You are on page 1of 111

MSAN-624 V. Dimitrova P.

CONJOINT ANALYSIS

EXAMPLE 1: Conjoint analysis, one respondent (dataset = drink)

This conjoint study pertains to soft drinks. The alternatives are:

1) Brand: Coke vs. Pepsi

2) Packaging: Bottle vs. Can

3) Calories (per ounce): 0 vs. 15

4) Price (per ounce): 8c, 10c or 12c

PROC IMPORT DATAFILE='e:\sasdatasets\MSAN624\Lecture 2.xls'


OUT=sasuser.new REPLACE; SHEET='drink'; GETNAMES=yes; run;
DATA SFG; SET sasuser.new;

ODS GRAPHICS ON;


ODS Exclude notes anova liberalanova conservanova
mvanova liberalmvanova conservmvanova;
PROC TRANSREG utilities PLOT=transformation;
MODEL monotone(Rank / reflect) = class(Brand Pack Calories Price / zero=sum);
OUTPUT OUT=T ireplace COE Predicted CLM Residuals Leverage;
run;
ODS GRAPHICS OFF;

PROC PRINT label;


var Rank TRank PRank Brand Pack Calories Price RRank Leverage;
label TRank= 'Transformed Ranks' PRank = 'Predicted Ranks';
run;

1
MSAN-624 V. Dimitrova P. 2

The ODS Exclude … conservmvanova statement suppresses the printing of unnecessary

output. The utilities keyword in the PROC TRANSREG statement is very important

because it instructs SAS to estimate a conjoint analysis. We use the monotone(Rank /

reflect) transformation of the dependent variable Rank because it is ranked (ordinal)

data; reflect reverses the ranks so that the highest rank indicates the most preferred

alternative. class (Brand Pack Calories Price) codes the independent variables as

dummy variables and zero=sum constrains the sum of the coefficient estimates, for

each variable, to be equal to 0. The OUTPUT statement creates a new dataset

containing the coded independent variables (ireplace), the coefficient estimates

(COE), predicted ranks (Predicted), confidence limits (CLM), residuals (Residuals), and

leverages (Leverage).

2
SAS Output Page 1 of 4

The SAS System

The TRANSREG Procedure

Dependent Variable Monotone(Rank)


Rank

Class Level Information


Class Levels Values
Brand 2 Coke Pepsi
Pack 2 Bottle Can
Calories 2 0 15
Price 3 8 10 12

Number of Observations Read 12


Number of Observations Used 12

TRANSREG Univariate Algorithm Iteration History for Monotone(Rank)


Iteration Average Maximum Criterion
Number Change Change R-Square Change Note
1 0.13066 0.42383 0.96853
2 0.00000 0.00000 1.00000 0.03147 Converged

Algorithm converged.

The TRANSREG Procedure Hypothesis Tests for Monotone(Rank)


Rank

Root MSE 0 R-Square 1.0000


Dependent Mean 6.50000 Adj R-Sq 1.0000
Coeff Var 0

Utilities Table Based on the Usual Degrees of Freedom


Importance
(% Utility
Label Utility Standard Error Range) Variable
Intercept 6.5000 0 Intercept
Brand Coke -1.1855 0 17.500 Class.BrandCoke
Brand Pepsi 1.1855 0 Class.BrandPepsi
Pack Bottle 0.6774 0 10.000 Class.PackBottle

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/26/2014
SAS Output Page 2 of 4

Pack Can -0.6774 0 Class.PackCan


Calories 0 2.3709 0 35.000 Class.Calories0
Calories 15 -2.3709 0 Class.Calories15
Price 8 2.2863 0 37.500 Class.Price8
Price 10 0.5081 0 Class.Price10
Price 12 -2.7943 0 Class.Price12

The standard errors are not adjusted for the fact that the dependent variable was transformed and so
are generally liberal (too small).

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/26/2014
SAS Output Page 3 of 4

The SAS System

The TRANSREG Procedure

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/26/2014
SAS Output Page 4 of 4

The SAS System

Obs Rank Transformed Predicted Brand Pack Calories Price Rank Leverage
Ranks Ranks Residuals
1 4 9.2943 9.2943 Coke Can 0 8 0 0.5
2 2 10.6491 10.6491 Coke Bottle 0 8 0 0.5
3 6 6.9234 6.9234 Pepsi Can 15 8 8.88178E-16 0.5
4 5 8.2782 8.2782 Pepsi Bottle 15 8 0 0.5
5 11 2.7742 2.7742 Coke Can 15 10 0 0.5
6 9 4.1291 4.1291 Coke Bottle 15 10 0 0.5
7 3 9.8871 9.8871 Pepsi Can 0 10 0 0.5
8 1 11.2419 11.2419 Pepsi Bottle 0 10 0 0.5
9 12 -0.5281 -0.5281 Coke Can 15 12 0 0.5
10 8 5.5686 5.5686 Coke Bottle 0 12 -1.7764E-15 0.5
11 7 6.5847 6.5847 Pepsi Can 0 12 -1.7764E-15 0.5
12 10 3.1976 3.1976 Pepsi Bottle 15 12 4.44089E-16 0.5
13 . . . . . . .
14 . . . . . . .

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/26/2014
MSAN-624 V. Dimitrova P. 3

EXAMPLE 2: How to create an experimental design for metric/non-metric conjoint?

An orange juice study involves four attributes:

1) Brand: Minute Maid (1), Fairlee (2), and President’s Choice (3)

2) Preparation: Fresh frozen (1), From concentrate (2), and Pasteurized and

frozen (3)

3) Packaging: Glass jar (1), Aluminum can (2), and Tetra-Pack (3)

4) Price: $0.60, $0.80, and $1.00

Respondents would have to rate 81 alternatives if we asked them to evaluate all


combinations (a full factorial). We therefore search for an optimal design. Generally,
we optimize designs either because there are too many products to be evaluated
and/or the complete set includes unrealistic alternatives.

Optimal designs are provided by computer algorithms. These designs are especially
useful when classical designs such as factorials cannot be implemented. Typically,
optimal design matrices are not orthogonal and effect estimates are correlated.
Optimal design generation involves straight optimizations based on a selected
optimality criterion and the model that will be fit; in other words, design optimality is
model dependent. Three optimality criteria are commonly employed. D optimality
maximizes the determinant of the information matrix X’X for a selected model or, in
other words, minimizes the variance-covariance matrix of the vector of parameter
estimates which equals σ2(X’X)-1. A optimality minimizes the trace of (X’X)-1; in other
words, it minimizes the sum of the variances of the parameter estimates for the
selected model. G optimality is based on σM, the maximum standard error for
prediction over the candidate set. The formulas for these 3 measures of design
optimality or efficiency are:

3
MSAN-624 V. Dimitrova P. 4

Here is a design optimization SAS program for the orange juice study.

DATA Include; INPUT Brand Prepare Pack Price;


DATALINES;
1 1 1 100
2 1 1 100
3 1 1 100
;

PROC PLAN ORDERED SEED=46062;


FACTORS Price = 3
Brand = 3
Prepare = 3
Pack = 3 / NOPRINT;
OUTPUT OUT=T0
Price Nvals = (60 80 100)
Brand Nvals = (1 2 3)
Prepare Nvals = (1 2 3)
Pack Nvals = (1 2 3);
run;

DATA T0; SET T0;


if (^((Prepare = 1) & (Price = 60)));
if (^((Pack = 1) & (Price = 60)));

PROC OPTEX SEED=27513 DATA=T0;


CLASS Brand Prepare Pack Price;
MODEL Brand Prepare Pack Price;
GENERATE n=Saturated iter=10 AUGMENT=Include;
EXAMINE number=1 design info var;
OUTPUT OUT=T number=1;

PROC PRINT DATA=T; run;

4
MSAN-624 V. Dimitrova P. 5

First we create a dataset containing all possible attribute combinations. Note that
attribute levels must be designated by numbers.

PROC PLAN ORDERED SEED=46062;


FACTORS Price = 3
Brand = 3
Prepare = 3
Pack = 3 / NOPRINT;
OUTPUT OUT=T0
Price Nvals = (60 80 100)
Brand Nvals = (1 2 3)
Prepare Nvals = (1 2 3)
Pack Nvals = (1 2 3);
run;

ORDERED instructs SAS to select attribute levels in order 1, 2, etc. Without the SEED
keyword we might not be able to replicate the design.

We exclude unrealistic combinations of the attribute levels:

DATA T0; SET T0;


if (^((Prepare = 1) & (Price = 60)));
if (^((Pack = 1) & (Price = 60)));

We also create a dataset containing product attribute combinations that we want to


include into our study.

DATA Include; INPUT Brand Prepare Pack Price;


DATALINES;
1 1 1 100
2 1 1 100
3 1 1 100
;

The AUGMENT=Include statement in PROC OPTEX incorporates these attribute


combinations into the experimental design.

5
MSAN-624 V. Dimitrova P. 6

PROC OPTEX SEED=27513 DATA=T0;


CLASS Brand Prepare Pack Price;
MODEL Brand Prepare Pack Price;
GENERATE n=Saturated iter=10 AUGMENT=Include;
EXAMINE number=1 design info var;
OUTPUT OUT=T number=1;

First we evaluate a saturated design comprising the same number of design points as

the number of parameters. This design estimates parameters without error and

therefore it is not preferred. We instruct SAS to generate 10 designs (iterations) and

allow us to EXAMINE the best performing design (number=1) by printing the design, its

information matrix, and the variance matrix of the coefficient estimates (EXAMINE

number=1 design info var). If we do not wish to examine the best performing design

we can simply print it (OUTPUT OUT=T number=1). Because saturated designs estimate

parameters without error, we may double the number of design points by replacing

n=Saturated with n = twice the number of design points of the saturated design.

6
SAS Output Page 1 of 5

The SAS System

The OPTEX Procedure

Class Level Information


Class Levels Values
Brand 3 1 2 3
Prepare 3 1 2 3
Pack 3 1 2 3
Price 3 60 80 100

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/26/2014
SAS Output Page 2 of 5

The SAS System

The OPTEX Procedure

Average Prediction
Design Number D-Efficiency A-Efficiency G-Efficiency Standard Error
1 65.4200 35.2941 34.2997 1.6833
2 65.4200 35.2941 34.2997 1.6833
3 65.4200 34.2857 35.9211 1.6283
4 65.4200 34.1232 35.0823 1.6663
5 65.4200 34.1232 35.0823 1.6663
6 61.3686 36.0000 39.0567 1.6576
7 61.3686 34.1772 37.7964 1.6697
8 61.3686 33.3333 33.7526 1.7086
9 61.3686 32.5301 33.7526 1.7437
10 56.0809 13.8462 17.9605 2.7080

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/26/2014
SAS Output Page 3 of 5

The SAS System

The OPTEX Procedure


Examining Design Number 1

Log determinant of the information matrix 1.5956E+01


Maximum prediction variance over candidates 8.5000
Average prediction variance over candidates 2.8333
Average variance of coefficients 0.3148
D-Efficiency 65.4200
A-Efficiency 35.2941

Point Number Brand Prepare Pack Price


4 3 2 2 60
5 2 3 3 60
6 2 3 2 100
7 2 2 1 80
8 2 1 3 80
9 1 2 3 100

Information Matrix
Intercept Brand1 Brand2 Prepare1 Prepare2 Pack1 Pack2 Price1 Price2
Intercept 9.0 0.0 4.2 2.4 -0.0 1.2 -2.1 -3.7 -2.1
Brand1 0.0 6.0 -0.0 0.0 0.0 -1.5 -2.6 -3.0 0.0
Brand2 4.2 -0.0 12.0 -1.7 -3.0 -0.9 -1.5 0.0 3.0
Prepare1 2.4 0.0 -1.7 9.0 -1.7 4.5 -4.3 -4.5 0.9
Prepare2 -0.0 0.0 -3.0 -1.7 9.0 -0.9 1.5 2.6 1.5
Pack1 1.2 -1.5 -0.9 4.5 -0.9 10.5 -0.9 -4.5 -0.9
Pack2 -2.1 -2.6 -1.5 -4.3 1.5 -0.9 7.5 2.6 -1.5
Price1 -3.7 -3.0 0.0 -4.5 2.6 -4.5 2.6 10.5 2.6
Price2 -2.1 0.0 3.0 0.9 1.5 -0.9 -1.5 2.6 7.5

Variance Matrix
Intercept Brand1 Brand2 Prepare1 Prepare2 Pack1 Pack2 Price1 Price2
Intercept 0.389 0.068 -0.275 -0.204 -0.196 0.068 0.039 0.068 0.275
Brand1 0.068 0.333 -0.000 0.083 -0.048 0.083 0.144 0.167 0.000
Brand2 -0.275 -0.000 0.333 0.241 0.194 -0.048 0.028 -0.000 -0.278

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/26/2014
SAS Output Page 4 of 5

Prepare1 -0.204 0.083 0.241 0.417 0.144 -0.083 0.144 0.083 -0.241
Prepare2 -0.196 -0.048 0.194 0.144 0.250 -0.048 -0.028 -0.048 -0.194
Pack1 0.068 0.083 -0.048 -0.083 -0.048 0.167 0.000 0.083 0.048
Pack2 0.039 0.144 0.028 0.144 -0.028 0.000 0.278 0.048 0.028
Price1 0.068 0.167 -0.000 0.083 -0.048 0.083 0.048 0.250 -0.048
Price2 0.275 0.000 -0.278 -0.241 -0.194 0.048 0.028 -0.048 0.417

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/26/2014
SAS Output Page 5 of 5

The SAS System

Obs Brand Prepare Pack Price


1 3 2 2 60
2 3 1 1 100
3 2 3 3 60
4 2 3 2 100
5 2 2 1 80
6 2 1 3 80
7 2 1 1 100
8 1 2 3 100
9 1 1 1 100

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/26/2014
SAS Output Page 1 of 5

The SAS System

The OPTEX Procedure

Class Level Information


Class Levels Values
Brand 3 1 2 3
Prepare 3 1 2 3
Pack 3 1 2 3
Price 3 60 80 100

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/27/2014
SAS Output Page 2 of 5

The SAS System

The OPTEX Procedure

Average Prediction
Design Number D-Efficiency A-Efficiency G-Efficiency Standard Error
1 65.4200 35.2941 34.2997 1.6833
2 65.4200 35.2941 34.2997 1.6833
3 65.4200 34.2857 35.9211 1.6283
4 65.4200 34.1232 35.0823 1.6663
5 65.4200 34.1232 35.0823 1.6663
6 61.3686 36.0000 39.0567 1.6576
7 61.3686 34.1772 37.7964 1.6697
8 61.3686 33.3333 33.7526 1.7086
9 61.3686 32.5301 33.7526 1.7437
10 56.0809 13.8462 17.9605 2.7080

