RESEARCH SUPPORT
Factor analysis support for new and translated scales:
EFA or CFA
EFA vs CFA is decided by how much is already known about an instrument's structure, whether the scale is new, adapted or translated, and the loadings and fit report that follow can be examined by a committee or reviewer item by item.
No Commitment Required
Strictly Confidential
In-House PhD-Level Statisticians
Fixed Project Fee
Same-Day Response
How Is the Structure of an Instrument Decided?
Factor analysis support begins with one question about your instrument: how much is already known about its structure. A newly developed scale has none, so an exploratory analysis proposes one. The factor analysis of a translated scale starts from a published structure that a confirmatory analysis can test in the new language group. Choosing between EFA and CFA therefore follows prior evidence, not preference. That choice is made jointly with the researcher by in-house PhD-level statisticians at OnlineSPSS, a specialist statistical consultancy for academic, clinical and professional research, and the reasoning is recorded beside the results. The instrument and its theory are the researcher’s.
WHEN TO ENGAGE
Instruments Whose Structure Is Still Open
What a factor analysis can show depends on whether the items are new, borrowed, translated or already validated by other researchers.
Items Newly Developed The item pool was assembled from theory and interviews, and the number and meaning of the factors are not yet known, so an exploratory analysis has to propose them.
Scale Adapted to a New Group A published scale will be used with another age group or setting, and its published structure should be tested against the new responses before scores are compared.
Questionnaire Translated A questionnaire has been translated and back-translated, and the original factors and their loadings must be shown to hold before scores from two language versions are treated alike.
Reviewer Doubts the Structure A reviewer or committee has questioned the number of factors or an item's loading, and the structure must be re-examined along with the reasoning behind the first decisions.
ITEMS AND STRUCTURE
Screening Items, Counting Factors, Testing the Structure
Once the route is fixed, the item checks and the structural tests are planned around it.
Route Decision
Prior evidence, sample size and the stage of development decide whether work begins with Exploratory Factor Analysis (EFA) or Confirmatory Factor Analysis (CFA).
Item Distributions
Items are screened for floor and ceiling effects, skew and sparse response categories, and the association measure is matched to ordinal items instead of assuming they are continuous.
Number of Factors
The number to retain is judged from several criteria together, such as a comparison with factors from random data, the scree pattern, interpretability and fit, never from one cutoff.
Structure Testing
The proposed model is tested in a separate sample or, for a translated scale, in the new language group, with fit indices reported alongside their limitations.
Comparing Groups
Scores from two language or age groups are compared only after Measurement Invariance shows equal loadings and thresholds, as a shared factor pattern alone does not make them comparable.
Reporting the Structure
Loadings, factor correlations and fit are reported in the form journals expect, with any cross-loading or dropped item explained in the text and not left for the reader to discover.
STRUCTURE PITFALLS
Choices That Weaken a Reported Factor Structure
Instrument papers draw the same objections to a doubtful structure again and again, and most can be settled in the plan before any solution is seen.
Holdout Split Made Too Late Only one sample exists, and the holdout half is drawn after an exploratory solution has been seen, or the split leaves too few respondents per item in either half. A random split fixed before any analysis, sized for each step, keeps the test fair.
Rotation Chosen by Default An orthogonal rotation is applied though the factors are expected to correlate, which forces them to be uncorrelated and can distort the loadings of the items.
Ordinal Items as Continuous Five-point items with a sparse end category are analyzed as continuous, which can bias loadings and fit, and an estimator for ordinal data is usually the better starting point.
Fit Cutoffs Used as Rules Published cutoffs for fit indices are applied as thresholds for every model, though they were offered as rough aids that depend on the model and the sample.
Items Dropped Until Fit Improves Items are removed or error terms freed until fit looks acceptable, which capitalizes on chance and must be reported as a modification, not presented as a result.
Too Few Respondents A small sample for the number of items and factors makes loadings unstable, and no rule of thumb replaces a look at communalities and how well each factor is determined.
DELIVERABLES
The Structure Report and Its Supporting Files
A reviewer of an instrument paper looks for the decisions behind the loadings, so each is recorded next to the output it produced.
