You are on page 1of 7

CONJOINT ANALYSIS

Conjoint analysis presents choice alternatives between products defined by


set of attributes
 Find product combination
 Experiment stimuli(filled data)
 Open collected data in spss
 Import/type the syntax
 output
DATA -> ORTHOGONAL DESIGN-> NAME THE FACCOTRS IN FACTOR NAME AND LABEL ->
ADD -> DEFINE VALUES AND LABEL->Create a new data file->reset the random number seed
to 20000000 (any value )-> options->minimum number of cases to generate(in general 18 to
22 ) combinations -> enter 18 in minimum number of cases to generate and 4 in number of
hold out case
*18 for developing model/ developing combination
*22 for testing
Data -> orthogonal->display

Card List
Card ID package brnad name price good house money back
design keeping seal guarantee
1 1 A* GLORY $1.39 no no
2 2 B* k2R $1.19 no yes
3 3 A* BISSELL $1.39 yes no
4 4 C* BISSELL $1.59 no no
5 5 B* GLORY $1.19 yes no
6 6 C* GLORY $1.19 no no
7 7 C* GLORY $1.39 no yes
8 8 B* k2R $1.39 no no
9 9 A* GLORY $1.59 no yes
10 10 A* k2R $1.19 no no
11 11 B* BISSELL $1.39 no yes
12 12 C* k2R $1.59 yes yes
13 13 A* k2R $1.59 no no
14 14 B* BISSELL $1.59 no no
15 15 A* BISSELL $1.19 yes yes
16 16 C* k2R $1.39 yes no
17 17 C* BISSELL $1.19 no no
18 18 B* GLORY $1.59 yes no
19a 19 A* GLORY $1.19 no yes
20a 20 C* k2R $1.39 no no
21a 21 A* k2R $1.19 yes no
22a 22 B* GLORY $1.39 no yes
a. Holdout
Syntax :
CONJOINT PLAN = 'C:\Users\KAVITHA SENTHIL\Desktop\conjoint.sav'
/DATA=*

/SEQUENCE=PREF1 To PREF22

/SUBJECT=ID

/FACTORS=

PACKAGE(DISCRETE)

BRAND(DISCRETE)

PRICE(LINEAR LESS)

SEAL(LINEAR MORE) MONEY(LINEAR MORE)

/PRINT=SUMMARYONLY.

Overall Statistics

Utilities
Utility Estimate Std. Error
package A* 1.550 .714
B* .000 .714
C* -1.550 .714
brand k2R .067 .714
GLORY -.267 .714
BISSELL .200 .714
price $1.19 .050 3.680
$1.39 .058 4.299
$1.59 .066 4.917
seal no .950 1.071
yes 1.900 2.143
money no 1.900 1.071
yes 3.800 2.143
(Constant) 5.642 4.776

Importance Values
package 39.479 First important

brand 19.684
price 5.667
seal 14.875
money 20.295 Second important
Averaged Importance Score
Add the options in the spss

Calculate Total utility score


=1.55-2.67+0.58+0.950+1.950=4.281
Continue this for all 22 combinations and select the highest value
Correlations
If less than 0.5 is not good
0.713 is good
Sig 0.00 related
 MULTINOMINAL LOGISTIC REGRESSION

Parameter Estimates
Preferred breakfasta B Std. Wald df Sig. Exp(B) 95% Confidence
Error Interval for Exp(B)
Lower Upper
Bound Bound
Breakfast Intercept - .322 13.105 1 .000
Bar 1.167
[agecat=1] .993 .318 9.747 1 .002 2.699 1.447 5.034
[agecat=2] 1.316 .322 16.671 1 .000 3.730 1.983 7.018
[agecat=3] .552 .342 2.602 1 .107 1.736 .888 3.393
[agecat=4] 0b . . 0 . . . .
[gender=0] -.135 .180 .561 1 .454 .874 .614 1.244
[gender=1] 0b . . 0 . . . .
[marital=0] .840 .194 18.808 1 .000 2.315 1.584 3.383
[marital=1] 0b . . 0 . . . .
[active=0] -.793 .183 18.693 1 .000 .452 .316 .648
[active=1] 0b . . 0 . . . .
Oatmeal Intercept 1.136 .238 22.790 1 .000
[agecat=1] - .534 64.121 1 .000 .014 .005 .040
4.272
[agecat=2] - .282 80.424 1 .000 .080 .046 .138
2.531
[agecat=3] - .218 29.713 1 .000 .304 .198 .466
1.191
[agecat=4] 0b . . 0 . . . .
[gender=0] -.006 .183 .001 1 .973 .994 .695 1.422
[gender=1] 0b . . 0 . . . .
[marital=0] -.260 .214 1.477 1 .224 .771 .507 1.173
[marital=1] 0b . . 0 . . . .
[active=0] .185 .188 .972 1 .324 1.204 .833 1.740
[active=1] 0b . . 0 . . . .
a. The reference category is: Cereal.
b. This parameter is set to zero because it is redundant.

Age category 1 (under 31) compared with age category4 above60 will likely prefer brekafast
Inactive lifestyle compared with active life style prefer break fast rather than cereals
significance 0.000
 AUTOMATIC LINEAR MODELING
DV IV
METRIC METRIC
CATEGORY/DUMMY LOADING

 Curve estimation
NUERAL NETWORKS
No data assumption
Linear/non linear
Disadvantage: interpretation is difficult.
Transform-> random number generators->set starting point->type fixed value 9191972
Transform-> compute variables->name as partition->in numeric expression type
2*rv.bernoulli(0.7)-1
Next neural network->multilayer perceptron the below screen

Include partition in partition ..it is created using the Bernoulli function

0.5 cutoff value for default


Reliability :

Reliability Statistics

Cronbach's N of Items
Alpha

.516 7

Multi collinearity
No collineriaty no common bias
Factor score is metric variable
Cluster is category variable
Amos
VAF= indirect effect/total effect
= 8 full median
2 to 8 partial
<2 no median
505.264

You might also like