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/27/2014
SAS Output Page 3 of 5

The SAS System

The OPTEX Procedure


Examining Design Number 6

Log determinant of the information matrix 1.5381E+01


Maximum prediction variance over candidates 6.5556
Average prediction variance over candidates 2.7475
Average variance of coefficients 0.3086
D-Efficiency 61.3686
A-Efficiency 36.0000

Point Number Brand Prepare Pack Price


4 3 2 2 60
5 2 3 2 80
6 2 3 1 80
7 2 1 3 80
8 1 3 3 60
9 1 2 3 100

Information Matrix
Intercept Brand1 Brand2 Prepare1 Prepare2 Pack1 Pack2 Price1 Price2
Intercept 9.0 1.2 2.1 1.2 -2.1 1.2 -2.1 -2.4 0.0
Brand1 1.2 7.5 -0.9 -1.5 -0.9 -3.0 -3.5 -1.5 -0.9
Brand2 2.1 -0.9 10.5 -0.9 -4.5 1.7 0.0 -0.9 7.5
Prepare1 1.2 -1.5 -0.9 10.5 -0.9 3.0 -3.5 -6.0 -3.5
Prepare2 -2.1 -0.9 -4.5 -0.9 7.5 -3.5 3.0 1.7 -3.0
Pack1 1.2 -3.0 1.7 3.0 -3.5 10.5 -0.9 -4.5 -0.9
Pack2 -2.1 -3.5 0.0 -3.5 3.0 -0.9 7.5 4.3 1.5
Price1 -2.4 -1.5 -0.9 -6.0 1.7 -4.5 4.3 9.0 1.7
Price2 0.0 -0.9 7.5 -3.5 -3.0 -0.9 1.5 1.7 9.0

Variance Matrix
Intercept Brand1 Brand2 Prepare1 Prepare2 Pack1 Pack2 Price1 Price2
Intercept 0.136 -0.000 -0.052 0.015 0.026 0.015 0.009 0.030 0.052
Brand1 -0.000 0.333 0.000 0.167 0.096 0.167 0.096 0.167 0.096
Brand2 -0.052 0.000 0.333 -0.032 0.056 -0.032 -0.019 0.032 -0.278

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/27/2014
SAS Output Page 4 of 5

Prepare1 0.015 0.167 -0.032 0.259 0.064 0.093 0.053 0.185 0.128
Prepare2 0.026 0.096 0.056 0.064 0.333 0.160 -0.130 0.128 0.111
Pack1 0.015 0.167 -0.032 0.093 0.160 0.259 -0.043 0.185 0.128
Pack2 0.009 0.096 -0.019 0.053 -0.130 -0.043 0.309 -0.086 -0.037
Price1 0.030 0.167 0.032 0.185 0.128 0.185 -0.086 0.370 0.064
Price2 0.052 0.096 -0.278 0.128 0.111 0.128 -0.037 0.064 0.444

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/27/2014
SAS Output Page 5 of 5

The SAS System

Obs Brand Prepare Pack Price


1 3 2 2 60
2 3 1 1 100
3 2 3 2 80
4 2 3 1 80
5 2 1 3 80
6 2 1 1 100
7 1 3 3 60
8 1 2 3 100
9 1 1 1 100

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/27/2014
MSAN-624 V. Dimitrova P. 7

EXAMPLE 3: How to create an experimental design for choice conjoint?

Use the same program except replace the GENERATE n=Saturated iter=10

AUGMENT=Include statement with two statements:

BLOCKS Structure=(9)3 iter=10;

GENERATE AUGMENT=Include;

The BLOCKS Structure = (9)3 statement instructs SAS to create 9 trials, each

comprising 3 options. The GENERATE AUGMENT=Include statement is necessary because

of the AUGMENT=Include option; otherwise it wouldn’t be needed. Note that you can

specify the number of iterations iter=10 either in the BLOCKS statement or in the

GENERATE statement.

7
MSAN-624 V. Dimitrova P. 8

EXAMPLE 4: Create an experimental design for the following conjoint study

Question

How likely are you to watch this network television program?

Attributes and Levels

Network: Genre

1 = ABC 1 = Drama

2 = CBS 2 = Comedy

3 = Fox 3 = Reality program

4 = NBC

Day of the Week Time:

1 = Monday 8 = 8 o’clock hour

2 = Tuesday 9 = 9 o’clock hour

3 = Wednesday 10 = 10 o’clock hour

4= Thursday

5 = Friday

Length:

30 = Half-hour show

60 = One-hour show

Constraints: Fox does not have network programming in the 10 o’clock hour.

Include CBS half-hour comedies on Monday in the 8 and 9 o’clock slots.

8
MSAN-624 V. Dimitrova P. 9

EXAMPLE 5: Conjoint analysis, multiple respondents (dataset = food)

PROC IMPORT DATAFILE='e:\sasdatasets\MSAN624\Lecture 2.xls'


OUT=sasuser.new REPLACE; SHEET='food'; GETNAMES=yes; run;
DATA SFG; SET sasuser.new;

ODS GRAPHICS ON;


ODS Exclude notes anova liberalanova conservanova
mvanova liberalmvanova conservmvanova;
PROC TRANSREG utilities Method=Morals PLOT=transformation OUTTEST=estimates;
MODEL monotone(Rank / reflect) = class(Ingredient Fat Price Calories /
zero=sum);
OUTPUT OUT=T ireplace COE Predicted CLM Residuals Leverage;
BY ID;
WEIGHT w;
run;
ODS GRAPHICS OFF;

PROC PRINT label;


var Rank TRank PRank Ingredient Fat Price Calories RRank Leverage;
label TRank= 'Transformed Ranks' PRank = 'Predicted Ranks';
run;

PROC PRINT DATA=estimates label;


ID ID; VAR value; FORMAT value 4.2;
where statistic='R-Square';
label value='R Square';
run;

DATA I0; SET estimates;


IF n(importance);

PROC TRANSPOSE DATA=I0 OUT=I; VAR importance; BY ID; run;


DATA I; SET I; rename Col1=Ingredient Col2=Fat Col3=Price Col4=Calories;

PROC PRINT DATA=I label; VAR ingredient fat Price calories;


label ingredient='Ingredient' fat='Fat' Price='Price'
calories='Calories'; run;

9
MSAN-624 V. Dimitrova P. 10

PROC MEANS DATA=I mean; VAR ingredient fat Price calories; run;

PROC CORR NOPRINT Kendall Pearson OUTK=K OUTP=P


DATA=T(where=(_TYPE_=' '));
var TRank;
With PRank;
BY ID;
run;

PROC PRINT DATA=K(where=(_TYPE_='CORR')) label; VAR ID TRank; label


TRank='Correlation'; run;

The WEIGHT w; statement identifies the cases to be included into the estimation

(the active cases) and the holdout cases that will be used for validation. We create

two new datasets OUTTEST=estimates and OUTPUT OUT=T. These datasets contain

statistics that we will need to evaluate the output. The following procedure

extracts the R-Square statistic from the OUTTEST=estimates dataset and creates a

table containing the R-Square statistic for each study participant.

PROC PRINT DATA=estimates label;


ID ID; VAR value; FORMAT value 4.2;
where statistic='R-Square';
label value='R Square';
run;

10
MSAN-624 V. Dimitrova P. 11

The next statements extract the importance statistics by respondent. First, we

delete all rows that do not contain an importance statistic. Then we transpose the

column of importance statistics into a row. Finally, we print the importance of the

different attributes by respondent as well as the average importance for each

attribute.

DATA I0; SET estimates;


IF n(importance);

PROC TRANSPOSE DATA=I0 OUT=I; VAR importance; BY ID; run;


DATA I; SET I; rename Col1=Ingredient Col2=Fat Col3=Price Col4=Calories;

PROC PRINT DATA=I label; VAR ingredient fat Price calories;


label ingredient='Ingredient' fat='Fat' Price='Price'
calories='Calories'; run;

PROC MEANS DATA=I mean; VAR ingredient fat Price calories; run;

11
MSAN-624 V. Dimitrova P. 12

The following procedures estimate the Kendall (for rank data) and Pearson (for

interval data) correlations between the Transformed Ranks (TRank) and the

Predicted Ranks (PRank). Correlations are estimated for holdout cases only.

PROC CORR NOPRINT Kendall Pearson OUTK=K OUTP=P


DATA=T(where=(_TYPE_=' '));
var TRank;
With PRank;
BY ID;
run;

PROC PRINT DATA=K(where=(_TYPE_='CORR')) label; VAR ID TRank;


label TRank='Correlation'; run;

12
SAS Output Page 1 of 58

The SAS System

The TRANSREG Procedure

ID=1

Dependent Variable Monotone(Rank)


Rank

Class Level Information


Class Levels Values
Ingredient 3 Beef Chicken Turkey
Fat 3 258
Price 3 1.99 2.29 2.59
Calories 2 250 350

Number of Observations Read 22


Number of Observations Used 18
Sum of Weights Read 18
Sum of Weights Used 18

TRANSREG MORALS Algorithm Iteration History for Monotone(Rank)


Iteration Average Maximum Criterion
Number Change Change R-Square Change Note
1 0.14463 0.35677 0.96181
2 0.01243 0.03072 0.99955 0.03774
3 0.00408 0.01042 0.99991 0.00036
4 0.00255 0.00729 0.99995 0.00004
5 0.00198 0.00614 0.99996 0.00002
6 0.00160 0.00506 0.99998 0.00001
7 0.00131 0.00415 0.99998 0.00001
8 0.00107 0.00340 0.99999 0.00001
9 0.00088 0.00279 0.99999 0.00000
10 0.00072 0.00229 1.00000 0.00000
11 0.00059 0.00187 1.00000 0.00000
12 0.00048 0.00153 1.00000 0.00000
13 0.00040 0.00126 1.00000 0.00000
14 0.00032 0.00103 1.00000 0.00000

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 2 of 58

15 0.00027 0.00084 1.00000 0.00000


16 0.00022 0.00069 1.00000 0.00000
17 0.00018 0.00057 1.00000 0.00000
18 0.00015 0.00046 1.00000 0.00000
19 0.00012 0.00038 1.00000 0.00000
20 0.00010 0.00031 1.00000 0.00000
21 0.00008 0.00026 1.00000 0.00000
22 0.00007 0.00021 1.00000 0.00000
23 0.00005 0.00017 1.00000 0.00000
24 0.00004 0.00014 1.00000 0.00000
25 0.00004 0.00011 1.00000 0.00000
26 0.00003 0.00009 1.00000 0.00000
27 0.00002 0.00008 1.00000 0.00000
28 0.00002 0.00006 1.00000 0.00000
29 0.00002 0.00005 1.00000 0.00000
30 0.00001 0.00004 1.00000 0.00000 Not Converged

WARNING: Failed to converge, however criterion change is less than 0.0001.

The TRANSREG Procedure Hypothesis Tests for Monotone(Rank)


Rank

Root MSE 0.00030960 R-Square 1.0000


Dependent Mean 11.38889 Adj R-Sq 1.0000
Coeff Var 0.00272

Utilities Table Based on the Usual Degrees of Freedom


Importance
(% Utility
Label Utility Standard Error Range) Variable
Intercept 11.3889 0.00007 Intercept
Ingredient Beef -2.1603 0.00010 12.261 Class.IngredientBeef
Ingredient Chicken 1.3543 0.00010 Class.IngredientChicken
Ingredient Turkey 0.8060 0.00010 Class.IngredientTurkey
Fat 2 7.0290 0.00010 49.046 Class.Fat2
Fat 5 0.0003 0.00010 Class.Fat5
Fat 8 -7.0293 0.00010 Class.Fat8

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 3 of 58

Price 1.99 3.5144 0.00010 24.523 Class.Price1D99


Price 2.29 0.0003 0.00010 Class.Price2D29
Price 2.59 -3.5147 0.00010 Class.Price2D59
Calories 250 2.0308 0.00007 14.170 Class.Calories250
Calories 350 -2.0308 0.00007 Class.Calories350

The standard errors are not adjusted for the fact that the dependent variable was transformed and so
are generally liberal (too small).

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 4 of 58

The SAS System

The TRANSREG Procedure

ID=1

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 5 of 58

The SAS System

The TRANSREG Procedure

ID=2

Dependent Variable Monotone(Rank)


Rank

Class Level Information


Class Levels Values
Ingredient 3 Beef Chicken Turkey
Fat 3 258
Price 3 1.99 2.29 2.59
Calories 2 250 350

Number of Observations Read 22


Number of Observations Used 18
Sum of Weights Read 18
Sum of Weights Used 18

TRANSREG MORALS Algorithm Iteration History for Monotone(Rank)


Iteration Average Maximum Criterion
Number Change Change R-Square Change Note
1 0.10816 0.27360 0.97880
2 0.00764 0.01947 0.99969 0.02089
3 0.00450 0.01317 0.99982 0.00013
4 0.00347 0.01076 0.99988 0.00006
5 0.00284 0.00879 0.99992 0.00004
6 0.00237 0.00723 0.99994 0.00003
7 0.00200 0.00598 0.99996 0.00002
8 0.00168 0.00497 0.99997 0.00001
9 0.00142 0.00413 0.99998 0.00001
10 0.00119 0.00345 0.99999 0.00001
11 0.00100 0.00288 0.99999 0.00000
12 0.00084 0.00241 0.99999 0.00000
13 0.00071 0.00201 1.00000 0.00000
14 0.00060 0.00169 1.00000 0.00000

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 6 of 58

15 0.00050 0.00141 1.00000 0.00000


16 0.00042 0.00118 1.00000 0.00000
17 0.00035 0.00099 1.00000 0.00000
18 0.00030 0.00083 1.00000 0.00000
19 0.00025 0.00070 1.00000 0.00000
20 0.00021 0.00059 1.00000 0.00000
21 0.00017 0.00049 1.00000 0.00000
22 0.00015 0.00041 1.00000 0.00000
23 0.00012 0.00035 1.00000 0.00000
24 0.00010 0.00029 1.00000 0.00000
25 0.00009 0.00024 1.00000 0.00000
26 0.00007 0.00020 1.00000 0.00000
27 0.00006 0.00017 1.00000 0.00000
28 0.00005 0.00014 1.00000 0.00000
29 0.00004 0.00012 1.00000 0.00000
30 0.00004 0.00010 1.00000 0.00000 Not Converged

WARNING: Failed to converge, however criterion change is less than 0.0001.

