Standard Test: Accelerated Shelf-life Testing

Standard Test: Accelerated Shelf-life Testing

Overview

This workflow supports accelerated shelf-life testing using the Difference From Control (DFC) method with a Category intensity scale. The objective is to determine whether panelists can detect a perceptible difference between a test sample and the control. 

Idea
If you are performing acceptance testing or need to use Line Scale attributes instead of Category, you will have to alter the steps to expand the questionnaire to collect information on multiple attributes as per your shelf-life testing requirements.


Notes
Real-time shelf-life time point testing workflow is also available if that method better suits your testing needs. Please refer to our main shelf-life testing page for guidelines and additional information about utilizing Compusense for shelf-life testing.


Info
Our recommendation is to link samples to products. Our examples are based on the assumption that you have products in the Products library.
However, if you do not have products in the Products library and your organization is not planning on using it, please ignore the parts of the workflow referring to sample/product linking. Everything else in the workflows is still applicable to your scenario.


Before You Begin

Accelerated shelf-life testing can be done ideally in a single test or in separate tests. The objective is to have the same panelists evaluate all samples to eliminate dealing with missing data. This can be achieved when all samples are possible to evaluate in a single sitting, or potentially on the same day if multiple sittings are needed.

You would use one test for all samples if there aren't that many samples and they are not fatiguing so that panelists can evaluate them all in one sitting or in multiple sittings in a single day. On the other hand, if there are too many samples to evaluate in a single sitting, or if the samples are otherwise fatiguing, the recommendation is to separate them into different tests. If all such samples were in a single test that you have to run over multiple days or weeks, you could have a risk of a panelist coming to one of the sessions, but being unable to attend another. You would end up with missing data. When time points are set up as separate tests, you will not have to deal with missing data due to someone's schedule conflict for a specific time point evaluation.

When analyzing data across tests, if you have an equal number of responses for each sample, there will be one Dunnett’s t value calculated and applied across all samples.

If the N is different (when you have an unequal number of responses for each sample), the Dunnett’s test will be run on a per pair basis. Dunnett’s t value will be calculated for each pair (and the column will not appear in the summary report because it is not the same across all samples). For example, control vs. sample 2 will use a different t value than control vs. sample 3. 

You can control which sample sets to exclude by using Filters.

The assumption in this workflow is that you will have a marked control available for panelists to compare the test samples against. If you will not have control available for your testing (i.e. if you will be using acceptance testing method, or intensity scale), you will have to alter the steps to expand the questionnaire to collect information on multiple attributes as per your shelf-life testing requirements. Please contact Compusense Support for assistance as needed.

The marked control sample should not be included inside the test. Panelists are not expected to collect data for the sample that is clearly marked as a control, but rather simply use it to compare other samples against it. Therefore, it should be presented to panelists as a clearly marked control, but should not be included in the test for data collection on it.

Your time point 0, however, will have to be a blind control with the sample type set to "control". Detailed instructions are outlined in this workflow. The blind control set as a "control" sample type will be used in the analysis to compare all the other time points (samples) against it. Other time point tests will have to have sample types set to something other than "control" for analysis to work properly.

If you link your samples to products, your analysis can be run from the Products library. If you do not link samples to products, your analysis can be run from the Advanced search & analysis across tests.



Use DFC Template Test For Test Setup

The Difference from control (DFC) template test comes with 4 preset samples, with a Category Intensity scale as a DFC question, and a follow up Choose n question that prompts those panelists who indicated that there was difference to select from the list of choices where that difference lies. Connections were used for this dynamic prompt. Follow the steps below to use this template test.


  1. In the Dashboard, click Create a new test. Scroll down to the Difference from control test and click Use.

  2. Go through the Overview and Samples & design tabs to set those components up as necessary:
    1. Update sample sets if needed. It is better to add more than enough sample sets than not have enough.

    2. Update the samples, such as the number of them, the names, blinding codes, and other information as necessary.

      Important:       Set the blind control sample's Sample type to Control. This will enable the Dunnett's analysis to work properly. If you missed to do this step during the setup, you can still do it after data is collected, but before test is set to Complete.


    3. Link the sample to the corresponding product in your Products library.

      If it does not work for you to create products in the Products library, that's OK. Continue with the setup without linking the samples to products. You will still get analysis in Compusense, just through a different set of steps, as described further down in the workflow.

    4. The design is by default Balanced. Review and update as necessary.

  3. In the Build tab:
    1. Review the Category Intensity and its Question options, and update as necessary to meet your testing needs.

    2. If you wish to ask additional or completely different follow up questions than the preset Choose n question, add them and set Connections as needed. Choose n and Choose 1 questions work well with our Products over time report.

      If you do not want to ask any follow up questions, delete the Choose n question.


    3. Update the text visible to panelists:
      1. You can update any of the preset text, such as:  Welcome screen  and Thank you screen.


      1. Update your text/instructions in the DFC question or any other follow up questions and screens that you may have added.


