“EFA or CFA?
I can’t tell.”
Match the method to what you already know about the scale and what you need to show.
- Exploratory versus confirmatory purposes
- Using an existing scale in a new population
- Sample size and item-count planning
FACTOR ANALYSIS AND SCALE VALIDATION · METHOD GUIDE
Factor analysis,
explained and defensible.
One-on-one help with exploratory and confirmatory factor analysis, reliability, and scale scoring. You make the measurement decisions and leave able to defend each one.
Book a free consultation30 minutes · Bring your output, your plan, or your committee's comments.
WHERE YOU MIGHT BE STUCK
Where factor analysis
gets stuck.
Most projects stall at one of three points.
Start with the one that fits.
“EFA or CFA?
“How many factors
should I keep?”
Retention is a judgment supported by several kinds of evidence.
- Parallel analysis, scree plots, and interpretability
- Rotation and cross-loadings
- Ordinal (Likert) items and polychoric correlations
“My fit or reliability
is poor.”
Find out whether the problem is the data, the scoring, or the model.
- Fit indices and what they cannot tell you
- Reliability beyond alpha, such as omega
- Reverse-scored and mis-specified items
DECISIONS THAT SHAPE YOUR ANALYSIS
Three choices committees look at.
01
Match the method to the purpose
Exploratory factor analysis explores a structure when you do not have an established model. Confirmatory factor analysis tests a structure you specified in advance. Using one to do the other’s job answers the wrong question.
02
Treat Likert items as what they are
Items with a few ordered categories often call for estimators and correlations designed for ordinal data, especially when responses are skewed. Treating them as continuous is sometimes defensible, but it is a decision to justify.
03
Reliability is not validity
Alpha and omega describe consistency. They do not show that a scale measures the intended construct, so report validity evidence separately.
START ON YOUR OWN
Free guides, tools,
and workflows.
Use these before or after your consultation.
EFA vs. CFA for your dissertation
When to use each method, how to retain factors, and the interpretation errors to avoid.
Read the guideReliability analysis, from alpha to ICC
Choose and report the right reliability coefficient for your design.
Read the guideReliability, factor analysis, and invariance
An editable R Markdown workflow for EFA, CFA, ordinal models, and measurement invariance.
See the workflowScale scoring and item analysis
Audit item direction, reverse-score, and compute documented scale scores.
See the workflowMy reliability is lower than expected
Check scoring, dimensionality, item behavior, and the limits of a single coefficient.
Read the CaseWhy is my SEM model fit poor?
Investigate misfit without chasing indices or making unsupported modifications.
Read the CaseACADEMIC INTEGRITY
Your analysis is your work,
and it should stay that way.
01
You are the author
Dr. Yel does not write your dissertation or any chapter of it. The decisions, the interpretations, and the prose are yours. His role is to advise, teach, and review.
02
You can explain every number
Nothing is handed over unexplained. We work through each method, assumption, and result until you can defend it on your own.
03
Your program's rules come first
Programs differ in what outside help they allow. Check your policy and let your chair know you're working with a statistical consultant.
Dr. Yel declines work that would misrepresent authorship or data, such as writing chapters, altering or fabricating data, or anything your program's policy prohibits.
A FEW THINGS YOU MIGHT BE WONDERING
Good questions.
Honest answers.
Should I use EFA or CFA for my dissertation?
It depends on whether you have a theory-based structure to test. Use EFA to explore a structure you do not yet have, and CFA to test one you specified in advance. Bring your scale and your research question and we will decide together.
Can you help with Likert-scale data?
Yes. We can decide whether ordinal methods are warranted for your items and how to justify the choice to your committee.
My Cronbach’s alpha is low. What should I do?
Check the scoring first, especially reverse-worded items, then the dimensionality, the item content, and the number of items. Do not delete items only to raise alpha.
Do I need factor analysis if I use a published scale?
Often you still report reliability in your own sample. You may also test the structure with CFA, particularly if you use the scale in a new population. Check what your committee expects.
Will you run the analysis for me?
Analyses are run together in live sessions, with you watching or at the keyboard, so you can explain every step. You keep the syntax or code from every session.
How much does factor analysis help cost?
The scope depends on your scale, your data, and the support you need. The free consultation is the place to discuss deliverables, pricing, and timing before you decide.
BRING THE SCALE THAT’S HOLDING YOU UP
Let’s make your measurement
ready to defend.
Meet with Dr. Yel for a free 30-minute consultation about your study, data, or committee feedback.
Book a free consultationCan't find a time that works? Email bookingrequest@dissertationstatshelper.com with a few times you're available and your time zone, and we'll find a time. It's a new address, so if you don't see our reply, please check your spam folder.