The TRANSREG Procedure Hypothesis Tests for Monotone(Rank)


Rank

Root MSE 0.00085335 R-Square 1.0000


Dependent Mean 11.77778 Adj R-Sq 1.0000
Coeff Var 0.00725

Utilities Table Based on the Usual Degrees of Freedom


Importance
(% Utility
Label Utility Standard Error Range) Variable
Intercept 11.7778 0.00020 Intercept
Ingredient Beef 0.0956 0.00028 8.489 Class.IngredientBeef
Ingredient Chicken -1.0878 0.00028 Class.IngredientChicken
Ingredient Turkey 0.9922 0.00028 Class.IngredientTurkey
Fat 2 7.4759 0.00028 62.398 Class.Fat2
Fat 5 0.3377 0.00028 Class.Fat5
Fat 8 -7.8136 0.00028 Class.Fat8

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 7 of 58

Price 1.99 2.0782 0.00028 16.968 Class.Price1D99


Price 2.29 0.0012 0.00028 Class.Price2D29
Price 2.59 -2.0794 0.00028 Class.Price2D59
Calories 250 1.4881 0.00020 12.146 Class.Calories250
Calories 350 -1.4881 0.00020 Class.Calories350

The standard errors are not adjusted for the fact that the dependent variable was transformed and so
are generally liberal (too small).

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 8 of 58

The SAS System

The TRANSREG Procedure

ID=2

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 9 of 58

The SAS System

The TRANSREG Procedure

ID=3

Dependent Variable Monotone(Rank)


Rank

Class Level Information


Class Levels Values
Ingredient 3 Beef Chicken Turkey
Fat 3 258
Price 3 1.99 2.29 2.59
Calories 2 250 350

Number of Observations Read 22


Number of Observations Used 18
Sum of Weights Read 18
Sum of Weights Used 18

TRANSREG MORALS Algorithm Iteration History for Monotone(Rank)


Iteration Average Maximum Criterion
Number Change Change R-Square Change Note
1 0.09589 0.20193 0.98442
2 0.00831 0.03054 0.99943 0.01501
3 0.00620 0.02154 0.99968 0.00025
4 0.00483 0.01587 0.99981 0.00012
5 0.00381 0.01203 0.99988 0.00007
6 0.00302 0.00929 0.99992 0.00004
7 0.00228 0.00725 0.99995 0.00003
8 0.00170 0.00542 0.99997 0.00002
9 0.00132 0.00357 0.99998 0.00001
10 0.00113 0.00292 0.99998 0.00001
11 0.00101 0.00250 0.99999 0.00000
12 0.00091 0.00220 0.99999 0.00000
13 0.00077 0.00196 0.99999 0.00000
14 0.00060 0.00169 0.99999 0.00000

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 10 of 58

15 0.00050 0.00142 0.99999 0.00000


16 0.00045 0.00124 0.99999 0.00000
17 0.00042 0.00113 0.99999 0.00000
18 0.00040 0.00106 1.00000 0.00000
19 0.00038 0.00099 1.00000 0.00000
20 0.00035 0.00093 1.00000 0.00000
21 0.00033 0.00087 1.00000 0.00000
22 0.00031 0.00082 1.00000 0.00000
23 0.00030 0.00077 1.00000 0.00000
24 0.00028 0.00073 1.00000 0.00000
25 0.00026 0.00069 1.00000 0.00000
26 0.00025 0.00065 1.00000 0.00000
27 0.00024 0.00061 1.00000 0.00000
28 0.00022 0.00058 1.00000 0.00000
29 0.00021 0.00055 1.00000 0.00000
30 0.00020 0.00051 1.00000 0.00000 Not Converged

WARNING: Failed to converge, however criterion change is less than 0.0001.

The TRANSREG Procedure Hypothesis Tests for Monotone(Rank)


Rank

Root MSE 0.00950 R-Square 1.0000


Dependent Mean 11.66667 Adj R-Sq 1.0000
Coeff Var 0.08145

Utilities Table Based on the Usual Degrees of Freedom


Importance
(% Utility
Label Utility Standard Error Range) Variable
Intercept 11.6667 0.00224 Intercept
Ingredient Beef -1.4178 0.00317 9.264 Class.IngredientBeef
Ingredient Chicken 0.9697 0.00317 Class.IngredientChicken
Ingredient Turkey 0.4481 0.00317 Class.IngredientTurkey
Fat 2 7.7971 0.00317 60.399 Class.Fat2
Fat 5 -0.0289 0.00317 Class.Fat5
Fat 8 -7.7682 0.00317 Class.Fat8

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 11 of 58

Price 1.99 2.4398 0.00317 18.877 Class.Price1D99


Price 2.29 -0.0148 0.00317 Class.Price2D29
Price 2.59 -2.4250 0.00317 Class.Price2D59
Calories 250 1.4766 0.00224 11.459 Class.Calories250
Calories 350 -1.4766 0.00224 Class.Calories350

The standard errors are not adjusted for the fact that the dependent variable was transformed and so
are generally liberal (too small).

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 12 of 58

The SAS System

The TRANSREG Procedure

ID=3

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 13 of 58

The SAS System

The TRANSREG Procedure

ID=4

Dependent Variable Monotone(Rank)


Rank

Class Level Information


Class Levels Values
Ingredient 3 Beef Chicken Turkey
Fat 3 258
Price 3 1.99 2.29 2.59
Calories 2 250 350

Number of Observations Read 22


Number of Observations Used 18
Sum of Weights Read 18
Sum of Weights Used 18

TRANSREG MORALS Algorithm Iteration History for Monotone(Rank)


Iteration Average Maximum Criterion
Number Change Change R-Square Change Note
1 0.08926 0.29201 0.98491
2 0.00511 0.01411 0.99974 0.01483
3 0.00423 0.00822 0.99981 0.00007
4 0.00362 0.00722 0.99986 0.00005
5 0.00315 0.00670 0.99989 0.00003
6 0.00274 0.00606 0.99992 0.00003
7 0.00239 0.00542 0.99994 0.00002
8 0.00208 0.00482 0.99995 0.00001
9 0.00181 0.00427 0.99996 0.00001
10 0.00158 0.00377 0.99997 0.00001
11 0.00138 0.00333 0.99998 0.00001
12 0.00121 0.00293 0.99998 0.00001
13 0.00105 0.00258 0.99999 0.00000
14 0.00092 0.00227 0.99999 0.00000

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 14 of 58

15 0.00080 0.00199 0.99999 0.00000


16 0.00070 0.00175 0.99999 0.00000
17 0.00062 0.00153 1.00000 0.00000
18 0.00054 0.00134 1.00000 0.00000
19 0.00047 0.00118 1.00000 0.00000
20 0.00041 0.00103 1.00000 0.00000
21 0.00036 0.00090 1.00000 0.00000
22 0.00032 0.00079 1.00000 0.00000
23 0.00028 0.00069 1.00000 0.00000
24 0.00024 0.00061 1.00000 0.00000
25 0.00021 0.00053 1.00000 0.00000
26 0.00018 0.00047 1.00000 0.00000
27 0.00016 0.00041 1.00000 0.00000
28 0.00014 0.00036 1.00000 0.00000
29 0.00012 0.00031 1.00000 0.00000
30 0.00011 0.00027 1.00000 0.00000 Not Converged

WARNING: Failed to converge, however criterion change is less than 0.0001.

The TRANSREG Procedure Hypothesis Tests for Monotone(Rank)


Rank

Root MSE 0.00280 R-Square 1.0000


Dependent Mean 11.72222 Adj R-Sq 1.0000
Coeff Var 0.02389

Utilities Table Based on the Usual Degrees of Freedom


Importance
(% Utility
Label Utility Standard Error Range) Variable
Intercept 11.7222 0.00066 Intercept
Ingredient Beef 0.6566 0.00093 12.895 Class.IngredientBeef
Ingredient Chicken -2.0187 0.00093 Class.IngredientChicken
Ingredient Turkey 1.3622 0.00093 Class.IngredientTurkey
Fat 2 3.0824 0.00093 22.719 Class.Fat2
Fat 5 -0.2082 0.00093 Class.Fat5
Fat 8 -2.8742 0.00093 Class.Fat8

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 15 of 58

Price 1.99 3.3716 0.00093 25.742 Class.Price1D99


Price 2.29 0.0062 0.00093 Class.Price2D29
Price 2.59 -3.3778 0.00093 Class.Price2D59
Calories 250 5.0659 0.00066 38.644 Class.Calories250
Calories 350 -5.0659 0.00066 Class.Calories350

The standard errors are not adjusted for the fact that the dependent variable was transformed and so
are generally liberal (too small).

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 16 of 58

The SAS System

The TRANSREG Procedure

ID=4

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 17 of 58

The SAS System

The TRANSREG Procedure

ID=5

Dependent Variable Monotone(Rank)


Rank

Class Level Information


Class Levels Values
Ingredient 3 Beef Chicken Turkey
Fat 3 258
Price 3 1.99 2.29 2.59
Calories 2 250 350

Number of Observations Read 22


Number of Observations Used 18
Sum of Weights Read 18
Sum of Weights Used 18

TRANSREG MORALS Algorithm Iteration History for Monotone(Rank)


Iteration Average Maximum Criterion
Number Change Change R-Square Change Note
1 0.08984 0.17877 0.98539
2 0.00741 0.01472 0.99961 0.01422
3 0.00581 0.01235 0.99975 0.00014
4 0.00463 0.01037 0.99983 0.00008
5 0.00372 0.00874 0.99989 0.00005
6 0.00302 0.00738 0.99992 0.00003
7 0.00247 0.00625 0.99994 0.00002
8 0.00204 0.00531 0.99996 0.00002
9 0.00169 0.00451 0.99997 0.00001
10 0.00142 0.00384 0.99998 0.00001
11 0.00119 0.00328 0.99998 0.00001
12 0.00100 0.00280 0.99999 0.00000
13 0.00085 0.00239 0.99999 0.00000
14 0.00072 0.00205 0.99999 0.00000

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 18 of 58

15 0.00062 0.00175 1.00000 0.00000


16 0.00053 0.00150 1.00000 0.00000
17 0.00045 0.00129 1.00000 0.00000
18 0.00038 0.00110 1.00000 0.00000
19 0.00033 0.00094 1.00000 0.00000
20 0.00028 0.00081 1.00000 0.00000
21 0.00024 0.00069 1.00000 0.00000
22 0.00021 0.00060 1.00000 0.00000
23 0.00018 0.00051 1.00000 0.00000
24 0.00015 0.00044 1.00000 0.00000
25 0.00013 0.00038 1.00000 0.00000
26 0.00011 0.00032 1.00000 0.00000
27 0.00010 0.00028 1.00000 0.00000
28 0.00008 0.00024 1.00000 0.00000
29 0.00007 0.00020 1.00000 0.00000
30 0.00006 0.00017 1.00000 0.00000 Not Converged

WARNING: Failed to converge, however criterion change is less than 0.0001.

The TRANSREG Procedure Hypothesis Tests for Monotone(Rank)


Rank

Root MSE 0.00156 R-Square 1.0000


Dependent Mean 11.22222 Adj R-Sq 1.0000
Coeff Var 0.01389

Utilities Table Based on the Usual Degrees of Freedom


Importance
(% Utility
Label Utility Standard Error Range) Variable
Intercept 11.2222 0.00037 Intercept
Ingredient Beef 0.3699 0.00052 8.956 Class.IngredientBeef
Ingredient Chicken 0.8514 0.00052 Class.IngredientChicken
Ingredient Turkey -1.2213 0.00052 Class.IngredientTurkey
Fat 2 2.0735 0.00052 17.908 Class.Fat2
Fat 5 -0.0026 0.00052 Class.Fat5
Fat 8 -2.0709 0.00052 Class.Fat8

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 19 of 58

Price 1.99 7.0234 0.00052 62.070 Class.Price1D99


Price 2.29 0.3180 0.00052 Class.Price2D29
Price 2.59 -7.3414 0.00052 Class.Price2D59
Calories 250 1.2805 0.00037 11.066 Class.Calories250
Calories 350 -1.2805 0.00037 Class.Calories350

The standard errors are not adjusted for the fact that the dependent variable was transformed and so
are generally liberal (too small).

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 20 of 58

The SAS System

The TRANSREG Procedure

ID=5

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 21 of 58

The SAS System

The TRANSREG Procedure

ID=6

Dependent Variable Monotone(Rank)


Rank

Class Level Information


Class Levels Values
Ingredient 3 Beef Chicken Turkey
Fat 3 258
Price 3 1.99 2.29 2.59
Calories 2 250 350

Number of Observations Read 22


Number of Observations Used 18
Sum of Weights Read 18
Sum of Weights Used 18

TRANSREG MORALS Algorithm Iteration History for Monotone(Rank)


Iteration Average Maximum Criterion
Number Change Change R-Square Change Note
1 0.11679 0.30132 0.96357
2 0.02337 0.05841 0.99690 0.03333
3 0.01596 0.03490 0.99847 0.00157
4 0.01175 0.02357 0.99916 0.00069
5 0.00880 0.01877 0.99953 0.00037
6 0.00664 0.01436 0.99973 0.00021
7 0.00500 0.01084 0.99985 0.00012
8 0.00376 0.00814 0.99992 0.00007
9 0.00282 0.00611 0.99995 0.00004
10 0.00212 0.00458 0.99997 0.00002
11 0.00159 0.00343 0.99998 0.00001
12 0.00119 0.00257 0.99999 0.00001
13 0.00090 0.00192 1.00000 0.00000
14 0.00067 0.00144 1.00000 0.00000

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 22 of 58

15 0.00050 0.00108 1.00000 0.00000


16 0.00038 0.00080 1.00000 0.00000
17 0.00028 0.00060 1.00000 0.00000
18 0.00021 0.00045 1.00000 0.00000
19 0.00016 0.00034 1.00000 0.00000
20 0.00012 0.00025 1.00000 0.00000
21 0.00009 0.00019 1.00000 0.00000
22 0.00007 0.00014 1.00000 0.00000
23 0.00005 0.00010 1.00000 0.00000
24 0.00004 0.00008 1.00000 0.00000
25 0.00003 0.00006 1.00000 0.00000
26 0.00002 0.00004 1.00000 0.00000
27 0.00002 0.00003 1.00000 0.00000
28 0.00001 0.00002 1.00000 0.00000
29 0.00001 0.00002 1.00000 0.00000 Converged

Algorithm converged.

