Showing results for 
Show  only  | Search instead for 
Did you mean: 
Submit your abstract to the call for content for Discovery Summit Americas by April 23. Selected abstracts will be presented at Discovery Summit, Oct. 21- 24.
Discovery is online this week, April 16 and 18. Join us for these exciting interactive sessions.
Choose Language Hide Translation Bar

How do I write the data table with repeats and replicates

I am doing a response surface methodology with an entire design replication. I have collected the data and have repeats for each run. Hence I was wondering on the data structure for inputting the repeats in JMP. Minitab structure is to input the repeats in each column. Is this the same for JMP?

Super User

Re: How do I write the data table with repeats and replicates

Hi @sianghongtay22,


It seems you're using replication and repetition indifferently, but these two terms refer to different techniques in DoE that can be used in combination:

  • Repetition is about making multiple response measurements on the same experimental run (same sample without any resetting between measurements). Repetitions only reduce the variation from the measurement system (by using the average of the repeated measurements for example).
    Your datatable will be structured by adding several columns for each response, corresponding to the repetitions of the measurement on the same experimental unit. You can then combine these columns of results by adding a calculated column of the mean, median, or any other aggregated measure.

  • Replication is about making multiple independent randomized experimental runs (multiple samples with resetting between each runs) for each treatment combination. Replications reduce the total experimental variation (process + measurements) in order to provide an estimate for pure error and reduce the prediction error (with more accurate parameters estimates).
    You can replicate an entire design by using the platform Augment Design and choosing the option "Replicate" after entering the factors and responses in the windows panel : Replicate a Design
    In the Custom Design platform, you also have the option to choose a number of runs that will be replicated (called "Replicate Runs"). This does not replicate the entire design, but chooses the optimal design points to replicate : Design Generation 
    The result of replication is the addition of new experimental units in the design (= rows), with same treatment combinations as other previous runs. The order should be randomized.


I hope this answer clarify the differences between the two and how to structure your table depending on your choices of replication and/or repetitions,

Scientific Expertise Engineer
L'Oréal - Data & Analytics
Super User

Re: How do I write the data table with repeats and replicates

Some clarification/augmentation to Victor's response.  It is important to think of experimental units when using the terms repeats and replicates.


For repeats, there is 1 experimental unit for each treatment combination.  That experimental units is measured multiple times.  How those repeated measurements are taken will affect which components of variation are quantified.  For e\xample, if the experimental unit is a batch, you could measure the exact same sample from the batch multiple times and this would likely capture the measurement variation.  If you sample the batch twice and measure each sample once, you would estimated both the within batch and the measurement variation (confounded).


For replicates, you will have multiple experimental units for the same treatment combination.  These may be randomized or collected in blocks.  Replication does not necessarily reduce the variation (in fact it may be quite the opposite).  For example, if the replicates are done over changing lots of raw material and raw material effects variation, you will actually increase the variation in the experiment.  Replication done randomly does allow for an estimation of the experimental error (hopefully with less bias providing a more robust statistical test), increases the inference space, however not can compromise precision.  It does not allow for the assignment of the error.  Replicates run in blocks, has the advantage of increasing inference space and increasing precision of the experiment (the block effect is accounted for in the model).  In addition, blocks treated as fixed effects, has the advantage of assigning the experimental errors.


Now for your questions:

There is no context for your situation, so it is impossible to provide specific advice.  However, to set-up the JMP table, you need a separate row for each replicate.  I add extra columns for repeats (for each row).  To analyze, stack (Tables>Stack) the repeat columns, assess consistency and then summarize the repeat data if appropriate (Tables>Summary>Mean & Variance).  Analyze the summary data.

"All models are wrong, some are useful" G.E.P. Box