4-6, 1-3), (5,6, 1-4),
and (6, 1-5).  As I recall, the original Latin Square was designed
for agricultural experiments, where the plots to be planted and
treated were taken in strictly designed order with offsets so that
(for instance) a diagonal ridge through the farm could not create
an artifact favoring one treatment.   - That is not a problem for
mice.  However, the present design builds in exactly one sequence
of "order" and cannot speak or carry-over effects except by looking
carefully (say) at "first trial".  Perhaps this matters; perhaps not.

3) Case 6 is mislabeled, "6 B 0.71 6 B 0.68"  should be
"6 RA 0.71 6 B 0.68" so that B follows RA as in every other line.

4) I don't see any problem with analyzing various effects in a
file with one Scent per line, putting in whichever effects you
want to test.  It is my impression that Latin Squares use
conventional partitioning of variance, and it is the data collection
that is particular rather than the resulting analysis. 

However, I have never had to do an analysis of a Latin Square,
so if you have some person or reference that tells you otherwise,
do pay attention to that, too.

--
Rich Ulrich

> Date: Fri, 5 Apr 2013 12:33:24 -0700

> From: [hidden email]
> Subject: Graduate student needing latin square help please
> To: [hidden email]
>
> I am a graduate student who is STRUGGLING to figure out how to run my
> collected data using a latin square. I did a mouse study that tested six
> mice, influenced by six scents, for six nights, over a six month period (six
> new mice were studied for each month). I simply do not know how to structure
> my data so that it will run. The response variable is th GUD's or difference
> in food intake. I could probably include time as a response variable but at
> this point I'm not willing to consider that as an additional factor. Can any
> one help???
>
> data RCB;
>
> input night scent $ food @@;
>
> datalines;
>
> 1 v 1.60 1 W 2.16 1 C 1.70 1 RA 1.08 1 B 1.17 1 RF 0.69
>
> 2 B 0.94 2 RF 1.58 2 V 1.33 2 W 0.51 2 C 0.07 2 RA 1.41
>
> 3 C 0.08 3 RA 1.53 3 B 1.78 3 RF 0.05 3 V 1.38 3 W 0.44
>
> 4 W 3.46 4 C 0.08 4 RA 1.85 4 B 1.43 4 RF 0.57 4 V 0.49
>
> 5 RF 1.04 5 V 1.69 5 W 2.67 5 C 2.03 5 RA 0.33 5 B 1.02
>
> 6 B 0.71 6 B 0.68 6 RF 2.56 6 V 0.76 6 W 0.61 6 C 0.57
>
> ;
>
> proc anova data=RCB;
>
> class night scent;
>
> model food = night scent;
>
> run;
>
>
> Leaving out the month of the year, I was able to run my data but I must
> figure out how to incorporate the months in too. The letters correspond
> with the scents, the single digit numbers correspond with the mouse/night
> and the three digit number is the GUD. Thank you
>
>
>
>
> --
> View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Graduate-student-needing-latin-square-help-please-tp5719290.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