The TRANSREG Procedure Hypothesis Tests for Monotone(Rank)


Rank

Root MSE 0.00015270 R-Square 1.0000


Dependent Mean 11.27778 Adj R-Sq 1.0000
Coeff Var 0.00135

Utilities Table Based on the Usual Degrees of Freedom


Importance
(% Utility
Label Utility Standard Error Range) Variable
Intercept 11.2778 0.00004 Intercept
Ingredient Beef 0.8897 0.00005 12.347 Class.IngredientBeef
Ingredient Chicken 1.2196 0.00005 Class.IngredientChicken
Ingredient Turkey -2.1094 0.00005 Class.IngredientTurkey
Fat 2 3.3292 0.00005 24.694 Class.Fat2
Fat 5 -0.0002 0.00005 Class.Fat5
Fat 8 -3.3290 0.00005 Class.Fat8
Price 1.99 6.6584 0.00005 49.389 Class.Price1D99

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 23 of 58

Price 2.29 -0.0001 0.00005 Class.Price2D29


Price 2.59 -6.6582 0.00005 Class.Price2D59
Calories 250 1.8295 0.00004 13.570 Class.Calories250
Calories 350 -1.8295 0.00004 Class.Calories350

The standard errors are not adjusted for the fact that the dependent variable was transformed and so
are generally liberal (too small).

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 24 of 58

The SAS System

The TRANSREG Procedure

ID=6

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 25 of 58

The SAS System

The TRANSREG Procedure

ID=7

Dependent Variable Monotone(Rank)


Rank

Class Level Information


Class Levels Values
Ingredient 3 Beef Chicken Turkey
Fat 3 258
Price 3 1.99 2.29 2.59
Calories 2 250 350

Number of Observations Read 22


Number of Observations Used 18
Sum of Weights Read 18
Sum of Weights Used 18

TRANSREG MORALS Algorithm Iteration History for Monotone(Rank)


Iteration Average Maximum Criterion
Number Change Change R-Square Change Note
1 0.09259 0.21516 0.98565
2 0.00550 0.01638 0.99981 0.01416
3 0.00366 0.01092 0.99992 0.00010
4 0.00243 0.00727 0.99996 0.00005
5 0.00162 0.00485 0.99998 0.00002
6 0.00108 0.00323 0.99999 0.00001
7 0.00072 0.00215 1.00000 0.00000
8 0.00048 0.00144 1.00000 0.00000
9 0.00032 0.00096 1.00000 0.00000
10 0.00021 0.00064 1.00000 0.00000
11 0.00014 0.00043 1.00000 0.00000
12 0.00009 0.00028 1.00000 0.00000
13 0.00006 0.00019 1.00000 0.00000
14 0.00004 0.00013 1.00000 0.00000

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 26 of 58

15 0.00003 0.00008 1.00000 0.00000


16 0.00002 0.00006 1.00000 0.00000
17 0.00001 0.00004 1.00000 0.00000
18 0.00001 0.00002 1.00000 0.00000 Converged

Algorithm converged.

The TRANSREG Procedure Hypothesis Tests for Monotone(Rank)


Rank

Root MSE 0.00012449 R-Square 1.0000


Dependent Mean 11.88889 Adj R-Sq 1.0000
Coeff Var 0.00105

Utilities Table Based on the Usual Degrees of Freedom


Importance
(% Utility
Label Utility Standard Error Range) Variable
Intercept 11.8889 0.00003 Intercept
Ingredient Beef 0.8956 0.00004 8.758 Class.IngredientBeef
Ingredient Chicken 0.2239 0.00004 Class.IngredientChicken
Ingredient Turkey -1.1195 0.00004 Class.IngredientTurkey
Fat 2 7.9495 0.00004 67.884 Class.Fat2
Fat 5 -0.2799 0.00004 Class.Fat5
Fat 8 -7.6695 0.00004 Class.Fat8
Price 1.99 1.5675 0.00004 14.599 Class.Price1D99
Price 2.29 0.2240 0.00004 Class.Price2D29
Price 2.59 -1.7915 0.00004 Class.Price2D59
Calories 250 1.0077 0.00003 8.759 Class.Calories250
Calories 350 -1.0077 0.00003 Class.Calories350

The standard errors are not adjusted for the fact that the dependent variable was transformed and so
are generally liberal (too small).

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 27 of 58

The SAS System

The TRANSREG Procedure

ID=7

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 28 of 58

The SAS System

The TRANSREG Procedure

ID=8

Dependent Variable Monotone(Rank)


Rank

Class Level Information


Class Levels Values
Ingredient 3 Beef Chicken Turkey
Fat 3 258
Price 3 1.99 2.29 2.59
Calories 2 250 350

Number of Observations Read 22


Number of Observations Used 18
Sum of Weights Read 18
Sum of Weights Used 18

TRANSREG MORALS Algorithm Iteration History for Monotone(Rank)


Iteration Average Maximum Criterion
Number Change Change R-Square Change Note
1 0.06609 0.14900 0.99149
2 0.00610 0.01607 0.99986 0.00836
3 0.00354 0.00938 0.99995 0.00009
4 0.00206 0.00547 0.99998 0.00003
5 0.00120 0.00319 0.99999 0.00001
6 0.00070 0.00186 1.00000 0.00000
7 0.00041 0.00108 1.00000 0.00000
8 0.00024 0.00063 1.00000 0.00000
9 0.00014 0.00037 1.00000 0.00000
10 0.00008 0.00021 1.00000 0.00000
11 0.00005 0.00012 1.00000 0.00000
12 0.00003 0.00007 1.00000 0.00000
13 0.00002 0.00004 1.00000 0.00000
14 0.00001 0.00002 1.00000 0.00000 Converged

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 29 of 58

Algorithm converged.

The TRANSREG Procedure Hypothesis Tests for Monotone(Rank)


Rank

Root MSE 0.00009167 R-Square 1.0000


Dependent Mean 11.16667 Adj R-Sq 1.0000
Coeff Var 0.00082093

Utilities Table Based on the Usual Degrees of Freedom


Importance
(% Utility
Label Utility Standard Error Range) Variable
Intercept 11.1667 0.00002 Intercept
Ingredient Beef -0.5023 0.00003 3.974 Class.IngredientBeef
Ingredient Chicken 0.4038 0.00003 Class.IngredientChicken
Ingredient Turkey 0.0984 0.00003 Class.IngredientTurkey
Fat 2 2.4129 0.00003 21.598 Class.Fat2
Fat 5 0.0984 0.00003 Class.Fat5
Fat 8 -2.5113 0.00003 Class.Fat8
Price 1.99 7.2980 0.00003 64.320 Class.Price1D99
Price 2.29 0.0689 0.00003 Class.Price2D29
Price 2.59 -7.3669 0.00003 Class.Price2D59
Calories 250 1.1523 0.00002 10.108 Class.Calories250
Calories 350 -1.1523 0.00002 Class.Calories350

The standard errors are not adjusted for the fact that the dependent variable was transformed and so
are generally liberal (too small).

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 30 of 58

The SAS System

The TRANSREG Procedure

ID=8

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 31 of 58

The SAS System

The TRANSREG Procedure

ID=9

Dependent Variable Monotone(Rank)


Rank

Class Level Information


Class Levels Values
Ingredient 3 Beef Chicken Turkey
Fat 3 258
Price 3 1.99 2.29 2.59
Calories 2 250 350

Number of Observations Read 22


Number of Observations Used 18
Sum of Weights Read 18
Sum of Weights Used 18

TRANSREG MORALS Algorithm Iteration History for Monotone(Rank)


Iteration Average Maximum Criterion
Number Change Change R-Square Change Note
1 0.08733 0.19850 0.98503
2 0.00883 0.02359 0.99972 0.01469
3 0.00529 0.01406 0.99990 0.00018
4 0.00317 0.00837 0.99996 0.00007
5 0.00191 0.00498 0.99999 0.00002
6 0.00115 0.00295 1.00000 0.00001
7 0.00069 0.00175 1.00000 0.00000
8 0.00042 0.00103 1.00000 0.00000
9 0.00026 0.00060 1.00000 0.00000
10 0.00017 0.00043 1.00000 0.00000
11 0.00011 0.00033 1.00000 0.00000
12 0.00008 0.00026 1.00000 0.00000
13 0.00007 0.00021 1.00000 0.00000
14 0.00005 0.00017 1.00000 0.00000

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 32 of 58

15 0.00004 0.00014 1.00000 0.00000


16 0.00004 0.00012 1.00000 0.00000
17 0.00003 0.00010 1.00000 0.00000
18 0.00003 0.00008 1.00000 0.00000
19 0.00002 0.00007 1.00000 0.00000
20 0.00002 0.00006 1.00000 0.00000
21 0.00002 0.00005 1.00000 0.00000
22 0.00001 0.00005 1.00000 0.00000
23 0.00001 0.00004 1.00000 0.00000
24 0.00001 0.00003 1.00000 0.00000
25 0.00001 0.00003 1.00000 0.00000 Converged

Algorithm converged.

The TRANSREG Procedure Hypothesis Tests for Monotone(Rank)


Rank

Root MSE 0.00025942 R-Square 1.0000


Dependent Mean 11.27778 Adj R-Sq 1.0000
Coeff Var 0.00230

Utilities Table Based on the Usual Degrees of Freedom


Importance
(% Utility
Label Utility Standard Error Range) Variable
Intercept 11.2778 0.00006 Intercept
Ingredient Beef -1.3071 0.00009 8.608 Class.IngredientBeef
Ingredient Chicken 0.6548 0.00009 Class.IngredientChicken
Ingredient Turkey 0.6523 0.00009 Class.IngredientTurkey
Fat 2 1.9593 0.00009 17.195 Class.Fat2
Fat 5 0.0004 0.00009 Class.Fat5
Fat 8 -1.9598 0.00009 Class.Fat8
Price 1.99 7.5053 0.00009 65.615 Class.Price1D99
Price 2.29 -0.0560 0.00009 Class.Price2D29
Price 2.59 -7.4493 0.00009 Class.Price2D59
Calories 250 0.9779 0.00006 8.582 Class.Calories250
Calories 350 -0.9779 0.00006 Class.Calories350

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 33 of 58

The standard errors are not adjusted for the fact that the dependent variable was transformed and so
are generally liberal (too small).

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 34 of 58

The SAS System

The TRANSREG Procedure

ID=9

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 35 of 58

The SAS System

The TRANSREG Procedure

ID=10

Dependent Variable Monotone(Rank)


Rank

Class Level Information


Class Levels Values
Ingredient 3 Beef Chicken Turkey
Fat 3 258
Price 3 1.99 2.29 2.59
Calories 2 250 350

Number of Observations Read 22


Number of Observations Used 18
Sum of Weights Read 18
Sum of Weights Used 18

TRANSREG MORALS Algorithm Iteration History for Monotone(Rank)


Iteration Average Maximum Criterion
Number Change Change R-Square Change Note
1 0.07704 0.17098 0.98891
2 0.00825 0.01736 0.99964 0.01073
3 0.00620 0.01261 0.99980 0.00016
4 0.00451 0.00922 0.99989 0.00009
5 0.00312 0.00641 0.99994 0.00005
6 0.00237 0.00411 0.99996 0.00002
7 0.00189 0.00393 0.99997 0.00001
8 0.00156 0.00356 0.99998 0.00001
9 0.00130 0.00315 0.99998 0.00001
10 0.00110 0.00274 0.99999 0.00000
11 0.00094 0.00238 0.99999 0.00000
12 0.00080 0.00205 0.99999 0.00000
13 0.00069 0.00177 1.00000 0.00000
14 0.00059 0.00152 1.00000 0.00000

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 36 of 58

15 0.00050 0.00130 1.00000 0.00000


16 0.00043 0.00112 1.00000 0.00000
17 0.00036 0.00096 1.00000 0.00000
18 0.00030 0.00084 1.00000 0.00000
19 0.00026 0.00074 1.00000 0.00000
20 0.00023 0.00063 1.00000 0.00000
21 0.00020 0.00054 1.00000 0.00000
22 0.00018 0.00047 1.00000 0.00000
23 0.00015 0.00040 1.00000 0.00000
24 0.00014 0.00035 1.00000 0.00000
25 0.00012 0.00030 1.00000 0.00000
26 0.00010 0.00026 1.00000 0.00000
27 0.00009 0.00023 1.00000 0.00000
28 0.00008 0.00020 1.00000 0.00000
29 0.00007 0.00017 1.00000 0.00000
30 0.00006 0.00015 1.00000 0.00000 Not Converged

WARNING: Failed to converge, however criterion change is less than 0.0001.

The TRANSREG Procedure Hypothesis Tests for Monotone(Rank)


Rank

Root MSE 0.00156 R-Square 1.0000


Dependent Mean 11.27778 Adj R-Sq 1.0000
Coeff Var 0.01379

Utilities Table Based on the Usual Degrees of Freedom


Importance
(% Utility
Label Utility Standard Error Range) Variable
Intercept 11.2778 0.00037 Intercept
Ingredient Beef 0.9716 0.00052 9.610 Class.IngredientBeef
Ingredient Chicken -1.3489 0.00052 Class.IngredientChicken
Ingredient Turkey 0.3774 0.00052 Class.IngredientTurkey
Fat 2 2.3221 0.00052 19.221 Class.Fat2
Fat 5 -0.0027 0.00052 Class.Fat5
Fat 8 -2.3194 0.00052 Class.Fat8

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 37 of 58

Price 1.99 6.9905 0.00052 59.071 Class.Price1D99


Price 2.29 0.2833 0.00052 Class.Price2D29
Price 2.59 -7.2738 0.00052 Class.Price2D59
Calories 250 1.4608 0.00037 12.099 Class.Calories250
Calories 350 -1.4608 0.00037 Class.Calories350

The standard errors are not adjusted for the fact that the dependent variable was transformed and so
are generally liberal (too small).

