HOME CONTACT ABOUT US SUPPORT SENSORYTEST.COM
| |||||
|
|
HOME CONTACT ABOUT US SUPPORT SENSORYTEST.COM |
|
|
 |
 SIMS SOFTWARE | QUESTION CHOICES | CLIENT COMMENTS |
| MARKETS | VIEW REPORTS | FREE EVALUATION | |
| TEST CHOICES | GET DEMO | SUPPORT |
| INCOMPLETE BLOCK PAIRED COMPARISON EXAMPLES
Incomplete blocking such as 3C2, 4C2, 5C2, etc. when Samples >=3, and presenting only two samples. |
Discrimination Paired Comparisons Incomplete Block Paired Comparisons View More Question Type Choices |
|
Incomplete block paired comparison is also commonly referred to as multiple paired comparision, multiple paired preference, pairwise ranking, and incomplete block paired preference.
Panelist Ballot 
Two alternative methods for doing this in SIMS 2000. 1) Ranking; Paired Comparison is similar to Pairwise Ranking. Use an Affective type Questionnaire, SIMS 2000 Ranking Attribute question, "rank all". 2) Hedonic; using SIMS 2000 hedonic attribute type, "select one". The rest of this webpage discusses the Hedonic method, functionality which was added to SIMS 2000 during April 2010. Questionnaire Set up the ballot as an Affective questionnaire using Hedonic Attributes instead of the typical two sample Discrimination PC. Affective questionnaire will allow for more flexible Incomplete Block Experimental Design and Experimental Plan Rotations. Create an Hedonic attribute type with two choices and label the choices them with the following special TAGs. <RELATIVE_SAMPLE_1_BLINDING_CODE> select the choice return value to be = 1 <RELATIVE_SAMPLE_2_BLINDING_CODE> select the choice return value to be = 2 The above example will present two choices with the blinding codes for the two samples in randomized order perfectly matching the Rotation Plan in the Test Definition. Example: Which Sample do you prefer OVERALL? ( ) 135 ( ) 246 Note: The "Attribute Seen With:" option should be set to "Relative Sample -- Last Sample", so that this question is only seen once at the end of the ballot. Note: Other handy TAGS are available as with all questions types in SIMS 2000. See the [Notes] buttons. See Print screen examples shown below for Questionnaire Setup Experimental Design and Plan Set up your Experimental Design specifying the true absolute number of samples, which is 3 or more for incomplete blocks of PC tests. Select an Experimental Plan, such as Combination 3 choose 2, or create your own. Number of samples presented is always 2 for PC. The Design can be either a Balanced Incomplete Block (BIB) or an Unbalanced Incomplete Block (UBIB). In a BIB Design, there are m = t(t-1)/2 possible pairs, such as for a 3C2 test there are 3 possible pairs, 4C2 6 pairs, etc. In a BIB Design, to be completely balanced, all pairs must be seen equal number of times. And to eliminate order effect, also present each pair in reserve (m*2) Create your Test Definition and review your Rotation Plan for accuracy. As always, "Test Your Test". Reporting and Data Since this is a new and advanced type of question and setup, some attention to details is required for data interpretation. The raw data saved is the actual 'return values', just like every hedonic question type in SIMS 2000, therefore the raw data return values are *not* the Sample #s, they are positional orders. • The SIMS 2000 Raw Data Export Report, by default, the columns for these hedonic questions will display the actual 'return values', not the sample selected. The raw data can be interpreted by reviewing the actual return values from each of your panelists while also reviewing which sample set pair the panelist was presented with from the Test Definition Rotation plan. • The SIMS 2000 Raw Data Export Report can optionally swap in the blinding-code associated with each panelist response, this may only be beneficial for interpretation if your experimental design selected the option for the Same Blinding Codes for all panelists. • The SIMS 2000 Data Frequency Reports will add additional report lines interpreting the data when these Special Hedonic Sample Choice Tags in Use is detected. Example: This design was Control vs 3 Tests. Show each panelist 3 pairs, select preference. C vs T1 C vs T2 C vs T3
The experimental design was 4 samples, 2 presented, 3 reps. 4C2R3. 10 panelists and 30 pairs evaluated.
Frequency Report:
Panelist selected frequency for each Absolute Sample:
Absolute Sample 1 Selected Frequency = 14 47%
Absolute Sample 2 Selected Frequency = 5 17%
Absolute Sample 3 Selected Frequency = 4 13%
Absolute Sample 4 Selected Frequency = 7 23%
The Frequency grid of Selected vs Non-selected Absolute Samples.
Read the grid from Left to Right.
Example: S1 selected over the S2, head-to-head, n times, and %.
vs | S1 | S2 | S3 | S4 |
-----------------------------------------------------------------
S1 | - | 5 (50%) | 6 (60%) | 3 (30%) |
S2 | 5 (50%) | - | - | - |
S3 | 4 (40%) | - | - | - |
S4 | 7 (70%) | - | - | - |
• The SIMS 2000 Test Data Results Reports will add additional report lines interpreting the data when these Special Hedonic Sample Choice Tags in Use is detected. Example: Replication: 1
Sample Set: 1 Panelist: Anonymous Panelist ... saw block number: 1
Response: 2 Attribute: Sweetness Preference(Hedonic Scale)
Label(2) = 101 = <RELATIVE_SAMPLE_2_BLINDING_CODE>
Absolute Sample 1 was Selected, Sample description: Control
Panelist samples presented: 2-102 vs 1-101
• Advanced analysis, see SIMS 2000 Reports | Discrimination Reports | Paired Comparison Attribute Analysis - Select option: [X] Show Affective Incomplete Block PCs. Select your desired sample pair for analysis. • For additional advanced analysis, use your own Incomplete Block Paired Comparisons statistical calculations using the raw data and proper data interpretation. • Other reports in SIMS 2000 may need to be scrutinized for proper data interpretation. |
|
Questionnaire Setup 
|
|
(The Scales on this Web page are not interactive.) References: Statistical Methods In Food and Consumer Research, Gacula, Singh, Bi, 2nd Ed, Chapter 11, pg 559 |