RESEARCH SUPPORT
JMP Data Analysis for Research
Designed Experiments and Exploratory Models
Factors and run counts planned before the first sample, models fitted to the runs and plots that show what the data support, each choice checked against how the experiment was performed and recorded in files that you can reopen and rerun later.
No Commitment Required
Strictly Confidential
In-House PhD-Level Statisticians
Fixed Project Fee
Same-Day Response
What Does an Analysis in JMP Involve?
JMP data analysis for research means using JMP, the interactive software from the SAS group, to design experiments and to model the data they produce, often in laboratory, process and engineering research settings. A project may begin with JMP design of experiments, where factors, ranges and the number of runs are fixed before anything is measured, or with a dataset already collected that is explored through distributions, graphs and fitted models. The analysis is carried out in-house at OnlineSPSS, a specialist statistical consultancy for academic, clinical and professional research, by PhD-level statisticians who check each model against the way the runs were performed and report effects with their confidence intervals.
WHEN TO ENGAGE
The State of the Experiment Decides the Work
A plan on paper calls for a design, completed runs call for a model that respects how they were carried out, and data logged outside any design call for exploration before modeling.
Factors Chosen, Runs Not Set A study in Life & Biological Sciences has several factors and costly runs, so the design type, the run count and the power to detect the effect of interest are settled before samples are prepared.
Runs Done, Model Open A screening or optimization experiment has been run and the model terms are undecided, so main effects, interactions and curvature are weighed against the number of runs available.
Process Data, No Design Measurements come from an instrument or process log and not from a designed experiment, so distributions, outliers and relationships are explored before a model is trusted for prediction.
Model Challenged in Review A coauthor or reviewer questions the model terms, the randomization or a prediction outside the tested factor ranges, and the fit is checked again against how the runs were performed.
DESIGN AND MODELING
Experiments, Models and Graphs in JMP
JMP statistical analysis for research joins design, modeling and graphing in one interactive environment, and each part is used for what it estimates and what the data can bear.
Custom and Screening Designs
The DOE menu offers custom, screening and response surface designs, and the choice rests on the number of factors, the runs the study can carry and the model expected, with a definitive screening design as one option when curvature may matter.
Design Evaluation
Before any run is made, Evaluate Design reports a candidate design's power and how its prediction variance spreads across the factor space, so a design with too little power for effects of the size that matters can be enlarged before material is committed.
Fit Model Analysis
Fit Model estimates the least squares or mixed model the design supports, as in Regression & Predictive Modeling, with whole plots entered as a random effect where randomization was restricted, and residual plots check the fit.
Profilers and Settings
The Prediction Profiler shows how the fitted response changes as each factor moves within the tested ranges, and its desirability functions weigh several responses at once when a compromise setting is sought.
Distributions and Graph Builder
Distribution and Graph Builder show shape, outliers and group differences before modeling, and they reveal early whether a response may need a transformation, which matters most for process data that never came from a designed experiment.
Scripts and Wider Analysis
JSL scripts make an analysis repeatable from the raw table to the final figure, and questions beyond the experiment itself belong to Quantitative Data Analysis.
DESIGN RISKS
Design Flaws a JMP Report Will Not Flag
None of these problems produces a warning in JMP, because each comes from how the runs were set up, repeated or reported.
Restricted Randomization Ignored A factor that is hard to change is set once for a group of runs, and a model that treats every run as independent understates the standard error of that factor's effect, so its precision and significance are overstated.
Repeats Taken for Replicates Repeat readings from one prepared sample are analyzed as independent runs, which shrinks the error estimate, whereas true replicates reset the factor settings and the preparation.
More Terms Than Runs Allow A model with nearly as many terms as runs leaves little information about error, so effects ranked by size lack reliable intervals and need confirmation runs.
Curvature Never Checked A two-level screening design without center points cannot reveal curvature, so the fitted surface places any optimum at an edge of the tested ranges, and center runs or a response surface design are needed to locate one inside them.
Many Responses, Many Tests Several responses and effects are examined and the ones crossing a threshold are reported, which inflates false positives unless the plan names the primary comparisons.
Platform Options Left Unstated Estimates depend on options such as the coding of effects, the estimation method and the version, so these are recorded beside every table and not left to memory.
DELIVERABLES
The Design Record and Model Files You Receive
The design table, the saved model and the printed plots are kept consistent with one another, so a figure in the manuscript can be traced to the runs that produced it.
Design Table and Rationale The factor table with levels, run order and randomization restrictions, and the reasoning behind the design type and the number of runs, for you to check and incorporate in a methods section.
Saved JMP Data Tables JMP data tables with the analysis scripts attached, so that any report can be reopened and rerun by a coauthor who has JMP, with the original data left untouched.
Model Selection Note Why each effect was kept or dropped, the criteria used and how the conclusions changed under simpler and richer models, in terms of your experiment.
Effect Estimates With Intervals Estimated factor effects with 95% confidence intervals, the error variance and the number of runs behind each estimate, laid out in the table style of your target journal.
Publication-Ready Plots Profiler, interaction and distribution plots redrawn for print, with axes, units and legends set for the manuscript and not left at screen defaults.
Confirmation Run Plan A short plan for confirmation runs at the settings the model favors, with the prediction intervals the new results should fall within if the model holds.
ILLUSTRATIVE EXAMPLE
Incubator Blocks and a Hard-to-Change Factor
An illustration of how a screening experiment on algal growth is designed in advance, and how its analysis adapts when one factor, temperature, cannot be reset for every flask.