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 38 of 58

The SAS System

The TRANSREG Procedure

ID=10

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 39 of 58

The SAS System

The TRANSREG Procedure

ID=11

Dependent Variable Monotone(Rank)


Rank

Class Level Information


Class Levels Values
Ingredient 3 Beef Chicken Turkey
Fat 3 258
Price 3 1.99 2.29 2.59
Calories 2 250 350

Number of Observations Read 22


Number of Observations Used 18
Sum of Weights Read 18
Sum of Weights Used 18

TRANSREG MORALS Algorithm Iteration History for Monotone(Rank)


Iteration Average Maximum Criterion
Number Change Change R-Square Change Note
1 0.34065 0.74850 0.23932
2 0.11870 0.49757 0.44786 0.20854
3 0.05316 0.28527 0.47745 0.02959
4 0.03622 0.27319 0.48472 0.00727
5 0.02755 0.23449 0.48987 0.00515
6 0.02151 0.05832 0.49352 0.00365
7 0.01050 0.02754 0.49396 0.00044
8 0.00510 0.01316 0.49406 0.00010
9 0.00247 0.00632 0.49408 0.00002
10 0.00119 0.00305 0.49409 0.00001
11 0.00058 0.00147 0.49409 0.00000
12 0.00028 0.00071 0.49409 0.00000
13 0.00014 0.00034 0.49409 0.00000
14 0.00007 0.00017 0.49409 0.00000

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 40 of 58

15 0.00003 0.00008 0.49409 0.00000


16 0.00002 0.00004 0.49409 0.00000
17 0.00001 0.00002 0.49409 0.00000 Converged

Algorithm converged.

The TRANSREG Procedure Hypothesis Tests for Monotone(Rank)


Rank

Root MSE 6.05419 R-Square 0.4941


Dependent Mean 12.16667 Adj R-Sq 0.1400
Coeff Var 49.76050

Utilities Table Based on the Usual Degrees of Freedom


Importance
(% Utility
Label Utility Standard Error Range) Variable
Intercept 12.1667 1.42699 Intercept
Ingredient Beef -2.0957 2.01806 17.339 Class.IngredientBeef
Ingredient Chicken 1.0479 2.01806 Class.IngredientChicken
Ingredient Turkey 1.0479 2.01806 Class.IngredientTurkey
Fat 2 3.5243 2.01806 44.657 Class.Fat2
Fat 5 -4.5722 2.01806 Class.Fat5
Fat 8 1.0479 2.01806 Class.Fat8
Price 1.99 2.2861 2.01806 30.998 Class.Price1D99
Price 2.29 1.0479 2.01806 Class.Price2D29
Price 2.59 -3.3340 2.01806 Class.Price2D59
Calories 250 -0.6351 1.42699 7.006 Class.Calories250
Calories 350 0.6351 1.42699 Class.Calories350

The standard errors are not adjusted for the fact that the dependent variable was transformed and so
are generally liberal (too small).

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 41 of 58

The SAS System

The TRANSREG Procedure

ID=11

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 42 of 58

The SAS System

The TRANSREG Procedure

ID=12

Dependent Variable Monotone(Rank)


Rank

Class Level Information


Class Levels Values
Ingredient 3 Beef Chicken Turkey
Fat 3 258
Price 3 1.99 2.29 2.59
Calories 2 250 350

Number of Observations Read 22


Number of Observations Used 18
Sum of Weights Read 18
Sum of Weights Used 18

TRANSREG MORALS Algorithm Iteration History for Monotone(Rank)


Iteration Average Maximum Criterion
Number Change Change R-Square Change Note
1 0.11072 0.34700 0.97173
2 0.01542 0.04033 0.99795 0.02622
3 0.01032 0.01858 0.99850 0.00056
4 0.00865 0.01543 0.99877 0.00026
5 0.00763 0.01359 0.99896 0.00020
6 0.00702 0.01253 0.99912 0.00016
7 0.00648 0.01154 0.99925 0.00013
8 0.00595 0.01063 0.99936 0.00011
9 0.00549 0.00979 0.99946 0.00009
10 0.00507 0.00907 0.99954 0.00008
11 0.00469 0.00840 0.99960 0.00007
12 0.00436 0.00767 0.99966 0.00006
13 0.00405 0.00706 0.99971 0.00005
14 0.00375 0.00652 0.99975 0.00004

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 43 of 58

15 0.00348 0.00603 0.99979 0.00004


16 0.00323 0.00557 0.99982 0.00003
17 0.00299 0.00516 0.99984 0.00003
18 0.00278 0.00477 0.99986 0.00002
19 0.00257 0.00442 0.99988 0.00002
20 0.00239 0.00409 0.99990 0.00002
21 0.00221 0.00378 0.99991 0.00001
22 0.00205 0.00350 0.99993 0.00001
23 0.00190 0.00324 0.99994 0.00001
24 0.00176 0.00300 0.99995 0.00001
25 0.00163 0.00278 0.99995 0.00001
26 0.00151 0.00257 0.99996 0.00001
27 0.00140 0.00238 0.99997 0.00001
28 0.00130 0.00221 0.99997 0.00000
29 0.00120 0.00204 0.99997 0.00000
30 0.00112 0.00189 0.99998 0.00000 Not Converged

WARNING: Failed to converge, however criterion change is less than 0.0001.

The TRANSREG Procedure Hypothesis Tests for Monotone(Rank)


Rank

Root MSE 0.03830 R-Square 1.0000


Dependent Mean 11.66667 Adj R-Sq 1.0000
Coeff Var 0.32831

Utilities Table Based on the Usual Degrees of Freedom


Importance
(% Utility
Label Utility Standard Error Range) Variable
Intercept 11.6667 0.00903 Intercept
Ingredient Beef 1.2484 0.01277 11.864 Class.IngredientBeef
Ingredient Chicken 0.7542 0.01277 Class.IngredientChicken
Ingredient Turkey -2.0026 0.01277 Class.IngredientTurkey
Fat 2 6.8180 0.01277 49.396 Class.Fat2
Fat 5 -0.1002 0.01277 Class.Fat5
Fat 8 -6.7178 0.01277 Class.Fat8

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 44 of 58

Price 1.99 3.3363 0.01277 24.239 Class.Price1D99


Price 2.29 -0.0304 0.01277 Class.Price2D29
Price 2.59 -3.3059 0.01277 Class.Price2D59
Calories 250 1.9868 0.00903 14.501 Class.Calories250
Calories 350 -1.9868 0.00903 Class.Calories350

The standard errors are not adjusted for the fact that the dependent variable was transformed and so
are generally liberal (too small).

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 45 of 58

The SAS System

The TRANSREG Procedure

ID=12

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 46 of 58

The SAS System

Obs Rank Transformed Predicted Ingredient Fat Price Calories Rank Leverage
Ranks Ranks Residuals
1 9 14.2269 14.7737 Beef 2 2.59 250 -0.5468 0.00000
2 17 7.7448 7.7450 Beef 5 2.59 250 -0.0002 0.44444
3 2 20.7075 20.7075 Turkey 2 1.99 350 -0.0001 0.44444
4 1 24.7690 24.7692 Turkey 2 1.99 250 -0.0002 0.44444
5 22 -0.3801 -0.3800 Turkey 8 2.59 350 -0.0001 0.44444
6 7 14.2269 14.2266 Chicken 2 2.59 350 0.0003 0.44444
7 3 18.2886 18.2882 Chicken 2 2.59 250 0.0003 0.44444
8 20 3.6833 3.6833 Chicken 8 2.29 350 -0.0001 0.44444
9 19 3.6833 3.6833 Beef 5 2.59 350 -0.0001 0.44444
10 6 14.2269 14.2264 Turkey 5 2.29 250 0.0005 0.44444
11 12 10.7127 10.7125 Beef 5 1.99 350 0.0001 0.00000
12 5 18.2886 18.2887 Beef 2 2.29 250 -0.0002 0.44444
13 21 3.6814 3.6817 Turkey 8 2.59 250 -0.0002 0.44444
14 14 7.7448 7.7446 Beef 8 1.99 250 0.0002 0.44444
15 16 7.7448 6.6496 Turkey 5 2.59 350 1.0952 0.00000
16 10 14.2269 14.2271 Chicken 5 1.99 350 -0.0002 0.44444
17 15 7.7448 7.7450 Chicken 8 2.29 250 -0.0002 0.44444
18 11 14.2269 14.7746 Chicken 5 2.29 250 -0.5477 0.00000
19 8 14.2269 14.2271 Beef 2 2.29 350 -0.0002 0.44444
20 13 10.1648 10.1647 Turkey 5 2.29 350 0.0001 0.44444
21 4 18.2886 18.2887 Chicken 5 1.99 250 -0.0002 0.44444
22 18 3.6833 3.6829 Beef 8 1.99 350 0.0004 0.44444
23 . . . . . . . .
24 . . . . . . . .
25 4 18.7582 18.7580 Beef 2 2.59 250 0.0002 0.00000
26 12 11.6193 11.6197 Beef 5 2.59 250 -0.0004 0.44444
27 3 20.8359 20.8361 Turkey 2 1.99 350 -0.0001 0.44444
28 1 23.8119 23.8122 Turkey 2 1.99 250 -0.0002 0.44444
29 21 1.3894 1.3889 Turkey 8 2.59 350 0.0005 0.44444
30 8 14.5978 14.5985 Chicken 2 2.59 350 -0.0007 0.44444
31 6 17.5747 17.5746 Chicken 2 2.59 250 0.0001 0.44444

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 47 of 58

32 22 1.3894 1.3895 Chicken 8 2.29 350 -0.0001 0.44444


33 16 8.6436 8.6436 Beef 5 2.59 350 0.0001 0.44444
34 7 14.5978 14.5969 Turkey 5 2.29 250 0.0009 0.44444
35 9 14.5933 12.8012 Beef 5 1.99 350 1.7921 0.00000
36 2 20.8389 20.8386 Beef 2 2.29 250 0.0004 0.44444
37 19 4.3654 4.3650 Turkey 8 2.59 250 0.0004 0.44444
38 17 7.6257 7.6261 Beef 8 1.99 250 -0.0003 0.44444
39 15 9.5402 9.5402 Turkey 5 2.59 350 -0.0000 0.00000
40 11 11.6193 11.6178 Chicken 5 1.99 350 0.0015 0.44444
41 20 4.3654 4.3656 Chicken 8 2.29 250 -0.0002 0.44444
42 13 11.6193 12.5169 Chicken 5 2.29 250 -0.8976 0.00000
43 5 17.8630 17.8624 Beef 2 2.29 350 0.0005 0.44444
44 14 11.6193 11.6208 Turkey 5 2.29 350 -0.0015 0.44444
45 10 14.5933 14.5939 Chicken 5 1.99 250 -0.0006 0.44444
46 18 4.6498 4.6499 Beef 8 1.99 350 -0.0002 0.44444
47 . . . . . . . .
48 . . . . . . . .
49 7 16.5270 17.0975 Beef 2 2.59 250 -0.5705 0.00000
50 15 9.2704 9.2715 Beef 5 2.59 250 -0.0011 0.44444
51 2 20.8769 20.8751 Turkey 2 1.99 350 0.0018 0.44444
52 1 23.8302 23.8282 Turkey 2 1.99 250 0.0020 0.44444
53 22 0.4453 0.4449 Turkey 8 2.59 350 0.0003 0.44444
54 8 16.5270 16.5319 Chicken 2 2.59 350 -0.0049 0.44444
55 3 19.4966 19.4850 Chicken 2 2.59 250 0.0116 0.44444
56 20 3.3873 3.3768 Chicken 8 2.29 350 0.0105 0.44444
57 17 6.3232 6.3184 Beef 5 2.59 350 0.0048 0.44444
58 9 13.5571 13.5476 Turkey 5 2.29 250 0.0096 0.44444
59 12 11.1837 11.1832 Beef 5 1.99 350 0.0005 0.00000
60 4 19.4966 19.5078 Beef 2 2.29 250 -0.0112 0.44444
61 21 3.3873 3.3981 Turkey 8 2.59 250 -0.0107 0.44444
62 16 6.4005 6.3971 Beef 8 1.99 250 0.0034 0.44444
63 14 9.2704 8.1842 Turkey 5 2.59 350 1.0862 0.00000
64 11 13.5571 13.5707 Chicken 5 1.99 350 -0.0136 0.44444
65 18 6.3232 6.3299 Chicken 8 2.29 250 -0.0067 0.44444
66 10 13.5571 14.0692 Chicken 5 2.29 250 -0.5121 0.00000

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 48 of 58

5 16.5554 16.5547 Beef 2 2.29 350 0.0007 0.44444


68 13 10.5916 10.5945 Turkey 5 2.29 350 -0.0029 0.44444
69 6 16.5270 16.5238 Chicken 5 1.99 250 0.0031 0.44444
70 19 3.4472 3.4440 Beef 8 1.99 350 0.0033 0.44444
71 . . . . . . . .
72 . . . . . . . .
73 3 17.9491 17.1494 Beef 2 2.59 250 0.7998 0.00000
74 10 13.8587 13.8587 Beef 5 2.59 250 -0.0000 0.44444
75 8 14.4727 14.4724 Turkey 2 1.99 350 0.0002 0.44444
76 1 24.6038 24.6043 Turkey 2 1.99 250 -0.0005 0.44444
77 21 1.7684 1.7665 Turkey 8 2.59 350 0.0020 0.44444
78 19 4.3415 4.3422 Chicken 2 2.59 350 -0.0007 0.44444
79 9 14.4727 14.4741 Chicken 2 2.59 250 -0.0014 0.44444
80 22 1.7684 1.7696 Chicken 8 2.29 350 -0.0011 0.44444
81 20 3.7266 3.7269 Beef 5 2.59 350 -0.0002 0.44444
82 4 17.9491 17.9483 Turkey 5 2.29 250 0.0008 0.44444
83 14 10.4025 10.4762 Beef 5 1.99 350 -0.0737 0.00000
84 2 20.5346 20.5333 Beef 2 2.29 250 0.0013 0.44444
85 12 11.8988 11.8983 Turkey 8 2.59 250 0.0004 0.44444
86 6 17.9367 17.9421 Beef 8 1.99 250 -0.0054 0.44444
87 18 4.4321 4.4325 Turkey 5 2.59 350 -0.0004 0.00000
88 17 7.7993 7.8009 Chicken 5 1.99 350 -0.0016 0.44444
89 11 11.9023 11.9014 Chicken 8 2.29 250 0.0008 0.44444
90 7 14.5677 14.5674 Chicken 5 2.29 250 0.0003 0.00000
91 13 10.4025 10.4015 Beef 2 2.29 350 0.0011 0.44444
92 16 7.8135 7.8164 Turkey 5 2.29 350 -0.0029 0.44444
93 5 17.9367 17.9328 Chicken 5 1.99 250 0.0039 0.44444
94 15 7.8135 7.8102 Beef 8 1.99 350 0.0033 0.44444
95 . . . . . . . .
96 . . . . . . . .
97 16 7.6053 7.6047 Beef 2 2.59 250 0.0005 0.00000
98 18 5.5268 5.5287 Beef 5 2.59 250 -0.0018 0.44444
99 6 17.8151 17.8174 Turkey 2 1.99 350 -0.0023 0.44444
100 1 20.3784 20.3783 Turkey 2 1.99 250 0.0001 0.44444
101 22 -0.6917 -0.6918 Turkey 8 2.59 350 0.0001 0.44444