      1. Question and attribute names are important to be descriptive for your reporting purposes, which can also be displayed to panelists through the Question options
         and Attribute options if desired.

    4. Go to the Attribute options (not the Question options) of the Category question, and update the choices as needed to meet your testing objective. Update the choice descriptors and any other options you find useful while in the Attribute options screen.
       

  4. Update the Panelists and Logistics tabs as necessary, preview and run the test .


Use A Blank Test For Test Setup

  1. Create a Standard test from a blank template. Go through the Overview and Samples & design tabs to set those components up as necessary:
    1. Add sample sets. It is better to add more than enough sample sets than not have enough.

    2. Add generic samples and link them to the relevant product, or import samples with the product information included in the file, or create samples from a product, or add samples from the Sample list library.

      Important: Set the blind control sample's Sample type to Control. This will enable the Dunnett's analysis to work properly. If you missed to do this step during the setup, you can still do it after data is collected, but before test is set to Complete.


      If it does not work for you to create products in the Products library, that's OK. Continue with the setup without linking the samples to products. You will still get analysis in Compusense, just through a different set of steps, as described further down in the workflow.

    3. The design is by default Balanced. Review and update as necessary.

    4. Update the blinding codes if necessary.

  2. In the Build tab:
    1. In the Questions bar on the left, expand the Category list, and click Intensity to add the Category Intensity question into your test. If you prefer to use the Line Scale, click the Line Scale instead.

    2. Click in the "Add question name" and name it well so you know what the question is all about.

    3. Click in the "Add attribute name" to update the attribute as desired.

    4. If you wish to ask any follow up questions, add them and set Connections as needed.

    5. Update the text visible to panelists:
      1. You can update any of the preset text, such as:  Welcome screen  and Thank you screen.


      1. Add your text/instructions in the DFC question or any other follow up questions and screens that you may have added.


      1. Question and attribute names are important to be descriptive for your reporting purposes, which can also be displayed to panelists through the Question options
         and Attribute options if desired.

    6. Go to the Attribute options (not the Question options), and update the choices so that there are 6 of them, or however many you need to meet your testing objective. Update the choice descriptors any other options you find useful while in the Attribute options screen.
       

    7. If you wish to ask panelists any follow up questions, add them in and use Connections to display them based on panelists' responses on the scale question.

  3. Update the Panelists and Logistics tabs as necessary, preview and run the test .




Subsequent Test Setup

Warning
Follow these steps only if it is not possible for you to evaluate all samples in one sitting or in one day (to ensure same panelists evaluate all samples and therefore avoid dealing with missing data).


  1. When you are completely happy with the first test setup (whether you created it from a template or from a blank test), clone the test to create a test or tests for additional set of accelerated shelf-life samples. Be sure to go through each tab in the new tests and update information as necessary.

  2. In the Samples & design tab, change the Sample type into something other than Control or Blind Control. Only the first test should have a sample type set to Control. If you do not make this update in subsequent time point tests, your analysis might be affected.


  3. Update other details in the Samples & design tab, such as blinding codes, sample name, etc.

  4. Update the Panelists and Logistics tabs as necessary, preview and run the test .


Shelf-life Data Analysis

When done with the data collection, pause each test that you may have ran and double-check the following:
  1. Whether you set the Sample type for one of the samples in the first test (or only test) to Control.

  2. Whether sample names reflect the correct time point. This will be very important during the analysis.


Set the test to Complete, and analyze the data by following a workflow that best suits your analysis needs:
  1. Dunnett's analysis for shelf-life time point testing data linked to products.

  2. Dunnett's analysis for real-time shelf-life time point testing data not linked to products.

  3. Products over time graphing based on action standards.



    • Related Articles

    • Standard Test: Shelf-life

      Overview Sensory shelf-life testing can be performed to determine the length of time that a product’s sensory characteristics stay within your organization's acceptable range. Typical examples of shelf-life testing uses: To determine the ideal ...

    • Standard Test: Real-time Shelf-life Time Point Testing

      Overview The focus of this workflow is real-time shelf-life time point testing using the difference style (Difference From Control) with Category intensity scale. The objective is to see whether the panel can identify a difference in a test sample ...

    • Analysis Across Tests: Dunnett's Analysis For Shelf-life Testing Data

      Overview The recommended analysis for shelf-life testing data is the Dunnett's test. Dunnett's test compares a (blind) control against each test sample. This workflow will help you select specific analysis options when generating the PowerPoint ...

    • Products Reports: Dunnett's Analysis For Shelf-life Testing Data

      Overview The recommended analysis for shelf-life testing data is the Dunnett's test. Dunnett's test compares a (blind) control from the first test (time point 0) against each test sample from to subsequent tests (time points). This workflow will help ...

    • Standard Test

      Overview Collect quantitative data on your products using the Standard test type. Use this test type to access a variety of question types for descriptive testing, quality control testing, or consumer testing. Question Compatibility The Standard test ...