Route Rationale Note A short note on why an exploratory or a confirmatory route was chosen, citing the prior evidence and the sample, with statistical wording for the methods, matched to the output, for you to check and incorporate.
Item Screening Report Distributions, response patterns and inter-item association for each item, with the reason that any item was dropped, recoded or flagged for revision.
Loadings and Fit Summary Pattern loadings, factor correlations and fit indices, with confidence intervals where the method provides them, arranged in a form suited to a journal.
Retention Decision Record The criteria compared to settle the number of factors and the result of each, so that a reader can see how the decision was reached and which criteria disagreed.
Model Specification File The specification of the confirmatory model in the program used, covering paths, fixed parameters and any modification together with the reason for it.
Scale Scoring Key How subscale scores are computed from the items, including reversed items and missing responses, so that scores are comparable across datasets and across waves of data collection.
ILLUSTRATIVE EXAMPLE
A Translated Caregiver Scale Against Its Published Structure
An illustration of how a translated questionnaire is tested against the structure reported for the original version.
ILLUSTRATIVE EXAMPLE · CAREGIVER BURDEN SCALE TRANSLATION RESEARCH · UNITED STATES
A researcher in gerontological nursing translated a caregiver burden questionnaire into Spanish for use with family caregivers, and the English original had reported three factors.
Because a published structure existed, the plan was confirmatory: the three-factor model was specified from the original report and fitted to the new responses, with the items treated as ordinal. The complication was that overall fit was acceptable, yet two items loaded weakly on their intended factor in the translated version. Rather than adjust the model until fit improved, the researcher and the statistician reviewed the wording of those items with the translation team, reported the weak loadings openly and labeled the one modification as post hoc, to be tested in a new sample. The conclusion was worded as support for the original structure with two items needing revision.
Outcome: The researcher received the route note, the item screening report, the loadings and fit summary with the post hoc modification labeled, the model specification and the scoring key.
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
Fitting and Reporting the Measurement Model
Models drawn as path diagrams are fitted in AMOS, while the programs below cover the exploratory step and scripted confirmatory work.
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
The Researcher Decides What Each Factor Represents
Naming a factor is an act of substantive theory, and a statistical solution can suggest a grouping of items without deciding what it means.
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 research. Authorship of the validation study or article remains yours.
Your Research Decisions The instrument and its theory, the naming of the factors, the interpretation of the structure and the conclusions about validity 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
Choosing Between EFA and CFA for an Instrument
What a factor structure shows is easily misjudged when the sample is reused, the version is translated or the model happens to fit well.
What does factor analysis support cover for a questionnaire or scale?
The work covers the choice between EFA and CFA, the screening of items, the retention of factors, the testing of the proposed structure and the reporting of loadings and fit. New, adapted and translated instruments all fall within it, with each decision recorded in wording that can go into a methods section.
How is the factor analysis of a translated scale planned?
A translated scale is planned from the structure published for the original: the translation is checked item by item, that structure is tested in the new language group, and invariance across the two versions is examined before their scores are pooled. Where item meanings are in doubt, cognitive interviews with a few respondents come first. If the original model fits poorly in the translated version, the misfitting items are traced back to the translation notes before any respecification, and any change is reported as post hoc.
Does good model fit show that a scale is valid?
No. Acceptable fit shows that the proposed structure is consistent with the correlations among the items, and several different structures can fit equally well. Validity rests on further evidence, such as item content, relations with other measures and performance in the groups where the scale will be used, so fit is reported as one piece of that evidence.
Can scores from two language versions be combined in one analysis?
Only when measurement invariance across the versions has been tested, not assumed from the translation process. The same factor pattern in each version is the first step, but comparing or pooling scores also needs equal loadings and equal thresholds or intercepts, at least for most items. Where some items differ, a partial invariance model or a separate analysis per version is reported, since combining versions without that check can produce group differences that arise from the questionnaire and not from the people.
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 instrument, the item wording and the dataset, 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
Published Structure or New Scale:
The Route Comes First
A scale with a published structure calls for a different analysis than one still in development, and naming the route while no solution has been examined keeps the confirmatory claim credible.
No Commitment Required
Strictly Confidential
Fixed Project Fee
In-House PhD-Level Statisticians
Same-Day Response
Serving Researchers Since 2014