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 49 of 58

17 5.5268 5.5253 Chicken 2 2.59 350 0.0015 0.44444


103 15 8.0863 8.0863 Chicken 2 2.59 250 -0.0000 0.44444
104 13 9.0402 9.0402 Chicken 8 2.29 350 -0.0000 0.44444
105 19 2.9674 2.9677 Beef 5 2.59 350 -0.0003 0.44444
106 11 11.5990 11.5968 Turkey 5 2.29 250 0.0022 0.44444
107 4 17.8151 17.3324 Beef 5 1.99 350 0.4826 0.00000
108 7 15.2646 15.2641 Beef 2 2.29 250 0.0006 0.44444
109 20 1.8697 1.8691 Turkey 8 2.59 250 0.0005 0.44444
110 3 17.8260 17.8251 Beef 8 1.99 250 0.0010 0.44444
111 21 1.3758 1.3765 Turkey 5 2.59 350 -0.0007 0.00000
112 5 17.8151 17.8140 Chicken 5 1.99 350 0.0011 0.44444
113 12 11.5990 11.6011 Chicken 8 2.29 250 -0.0021 0.44444
114 10 12.7032 13.6695 Chicken 5 2.29 250 -0.9663 0.00000
115 9 12.7032 12.7031 Beef 2 2.29 350 0.0001 0.44444
116 14 9.0351 9.0358 Turkey 5 2.29 350 -0.0007 0.44444
117 2 20.3745 20.3749 Chicken 5 1.99 250 -0.0004 0.44444
118 8 15.2646 15.2641 Beef 8 1.99 350 0.0005 0.44444
119 . . . . . . . .
120 . . . . . . . .
121 13 10.9976 10.6680 Beef 2 2.59 250 0.3296 0.00000
122 17 7.3384 7.3385 Beef 5 2.59 250 -0.0001 0.44444
123 5 17.3263 17.3265 Turkey 2 1.99 350 -0.0002 0.44444
124 1 20.9855 20.9855 Turkey 2 1.99 250 0.0001 0.44444
125 22 -2.6483 -2.6483 Turkey 8 2.59 350 0.0001 0.44444
126 16 7.3389 7.3389 Chicken 2 2.59 350 -0.0000 0.44444
127 11 10.9979 10.9979 Chicken 2 2.59 250 0.0000 0.44444
128 15 7.3389 7.3388 Chicken 8 2.29 350 0.0001 0.44444
129 19 3.6795 3.6796 Beef 5 2.59 350 -0.0001 0.44444
130 12 10.9976 10.9975 Turkey 5 2.29 250 0.0001 0.44444
131 7 16.9961 16.9962 Beef 5 1.99 350 -0.0000 0.00000
132 4 17.3263 17.3261 Beef 2 2.29 250 0.0002 0.44444
133 20 1.0107 1.0106 Turkey 8 2.59 250 0.0001 0.44444
134 6 17.3263 17.3264 Beef 8 1.99 250 -0.0001 0.44444
135 21 0.6804 0.6804 Turkey 5 2.59 350 -0.0000 0.00000
136 3 17.3263 17.3261 Chicken 5 1.99 350 0.0002 0.44444

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 50 of 58

14 10.9976 10.9978 Chicken 8 2.29 250 -0.0002 0.44444


138 9 13.6675 14.3265 Chicken 5 2.29 250 -0.6590 0.00000
139 10 13.6671 13.6672 Beef 2 2.29 350 -0.0000 0.44444
140 18 7.3384 7.3385 Turkey 5 2.29 350 -0.0001 0.44444
141 2 20.9850 20.9850 Chicken 5 1.99 250 -0.0001 0.44444
142 8 13.6675 13.6674 Beef 8 1.99 350 0.0001 0.44444
143 . . . . . . . .
144 . . . . . . . .
145 4 19.9500 19.9501 Beef 2 2.59 250 -0.0001 0.00000
146 12 11.7209 11.7207 Beef 5 2.59 250 0.0002 0.44444
147 5 19.2787 19.2786 Turkey 2 1.99 350 0.0001 0.44444
148 2 21.2941 21.2940 Turkey 2 1.99 250 0.0001 0.44444
149 22 0.3006 0.3006 Turkey 8 2.59 350 0.0000 0.44444
150 7 17.2630 17.2631 Chicken 2 2.59 350 -0.0001 0.44444
151 6 19.2784 19.2784 Chicken 2 2.59 250 -0.0001 0.44444
152 20 3.6597 3.6596 Chicken 8 2.29 350 0.0001 0.44444
153 15 9.7052 9.7054 Beef 5 2.59 350 -0.0001 0.44444
154 13 11.7209 11.7212 Turkey 5 2.29 250 -0.0002 0.44444
155 10 13.0643 13.0644 Beef 5 1.99 350 -0.0001 0.00000
156 1 21.9657 21.9657 Beef 2 2.29 250 0.0000 0.44444
157 21 2.3159 2.3159 Turkey 8 2.59 250 0.0000 0.44444
158 17 7.6901 7.6901 Beef 8 1.99 250 -0.0001 0.44444
159 16 7.6902 7.6902 Turkey 5 2.59 350 0.0000 0.00000
160 11 12.3927 12.3927 Chicken 5 1.99 350 0.0000 0.44444
161 18 5.6750 5.6750 Chicken 8 2.29 250 0.0001 0.44444
162 9 13.0647 13.0646 Chicken 5 2.29 250 0.0001 0.00000
163 3 19.9503 19.9503 Beef 2 2.29 350 0.0000 0.44444
164 14 9.7059 9.7058 Turkey 5 2.29 350 0.0001 0.44444
165 8 14.4081 14.4081 Chicken 5 1.99 250 0.0000 0.44444
166 19 5.6747 5.6748 Beef 8 1.99 350 -0.0001 0.44444
167 . . . . . . . .
168 . . . . . . . .
169 17 5.4642 6.8627 Beef 2 2.59 250 -1.3985 0.00000
170 18 4.5482 4.5483 Beef 5 2.59 250 -0.0000 0.44444
171 3 19.8236 19.8237 Turkey 2 1.99 350 -0.0000 0.44444

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 51 of 58

1 22.1283 22.1283 Turkey 2 1.99 250 -0.0000 0.44444


173 22 0.2346 0.2346 Turkey 8 2.59 350 0.0000 0.44444
174 16 5.4642 5.4642 Chicken 2 2.59 350 0.0000 0.44444
175 15 7.7689 7.7688 Chicken 2 2.59 250 0.0000 0.44444
176 14 7.9758 7.9758 Chicken 8 2.29 350 0.0001 0.44444
177 21 2.2436 2.2436 Beef 5 2.59 350 -0.0000 0.44444
178 10 12.5847 12.5848 Turkey 5 2.29 250 -0.0001 0.44444
179 6 16.6034 16.9085 Beef 5 1.99 350 -0.3051 0.00000
180 7 14.2986 14.2985 Beef 2 2.29 250 0.0001 0.44444
181 19 2.5392 2.5392 Turkey 8 2.59 250 0.0000 0.44444
182 5 16.6034 16.6033 Beef 8 1.99 250 0.0001 0.44444
183 20 2.5392 2.8444 Turkey 5 2.59 350 -0.3051 0.00000
184 4 17.8147 17.8146 Chicken 5 1.99 350 0.0000 0.44444
185 13 10.2802 10.2804 Chicken 8 2.29 250 -0.0001 0.44444
186 9 12.8901 12.8902 Chicken 5 2.29 250 -0.0000 0.00000
187 11 11.9938 11.9939 Beef 2 2.29 350 -0.0001 0.44444
188 12 10.2802 10.2801 Turkey 5 2.29 350 0.0001 0.44444
189 2 20.1193 20.1192 Chicken 5 1.99 250 0.0000 0.44444
190 8 14.2986 14.2987 Beef 8 1.99 350 -0.0001 0.44444
191 . . . . . . . .
192 . . . . . . . .
193 17 5.4587 5.4587 Beef 2 2.59 250 -0.0000 0.00000
194 20 3.4994 3.4998 Beef 5 2.59 250 -0.0004 0.44444
195 3 20.4165 20.4168 Turkey 2 1.99 350 -0.0003 0.44444
196 1 22.3727 22.3727 Turkey 2 1.99 250 0.0000 0.44444
197 22 1.5432 1.5432 Turkey 8 2.59 350 0.0000 0.44444
198 16 5.4647 5.4647 Chicken 2 2.59 350 -0.0000 0.44444
199 15 7.4205 7.4206 Chicken 2 2.59 250 -0.0001 0.44444
200 14 8.9388 8.9388 Chicken 8 2.29 350 -0.0001 0.44444
201 21 1.5440 1.5439 Beef 5 2.59 350 0.0001 0.44444
202 10 12.8522 12.8525 Turkey 5 2.29 250 -0.0003 0.44444
203 5 16.4985 16.4984 Beef 5 1.99 350 0.0001 0.00000
204 8 12.8522 12.8519 Beef 2 2.29 250 0.0003 0.44444
205 18 3.4994 3.4990 Turkey 8 2.59 250 0.0004 0.44444
206 6 16.4940 16.4941 Beef 8 1.99 250 -0.0001 0.44444

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 52 of 58

19 3.4994 3.5033 Turkey 5 2.59 350 -0.0039 0.00000


208 4 18.4604 18.4603 Chicken 5 1.99 350 0.0001 0.44444
209 13 10.8946 10.8947 Chicken 8 2.29 250 -0.0002 0.44444
210 9 12.8522 12.8549 Chicken 5 2.29 250 -0.0027 0.00000
211 12 10.8961 10.8960 Beef 2 2.29 350 0.0001 0.44444
212 11 10.8967 10.8966 Turkey 5 2.29 350 0.0001 0.44444
213 2 20.4165 20.4162 Chicken 5 1.99 250 0.0003 0.44444
214 7 14.5382 14.5382 Beef 8 1.99 350 -0.0000 0.44444
215 . . . . . . . .
216 . . . . . . . .
217 14 8.7588 8.7584 Beef 2 2.59 250 0.0004 0.00000
218 17 6.4326 6.4336 Beef 5 2.59 250 -0.0010 0.44444
219 2 19.5071 19.5070 Turkey 2 1.99 350 0.0001 0.44444
220 1 22.4292 22.4286 Turkey 2 1.99 250 0.0007 0.44444
221 22 0.6015 0.6011 Turkey 8 2.59 350 0.0003 0.44444
222 20 3.5138 3.5163 Chicken 2 2.59 350 -0.0025 0.44444
223 15 6.4384 6.4379 Chicken 2 2.59 250 0.0005 0.44444
224 16 6.4326 6.4320 Chicken 8 2.29 350 0.0006 0.44444
225 19 3.5138 3.5120 Beef 5 2.59 350 0.0018 0.44444
226 10 13.3951 13.3966 Turkey 5 2.29 250 -0.0015 0.44444
227 4 18.3772 17.7763 Beef 5 1.99 350 0.6009 0.00000
228 6 16.3157 16.3156 Beef 2 2.29 250 0.0001 0.44444
229 18 3.5236 3.5227 Turkey 8 2.59 250 0.0009 0.44444
230 3 18.3815 18.3812 Beef 8 1.99 250 0.0003 0.44444
231 21 2.9176 2.9178 Turkey 5 2.59 350 -0.0002 0.00000
232 7 15.4572 15.4558 Chicken 5 1.99 350 0.0014 0.44444
233 13 9.3538 9.3536 Chicken 8 2.29 250 0.0002 0.44444
234 11 11.6699 11.6703 Chicken 5 2.29 250 -0.0004 0.00000
235 9 13.3951 13.3940 Beef 2 2.29 350 0.0011 0.44444
236 12 10.4745 10.4750 Turkey 5 2.29 350 -0.0005 0.44444
237 5 18.3772 18.3774 Chicken 5 1.99 250 -0.0002 0.44444
238 8 15.4572 15.4596 Beef 8 1.99 350 -0.0024 0.44444
239 . . . . . . . .
240 . . . . . . . .
241 15 9.4999 9.6261 Beef 2 2.59 250 -0.1262 0.00000

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 53 of 58

22 -9.3618 1.5297 Beef 5 2.59 250 -10.8915 0.44444


243 11 16.9292 19.6600 Turkey 2 1.99 350 -2.7309 0.44444
244 9 16.9292 18.3898 Turkey 2 1.99 250 -1.4606 0.44444
245 21 9.4999 11.5636 Turkey 8 2.59 350 -2.0637 0.44444
246 5 16.9292 14.0400 Chicken 2 2.59 350 2.8892 0.44444
247 14 9.4999 12.7697 Chicken 2 2.59 250 -3.2698 0.44444
248 17 9.4999 15.9454 Chicken 8 2.29 350 -6.4455 0.44444
249 13 9.4999 2.7999 Beef 5 2.59 350 6.7000 0.44444
250 16 9.4999 9.0551 Turkey 5 2.29 250 0.4448 0.44444
251 6 16.9292 8.4200 Beef 5 1.99 350 8.5092 0.00000
252 10 16.9292 14.0080 Beef 2 2.29 250 2.9212 0.44444
253 7 16.9292 10.2933 Turkey 8 2.59 250 6.6358 0.44444
254 19 9.4999 12.7697 Beef 8 1.99 250 -3.2698 0.44444
255 1 16.9292 5.9436 Turkey 5 2.59 350 10.9856 0.00000
256 20 9.4999 11.5636 Chicken 5 1.99 350 -2.0637 0.44444
257 3 16.9292 14.6751 Chicken 8 2.29 250 2.2540 0.44444
258 12 9.4999 9.0551 Chicken 5 2.29 250 0.4448 0.00000
259 8 16.9292 15.2782 Beef 2 2.29 350 1.6509 0.44444
260 18 9.4999 10.3254 Turkey 5 2.29 350 -0.8255 0.44444
261 4 16.9292 10.2933 Chicken 5 1.99 250 6.6358 0.44444
262 2 16.9292 14.0400 Beef 8 1.99 350 2.8892 0.44444
263 . . . . . . . .
264 . . . . . . . .
265 7 17.6746 18.4139 Beef 2 2.59 250 -0.7393 0.00000
266 12 11.5059 11.4957 Beef 5 2.59 250 0.0102 0.44444
267 3 17.8861 17.8316 Turkey 2 1.99 350 0.0545 0.44444
268 2 21.7515 21.8051 Turkey 2 1.99 250 -0.0536 0.44444
269 22 -2.3369 -2.3464 Turkey 8 2.59 350 0.0094 0.44444
270 10 13.9583 13.9463 Chicken 2 2.59 350 0.0120 0.44444
271 4 17.8861 17.9198 Chicken 2 2.59 250 -0.0337 0.44444
272 20 3.6854 3.6859 Chicken 8 2.29 350 -0.0005 0.44444
273 18 7.5129 7.5222 Beef 5 2.59 350 -0.0093 0.44444
274 14 11.5059 11.5202 Turkey 5 2.29 250 -0.0143 0.44444
275 9 14.1475 14.1643 Beef 5 1.99 350 -0.0169 0.00000
276 1 21.7515 21.6894 Beef 2 2.29 250 0.0621 0.44444

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 54 of 58

21 1.6385 1.6272 Turkey 8 2.59 250 0.0114 0.44444


278 13 11.5059 11.5202 Beef 8 1.99 250 -0.0143 0.44444
279 19 4.2713 4.2712 Turkey 5 2.59 350 0.0000 0.00000
280 11 13.6602 13.6702 Chicken 5 1.99 350 -0.0100 0.44444
281 15 7.6609 7.6595 Chicken 8 2.29 250 0.0014 0.44444
282 8 14.2691 14.2771 Chicken 5 2.29 250 -0.0080 0.00000
283 6 17.6746 17.7159 Beef 2 2.29 350 -0.0413 0.44444
284 17 7.5393 7.5467 Turkey 5 2.29 350 -0.0074 0.44444
285 5 17.6746 17.6438 Chicken 5 1.99 250 0.0308 0.44444
286 16 7.5393 7.5467 Beef 8 1.99 350 -0.0074 0.44444
287 . . . . . . . .
288 . . . . . . . .