ILLUSTRATIVE EXAMPLE · MICROALGAE GROWTH MEDIUM RESEARCH · UNITED STATES
A postdoctoral researcher in bioprocess science wanted to learn which of five factors, including temperature, light and nitrate level, most affected the biomass yield of a microalgae culture, with a limited number of flasks.
A custom screening design was generated in JMP after the number of runs was weighed against the power to detect an effect of the size that mattered. The complication was practical: only two incubators were available and each could hold one temperature at a time, so temperature was set once per incubator run and the flasks were spread over several successive incubator runs. The design was therefore built with temperature as a hard-to-change factor, and the model treated each incubator run as a random whole-plot effect, so that the temperature effect was judged against the variation between incubator runs and not against the flasks, with the wider interval that a small number of incubator runs implies. Interactions were kept only where the data supported them, and the settings favored by the profiler were framed as predictions to be confirmed by further runs.
Outcome: The researcher received the design table, the factor effects with 95% confidence intervals, the profiler plots, the saved JMP scripts and a plan for confirmation runs at the favored settings and their prediction intervals.
This example does not describe a real client or project. We never share details of actual client work, in line with the confidentiality we promise every researcher. This illustrative example is included to show the type, scope, and standard of work involved in this kind of analysis.
SOFTWARE
Refitting a JMP Model Outside JMP
A coauthor or reviewer may want the model refitted in another program, and Minitab is a natural destination for teams that run their designed experiments there. Other programs are listed under Software.
HOW WE ASSESS YOUR PROJECT
We Review Your Actual Research
Your assessment is based on a review of your actual research materials and analytical requirements, not a generic project category.
You Submit Your Project Details
Share whatever is available at the current stage, such as the research aim, protocol or proposal, dataset, variable list, manuscript or statistical requirements.
We Review Your Research and Define the Scope
Your statistically trained Research Coordinator reviews the materials to understand the study design, research questions, data structure, analytical requirements and expected deliverables. If clarification is needed, additional documents or information can be requested before the assessment is finalized.
Your Analysis Is Completed and Delivered
Once the scope and fixed project fee are agreed, the in-house statistician completes the work and the Research Coordinator remains the communication point through delivery, with the analysis, deliverables and timeline as agreed.
RESEARCH RESPONSIBILITY
Which Factors Matter Is a Scientific Decision
Choosing the factors, their ranges and the response that counts draws on knowledge of the system under study, and the conclusions drawn from the estimates belong to the research team.
Agreed Statistical Support OnlineSPSS provides the agreed methodological, analytical and statistical reporting support, defined in a scope that is confirmed before any work begins.
No Intellectual Contribution OnlineSPSS does not claim any intellectual contribution to your thesis, article or report. Authorship remains the researcher's.
Your Research Decisions The research question, the choice of factors and ranges, the scientific meaning of the estimates, the discussion and the conclusions remain yours, with overall responsibility.
Confidential Handling Research data and project details are handled confidentially, project information is not published, and an NDA is available where required.
FREQUENTLY ASKED QUESTIONS
Timing, Profiler Settings and JMP Editions
The design is best fixed before samples are prepared, a profiler setting is a prediction to be confirmed, and the edition you hold limits some models.
Where does JMP data analysis for research start and stop?
JMP data analysis for research at OnlineSPSS means designing experiments in JMP, fitting and checking the models for the data they produce, and preparing the results for a manuscript or thesis. Both stages are covered, the design before the runs and the modeling after them, and a project may need only one. A dataset collected years earlier with no design behind it can also be modeled.
Should the design be set before any samples are prepared?
Yes, because that is the stage at which statistical input changes the most. The factors, their ranges, the number of runs and the power to detect an effect of interest are settled in JMP’s design tools before material is committed, and center points or true replicates are built into the design instead of added at analysis.
Is the optimum from the Prediction Profiler a final result?
No. The profiler reads a fitted model, so a proposed setting is a prediction with uncertainty, and it is trustworthy only within the factor ranges that were tested. A confirmation run at the proposed setting, compared with its prediction interval, is the usual way to learn whether the model holds there.
Which JMP edition and version does my project need?
The analysis you need decides this, and some modeling platforms, such as Generalized Regression, are offered in JMP Pro and not in every edition. Menus and options also differ between versions. The assessment therefore asks which edition and version you hold, and the report records the ones used.
How do you protect my research data and confidentiality?
Your data and project details are handled confidentially, and project information is not published. Researcher names and institutions are not used publicly without permission. An NDA is available where required and can be arranged before sensitive materials are shared. Read more about confidentiality and data security.
What happens when I request an assessment, and how long does it take?
You share your research question, design and relevant materials, such as the factor list, any JMP data tables, the run log and the output you already have, and a statistically trained Research Coordinator reviews the request and responds the same day. The analysis, deliverables, timeline and one fixed project fee are then confirmed before any work begins, so the timeline follows from the agreed scope.
Still have a question? Find more answers in our full FAQ.
NEXT STEP
Design Your JMP Experiment
Around Its Constraints
Restricted randomization and repeated readings are easier to plan into a JMP design than to account for after the runs, though a completed experiment can still be analyzed with a model that reflects how it was performed.
No Commitment Required
Strictly Confidential
Fixed Project Fee
In-House PhD-Level Statisticians
Same-Day Response
Serving Researchers Since 2014