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 55 of 58

The SAS System

ID R Square
1 1.00
2 1.00
3 1.00
4 1.00
5 1.00
6 1.00
7 1.00
8 1.00
9 1.00
10 1.00
11 0.49
12 1.00

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 56 of 58

The SAS System

Obs Ingredient Fat Price Calories


1 12.2613 49.0456 24.5230 14.1700
2 8.4885 62.3981 16.9676 12.1458
3 9.2643 60.3991 18.8775 11.4591
4 12.8950 22.7189 25.7425 38.6436
5 8.9562 17.9081 62.0699 11.0659
6 12.3468 24.6941 49.3888 13.5704
7 8.7582 67.8835 14.5991 8.7592
8 3.9743 21.5977 64.3199 10.1080
9 8.6081 17.1955 65.6147 8.5817
10 9.6095 19.2212 59.0706 12.0987
11 17.3389 44.6568 30.9979 7.0063
12 11.8637 49.3964 24.2392 14.5007

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 57 of 58

The SAS System

The MEANS Procedure

Variable Mean
Ingredient 10.3637367
Fat 38.0929242
Price 38.0342160
Calories 13.5091231

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
SAS Output Page 58 of 58

The SAS System

Obs ID Correlation
4 1 0.91287
8 2 1.00000
12 3 1.00000
16 4 1.00000
20 5 1.00000
24 6 1.00000
28 7 1.00000
32 8 1.00000
36 9 1.00000
40 10 1.00000
44 11 -0.81650
48 12 1.00000

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/... 5/29/2014
MSAN-624 V. Dimitrova P. 13

EXAMPLE 6: Conjoint analysis, multiple respondents (soft drinks)

We conduct the conjoint analysis similar to Example 5. The new components are a

segmentation analysis and a market share analysis. The following program

simulates market shares using the maximum utility (“first choice”) method. The

other two market share models are the Bradley-Terry-Luce (BTL) and logit

methods. Note that the maximum utility method is scale free, whereas the BTL

and logit methods are not invariant under linear transformations of the predicted

utility.

PROC RANK DATA=T OUT=r TIES=mean; VAR PRank; RANKS newRank; BY ID; run;

DATA r; SET r; rename _NAME_=Profile;

PROC FORMAT; VALUE $Profile 'Row1' = 'Coca-Cola, Diet, $3.50' 'Row6' = 'Coca-
Cola, Regular, $3.50' 'Row11' = 'Pepsi, Diet, $3.50' 'Row16' = 'Pepsi,
Regular, $3.50'; run;

DATA r; SET r; FORMAT Profile $Profile.;

PROC FREQ DATA=r(where=(newRank=20)) ORDER=FREQ; TABLES Profile / OUTPCT;


run;

PROC PRINT; run;

13
MSAN-624 V. Dimitrova P. 14

Next we conduct segmentation analysis.

DATA SA; SET T (where=(_TYPE_='M COEFFI'));

PROC STANDARD DATA=SA OUT=C1 MEAN=0 STD=1;

VAR BrandCoke BrandPepsi Price_3_50 Price_3_75 Price_4_00 Price_4_25

Price_4_50 ContentDiet ContentRegular; run;

PROC FASTCLUS DATA=C1 OUT=C2 MAXC=50 RADIUS=.5 DRIFT NOPRINT;

VAR BrandCoke BrandPepsi Price_3_50 Price_3_75 Price_4_00 Price_4_25

Price_4_50 ContentDiet ContentRegular; run;

PROC MEANS DATA=C2 MEAN NOPRINT;

VAR BrandCoke BrandPepsi Price_3_50 Price_3_75 Price_4_00 Price_4_25

Price_4_50 ContentDiet ContentRegular; CLASS cluster;

OUTPUT OUT=C3 MEAN (BrandCoke BrandPepsi Price_3_50 Price_3_75 Price_4_00

Price_4_25 Price_4_50 ContentDiet ContentRegular) = BrandCoke BrandPepsi

Price_3_50 Price_3_75 Price_4_00 Price_4_25 Price_4_50 ContentDiet

ContentRegular;

run;

PROC CLUSTER DATA=C3 (where=(_TYPE_=1)) METHOD=Ward PLOTS=NONE RSQUARE PSEUDO

PRINT=10 OUTTREE=sasuser.tree;

VAR BrandCoke BrandPepsi Price_3_50 Price_3_75 Price_4_00 Price_4_25

Price_4_50 ContentDiet ContentRegular; run;

14
MSAN-624 V. Dimitrova P. 15

For large samples (n > 100), we conduct initial cluster analysis with PROC FASTCLUS.

PROC FASTCLUS DATA=C1 OUT=C2 MAXC=50 RADIUS=.5 DRIFT NOPRINT;

VAR BrandCoke BrandPepsi Price_3_50 Price_3_75 Price_4_00 Price_4_25

Price_4_50 ContentDiet ContentRegular; run;

MAXC=50 specify the maximum number of clusters to create. RADIUS=.5 sets the

minimum distance when selecting new seeds, and DRIFT allows cluster seeds to

drift.

We then average the clustering variables for each of the newly formed FASTCLUS

clusters.

PROC MEANS DATA=C2 MEAN NOPRINT;

VAR BrandCoke BrandPepsi Price_3_50 Price_3_75 Price_4_00 Price_4_25

Price_4_50 ContentDiet ContentRegular; CLASS cluster;

OUTPUT OUT=C3 MEAN (BrandCoke BrandPepsi Price_3_50 Price_3_75 Price_4_00

Price_4_25 Price_4_50 ContentDiet ContentRegular) = BrandCoke BrandPepsi

Price_3_50 Price_3_75 Price_4_00 Price_4_25 Price_4_50 ContentDiet

ContentRegular;

run;

15
MSAN-624 V. Dimitrova P. 16

Then we use a hierarchical clustering procedure in order to determine the best

number of clusters. Note that we cluster the newly formed FASTCLUS clusters. I

recommend METHOD=EML (Equal-Variances Maximum Likelihood) and METHOD=WARD

(Ward’s Minimum Variance method).

PROC CLUSTER DATA=C3 (where=(_TYPE_=1)) METHOD=EML PLOTS=NONE RSQUARE PSEUDO

PRINT=10 OUTTREE=sasuser.tree;

VAR BrandCoke BrandPepsi Price_3_50 Price_3_75 Price_4_00 Price_4_25

Price_4_50 ContentDiet ContentRegular; run;

RSQUARE outputs the Semi-Partial R2 and the R2 criteria. For each level of the

hierarchy, the Semi-Partial R2 shows the proportional reduction in variance due to

joining the two clusters (a small reduction in variance means that the two clusters

are similar) and the R2 statistic shows the remaining variance that has not been

accounted for.

PSEUDO prints the Pseudo-F and the Pseudo-T2 statistics. The Pseudo-F statistic is

not distributed as an F random variable. It equals the ratio of between-cluster

variation to within-cluster variation. We look for large values suggesting that the

mean vectors of all the clusters are different. The Pseudo-T2 statistic measures

the within-cluster variance of the newly formed cluster relative to the sum of

within-cluster variance of the two component clusters. Large values of the

Pseudo-T2 statistic mean that the two clusters should not be combined because

their mean vectors are different. We look for small values.

16
MSAN-624 V. Dimitrova P. 17

Both the EML and the Ward methods indicate to 3 clusters.

17
MSAN-624 V. Dimitrova P. 18

Let’s compare this output to an output from the EML and the Ward methods if we

didn’t use FASTCLUS but directly clustered the initial 132 cases.

18
MSAN-624 V. Dimitrova P. 19

The conclusions are very similar, so for this dataset we do not need to use

FASTCLUS. However, FASTCLUS is recommended for larger samples, especially

samples exceeding 500 cases.

Next we plot the cluster means and interpret the clusters. We also cross the 3-

cluster solution with the 6-cluster solution suggested by the Pseudo F statistic.

PROC TREE DATA=sasuser.tree NCL=3 OUT=CL1;

COPY BrandCoke BrandPepsi Price_3_50 Price_3_75 Price_4_00 Price_4_25

Price_4_50 ContentDiet ContentRegular;

run;

PROC MEANS MEAN DATA=CL1; CLASS cluster;

VAR BrandCoke BrandPepsi Price_3_50 Price_3_75 Price_4_00 Price_4_25

Price_4_50 ContentDiet ContentRegular;

OUTPUT OUT=M; run;

DATA M1; SET M(where=(_TYPE_=1));

PROC TRANSPOSE DATA=M1(where=(_STAT_='MEAN')) OUT=M2; run;

PROC SORT DATA=M2; BY _NAME_; run;

PROC FORMAT; VALUE $_NAME_ 'BrandCoke' = 'Coca Cola' 'BrandPepsi' = 'Pepsi'

'Price_3_50' = '$3.50' 'Price_3_75' = '$3.75' 'Price_4_00' = '$4.00'

'Price_4_25' = '$4.25' 'Price_4_50' = '$4.50' 'ContentDiet' = 'Diet'

'ContentRegular' = 'Regular';

19
MSAN-624 V. Dimitrova P. 20

DATA M2; SET M2; FORMAT _NAME_ $_NAME_.; label _NAME_ = 'Attribute Levels';

Symbol1 interpol=join width=2 value=triangle line=1 c=darkred;

Symbol2 interpol=join width=2 value=circle line=6 c=indigo;

Symbol3 interpol=join width=2 value=square line=2 c=Green;

Legend1 label=none value=(Tick=1 "Cluster 1" Tick=2 "Cluster 2" Tick=3

"Cluster 3");

Axis1 label=none;

ods graphics on;

PROC GPlot data=m2(where=(_LABEL_ NE ' '));

PLOT (Col1-Col3)*_NAME_ / overlay legend=Legend1;

run;

ods graphics off;

PROC TREE DATA=sasuser.tree NCL=6 OUT=CL2;

COPY BrandCoke BrandPepsi Price_3_50 Price_3_75 Price_4_00 Price_4_25

Price_4_50 ContentDiet ContentRegular;

run;

DATA new; MERGE CL1 (rename=(cluster=CL3)) CL2 (rename=(cluster=CL6)); run;

PROC FREQ DATA=new; TABLES CL3 * CL6; run;

20
MSAN-624 V. Dimitrova P. 21

EXAMPLE 7: Cluster analysis (TOYS)

The new element is the correspondence analysis. First we label all clusters and

attribute levels. Then we conduct the correspondence analysis using PROC CORRESP.

PROC FORMAT; VALUE cluster 1='Down-To-Earth' 2='Enthusiasts' 3='Creatives'

4='Intellectuals' 5='Apathetic'; run;

DATA CL5; SET CL5; FORMAT cluster cluster.;

label X1='Male 8' X2='Female 8' X3='Male 9' X4='Female 9' X5='Male 10'

X6='Female 10' X7='Male 11' X8='Female 11' X9='Male 12' X10='Female 12';

PROC MEANS DATA=CL5 MEAN SUM NOPRINT; VAR challenging easyuse imagination

educational Social X1-X10; CLASS cluster;

OUTPUT OUT=CRP MEAN (challenging easyuse imagination educational

Social)=challenging easyuse imagination educational Social

SUM (X1-X10) = X1-X10;

run;

ODS GRAPHICS ON;

PROC CORRESP DATA=CRP (where=(_TYPE_=1)); VAR X1-X10; ID cluster; run;

ODS GRAPHICS OFF;

21
MSAN-624 V. Dimitrova P. 22

In correspondence analysis, total inertia refers to the amount of variance in the

correspondence data table explained by the total model. Each dimension's inertia

refers to the amount of that total variance explained by the dimension. In the

example, our model accounts for 4.8% of the variance in the data table. Of that,

Dimension 1 accounts for 50% (approximately 2.4% of the total variance), and the

first two dimensions account for 83.30% of the explained variance. The chi-square

test evaluates the hypothesis that the total inertia value is / is not different than

zero. In our example, we have χ (36) = 23.27 < 51.71 suggesting that our model

accounts for some of the variance. The coordinates for all categories (points) are

displayed. The Summary statistics table for the row / column points displays:

1) Quality: The quality of the category’s representation in the two-dimensional

display equaling the sum of its two squared cosines. These are the cosines

of the angles between each axis and a vector from the origin to the point.

2) Mass: The category’s proportion of the entire sample size.

3) Inertia: The percentage of inertia that the category accounts for.

EXAMPLE 8: Cluster analysis (Branch)

22
SAS Output Page 1 of 2

The SAS System

The CLUSTER Procedure


Equal Variance Maximum Likelihood Method

Eigenvalues of the Covariance Matrix


Eigenvalue Difference Proportion Cumulative
1 1.91103110 0.73772584 0.3822 0.3822
2 1.17330526 0.25422101 0.2347 0.6169
3 0.91908425 0.34182441 0.1838 0.8007
4 0.57725984 0.15794029 0.1155 0.9161
5 0.41931955 0.0839 1.0000

Root-Mean-Square Total-Sample Standard Deviation 1

Root-Mean-Square Distance Between Observations 3.162278

Cluster History
Number Pseudo Pseudo Log
of Clusters Semipartial R- F t- Likelihood Log
Clusters Joined Freq R-Square Square Statistic Squared Ratio Likelihood Tie
10 CL11 CL144 262 0.0178 .543 69.7 23.4 28.35 -9796
9 CL22 CL12 120 0.0223 .521 71.9 20.1 23.45 -9820
8 CL9 CL15 139 0.0233 .497 74.9 18.5 17.05 -9837
7 CL24 CL36 35 0.0167 .481 81.9 25.9 40.26 -9877
6 CL108 CL62 33 0.0179 .463 91.7 57.1 52.35 -9929
5 CL10 CL16 324 0.0778 .385 83.4 92.2 47.70 -9977
4 CL8 CL53 146 0.0232 .362 101 16.7 43.30 -10020
3 CL7 CL4 181 0.0608 .301 115 41.7 66.97 -10087
2 CL5 CL3 505 0.2105 .091 53.3 158 49.26 -10137
1 CL2 CL6 538 0.0905 .000 . 53.3 7.044 -10144

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/_... 6/3/2014
SAS Output Page 2 of 2

The SAS System

The CLUSTER Procedure


Ward's Minimum Variance Cluster Analysis

Eigenvalues of the Covariance Matrix


Eigenvalue Difference Proportion Cumulative
1 1.91103110 0.73772584 0.3822 0.3822
2 1.17330526 0.25422101 0.2347 0.6169
3 0.91908425 0.34182441 0.1838 0.8007
4 0.57725984 0.15794029 0.1155 0.9161
5 0.41931955 0.0839 1.0000

Root-Mean-Square Total-Sample Standard Deviation 1

Root-Mean-Square Distance Between Observations 3.162278

Cluster History
Number
of Semipartial Pseudo F Pseudo
Clusters Clusters Joined Freq R-Square R-Square Statistic t-Squared Tie
10 CL27 CL33 90 0.0197 .610 91.7 51.7
9 CL14 CL29 46 0.0212 .589 94.6 22.4
8 CL10 CL37 109 0.0262 .562 97.3 44.7
7 CL13 CL19 93 0.0304 .532 101 32.4
6 CL17 CL12 93 0.0411 .491 103 42.4
5 CL9 CL7 139 0.0453 .446 107 34.8
4 CL16 CL8 177 0.0548 .391 114 69.2
3 CL4 CL11 306 0.0634 .327 130 75.2
2 CL6 CL5 232 0.1076 .220 151 70.2
1 CL2 CL3 538 0.2199 .000 . 151

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/_... 6/3/2014
SAS Output Page 1 of 10

The CLUSTER Procedure


Ward's Minimum Variance Cluster Analysis

Eigenvalues of the Covariance Matrix


Eigenvalue Difference Proportion Cumulative
1 0.84798092 0.35322043 0.4209 0.4209
2 0.49476049 0.13924235 0.2456 0.6664
3 0.35551814 0.15503813 0.1764 0.8429
4 0.20048001 0.08435091 0.0995 0.9424
5 0.11612910 0.0576 1.0000

Root-Mean-Square Total-Sample Standard Deviation 0.634802

Root-Mean-Square Distance Between Observations 2.007421

Cluster History
Number
of Semipartial Pseudo F Pseudo
Clusters Clusters Joined Freq R-Square R-Square Statistic t-Squared Tie
10 CL14 OB41 78 0.0318 .738 334 108
9 CL17 CL11 104 0.0326 .705 319 36.5
8 CL16 CL15 97 0.0393 .666 304 83.1
7 CL9 CL12 138 0.0428 .623 295 38.3
6 CL13 CL35 662 0.0452 .578 293 1312
5 CL10 CL20 104 0.0595 .519 288 80.6
4 CL6 CL8 759 0.0799 .439 279 398
3 CL4 CL5 863 0.1228 .316 248 288
2 CL3 OB23 938 0.1436 .172 224 275
1 CL2 CL7 1076 0.1723 .000 . 224

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/_... 6/3/2014
SAS Output Page 2 of 10

The TREE Procedure


Ward's Minimum Variance Cluster Analysis

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/_... 6/3/2014
SAS Output Page 3 of 10

The TREE Procedure


Ward's Minimum Variance Cluster Analysis

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/_... 6/3/2014
SAS Output Page 4 of 10

The FREQ Procedure

Frequency Table of CL3 by CL5


Percent
CL5
Row Pct
Col Pct CL3 1 2 3 4 5 Total
1 24 0 0 0 0 24
47.06 0.00 0.00 0.00 0.00 47.06
100.00 0.00 0.00 0.00 0.00
100.00 0.00 0.00 0.00 0.00
2 0 11 10 5 0 26
0.00 21.57 19.61 9.80 0.00 50.98
0.00 42.31 38.46 19.23 0.00
0.00 100.00 100.00 100.00 0.00
3 0 0 0 0 1 1
0.00 0.00 0.00 0.00 1.96 1.96
0.00 0.00 0.00 0.00 100.00
0.00 0.00 0.00 0.00 100.00
Total 24 11 10 5 1 51
47.06 21.57 19.61 9.80 1.96 100.00

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/_... 6/3/2014
SAS Output Page 5 of 10

The MEANS Procedure

CLUSTER N Obs Variable Label Mean


1 24 challenging challenging -0.2531116
easyuse easyuse -1.5630640
imagination imagination -0.7643012
educational educational -0.6444729
Social Social -0.7084017
2 11 challenging challenging -0.3744627
easyuse easyuse 0.4455197
imagination imagination -1.6330122
educational educational -0.1817246
Social Social 0.4928089
3 10 challenging challenging -1.3806742
easyuse easyuse 0.4767944
imagination imagination 0.5875779
educational educational -0.5291654
Social Social -0.7601491
4 5 challenging challenging 0.3646494
easyuse easyuse 0.0368852
imagination imagination 0.3074871
educational educational 0.3905650
Social Social -0.5288316
5 1 challenging challenging 1.0352606
easyuse easyuse 0.7470770
imagination imagination 0.9562708
educational educational 1.2166481
Social Social 0.7265595

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/_... 6/3/2014
SAS Output Page 6 of 10

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/_... 6/3/2014
SAS Output Page 7 of 10

The CORRESP Procedure

Inertia and Chi-Square Decomposition


Singular Principal Chi- Cumulative 10 20 30 40 50
Value Inertia Square Percent Percent ----+----+----+----+----+---
0.15548 0.02418 26.0129 50.30 50.30 *************************
0.12592 0.01586 17.0614 32.99 83.30 ****************
0.08402 0.00706 7.5962 14.69 97.99 *******
0.03110 0.00097 1.0407 2.01 100.00 *
Total 0.04806 51.7113 100.00

Degrees of Freedom = 36

Row Coordinates
Dim1 Dim2
Down-To-Earth -0.1328 -0.2611
Enthusiasts -0.2181 0.2743
Creatives 0.4054 0.0088
Intellectuals -0.0206 0.0054
Apathetic 0.1459 0.0657

Summary Statistics for the Row Points


Quality Mass Inertia
Down-To-Earth 0.9746 0.1283 0.2350
Enthusiasts 0.9726 0.0901 0.2369
Creatives 0.9533 0.0967 0.3468
Intellectuals 0.2982 0.6152 0.0194
Apathetic 0.2294 0.0697 0.1619

Partial Contributions to Inertia for


the Row Points
Dim1 Dim2
Down-To-Earth 0.0936 0.5515
Enthusiasts 0.1774 0.4279
Creatives 0.6570 0.0005
Intellectuals 0.0108 0.0011
Apathetic 0.0614 0.0190

Indices of the Coordinates That Contribute


Most to Inertia for the Row Points
Dim1 Dim2 Best
Down-To-Earth 0 2 2
Enthusiasts 2 2 2
Creatives 1 0 1
Intellectuals 0 0 1
Apathetic 0 0 1

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/_... 6/3/2014
SAS Output Page 8 of 10

Squared Cosines for the Row Points


Dim1 Dim2
Down-To-Earth 0.2003 0.7743
Enthusiasts 0.3766 0.5960
Creatives 0.9529 0.0005
Intellectuals 0.2790 0.0192
Apathetic 0.1907 0.0387

Column Coordinates
Dim1 Dim2
Male 8 -0.0383 0.1576
Female 8 0.3450 -0.0287
Male 9 -0.0590 0.2038
Female 9 0.2204 -0.0099
Male 10 0.0989 0.1047
Female 10 0.0930 0.0020
Male 11 -0.0791 -0.0730
Female 11 -0.0004 -0.2781
Male 12 -0.2041 0.0061
Female 12 -0.1335 -0.0811

Summary Statistics for the Column Points


Quality Mass Inertia
Male 8 0.8133 0.0743 0.0500
Female 8 0.9600 0.0613 0.1594
Male 9 0.8323 0.1208 0.1359
Female 9 0.9609 0.1264 0.1332
Male 10 0.4459 0.0781 0.0756
Female 10 0.6183 0.0799 0.0233
Male 11 0.3376 0.0948 0.0677
Female 11 0.9384 0.0855 0.1466
Male 12 0.9340 0.1338 0.1243
Female 12 0.8752 0.1450 0.0841

Partial Contributions to Inertia for


the Column Points
Dim1 Dim2
Male 8 0.0045 0.1164
Female 8 0.3020 0.0032
Male 9 0.0174 0.3163
Female 9 0.2539 0.0008
Male 10 0.0316 0.0539
Female 10 0.0286 0.0000
Male 11 0.0245 0.0319

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/_... 6/3/2014
SAS Output Page 9 of 10

Female 11 0.0000 0.4170


Male 12 0.2305 0.0003
Female 12 0.1070 0.0601

Indices of the Coordinates That Contribute


Most to Inertia for the Column Points
Dim1 Dim2 Best
Male 8 0 2 2
Female 8 1 0 1
Male 9 0 2 2
Female 9 1 0 1
Male 10 0 0 2
Female 10 0 0 1
Male 11 0 0 2
Female 11 0 2 2
Male 12 1 0 1
Female 12 1 0 1

Squared Cosines for the Column Points


Dim1 Dim2
Male 8 0.0454 0.7679
Female 8 0.9534 0.0066
Male 9 0.0643 0.7680
Female 9 0.9590 0.0019
Male 10 0.2104 0.2355
Female 10 0.6181 0.0003
Male 11 0.1821 0.1554
Female 11 0.0000 0.9384
Male 12 0.9331 0.0008
Female 12 0.6395 0.2357

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/_... 6/3/2014
SAS Output Page 10 of 10

The CORRESP Procedure

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/_... 6/3/2014
MSAN-624 V. Dimitrova P. 23

EXAMPLE 9: Finite Mixture Models (Branch)

PROC FMM DATA=SFG PLOTS=all;

MODEL debtinc = / KMIN=1 KMAX=3;

OUTPUT OUT=CL PROB(components) / ALLSTATS;

run;

The KMIN=1 and KMAX=3 options in the MODEL statement specify the minimum and

maximum number of mixture components to be evaluated.

The OUTPUT statement requests that all relevant statistics are output into a new

dataset, including the probability for each mixture component.

We use a finite mixture model to examine the variable debt-to-income ratio

(debtInc). The Akaike Information Criterion (AIC) and the Bayesian Information

Criterion (BIC) indicate that the distribution of the debt-to-income ratio variable is

a mixture of two distributions. One distribution representing consumers with a

lower debt-to-income ratio has a mean of 5.95% and a variance of 9.72. The other

distribution representing people with a higher debt-to-income ratio has a mean of

9.72% and a variance of 45.08. The two debt-to-income ratio segments are

approximately equal in size. The lower debt-to-income ratio segment accounts for

53.74% of the population, whereas the higher debt-to-income ratio segment is

46.26% of the population.

The output dataset CL contains information for each consumer in the

dataset, including the assignment of the consumer to Segment 1 or Segment 2.

23
MSAN-624 V. Dimitrova P. 24

Because the model does not involve covariates, the predicted values for the two

components are the respective means of those components. Respondent 1 has

99.77% probability of belonging to Segment 2, and therefore he/she has been

assigned to that segment. Accordingly, note that Respondent 1 has a relatively

large residual for Component 1 and smaller residual for Component 2.

EXAMPLE 10: Finite Mixture Models (Branch)

Evaluate the mixture for the Income variable, and then evaluate a model with

Income as a covariate and debt-to-income ratio as the dependent variable.

24
SAS Output Page 1 of 3

The SAS System

The FMM Procedure

Model Information
Data Set WORK.SFG
Response Variable debtinc
Type of Model Homogeneous Mixture
Distribution Normal
Min Components 1
Max Components 3
Link Function Identity
Estimation Method Maximum Likelihood

Number of Observations Read 1500


Number of Observations Used 1500

Component Description
for Mixture Models
Model ID Normal
1 1
2 2
3 3

Component Evaluation for Mixture Models


Number of
Components Parameters
Model Max
ID Total Eff. Total Eff. -2 Log L AIC AICC BIC Pearson Gradient
1 1 1 2 2 9949.52 9953.52 9953.53 9964.15 1500.00 2.24E-13
2 2 2 5 5 9650.26 9660.26 9660.30 9686.83 1499.99 0.00053
3 3 3 8 8 9650.26 9666.26 9666.36 9708.77 1500.02 0.00395

The model with 2 components (ID=2) was selected as 'best' based on the BIC (smaller is better).

Convergence criterion (GCONV=1E-8) satisfied.

Fit Statistics
-2 Log Likelihood 9650.3

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/_... 6/3/2014
SAS Output Page 2 of 3

AIC (smaller is better) 9660.3


AICC (smaller is better) 9660.3
BIC (smaller is better) 9686.8
Pearson Statistic 1500.0
Effective Parameters 5
Effective Components 2

Parameter Estimates for 'Normal' Model


Component Parameter Estimate Standard Error z Value Pr > |z|
1 Intercept 5.9491 0.2810 21.17 <.0001
2 Intercept 14.5540 0.5763 25.25 <.0001
1 Variance 9.7237 1.2458
2 Variance 45.0789 3.0091

Parameter Estimates for Mixing Probabilities


Linked Scale
Parameter Estimate Standard Error z Value Pr > |z| Probability
Probability 0.1501 0.1743 0.86 0.3894 0.5374

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/_... 6/3/2014
SAS Output Page 3 of 3

file:///C:/Users/vkaltche.LMUMAIN/AppData/Local/Temp/SAS%20Temporary%20Files/_... 6/3/2014

You might also like