Skip to content
No Vape

Read random slopes and intercepts in vaping research

Distinguish modeled differences in starting level from differences in a relationship, without treating random effects as randomized assignment or a personalized treatment.

No Vape · Updated

Open book on a cafe table
Illustrative image · Lucille Emi Oh / Unsplash

Find the person or group attached to the term

A repeated-observation model can describe occasions nested within people or people nested within groups. Read which unit carries the random intercept or slope. An intercept is tied to the model’s reference conditions; it is not automatically a participant’s first chronological measurement or a permanent personal trait.

Piper and colleagues’ observational study of smoking and vaping contexts reports hierarchical logistic models with random slopes and intercepts. It provides a real example of the terminology in research about product-use events. Those modeled associations are not a recommendation to vape instead of smoke, a quitting intervention result or a measurement of the reader.

Megan E. Piper and colleagues / Addiction: Understanding contexts of smoking and vaping among dual users: analysis of ecological momentary assessment data

Draw a level difference before drawing a slope difference

For an invented sketch, imagine two people’s fitted lines across a generic predictor. With different intercepts and the same slope, the lines can sit at different levels while changing at the same rate. Allowing different slopes lets their modeled relationships with that predictor differ. The drawing contains no actual participant data or outcome forecast.

Stata’s official multilevel example extends a random-intercept model by adding a random slope for a named predictor. It illustrates that an additional varying coefficient is a particular model structure, rather than just another label for having repeated rows. Keep the functional form and outcome scale in view; a logistic model does not become an ordinary straight-line probability model because the sketch is simple.

StataCorp: Linear multilevel models

Keep modeled variation separate from a subgroup explanation

A random slope does not automatically explain why relationships vary, identify a clinical subgroup or prove that changing the predictor causes a result. A model may also include fixed effects or interactions with their own questions. Read the reported structure instead of describing every between-person difference as the same analysis.

The vaping-context report explains that some more complex models did not converge and were fitted with random intercepts only. Retain what was actually fitted for the particular result you are quoting. Do not call a fallback model a random-slope result just because the original plan or another table used that structure.

Megan E. Piper and colleagues / Addiction: Understanding contexts of smoking and vaping among dual users: analysis of ecological momentary assessment data

Write a precise structure statement

Use a short note naming the repeated unit, the varying level, the predictor whose slope varies and the actual reported specification. A title saying multilevel is not enough to fill those fields. Leave an unspecified element open rather than assigning a plausible random effect yourself.

This does not audit the model or decide whether its complexity was appropriate. It prevents random-effects terminology from becoming a claim about randomization, diagnosis or an individualized benefit. Appropriate statistical and clinical expertise is needed for consequential use of the findings. Your own repeated notes can support ordinary planning without being described as fitted personal random slopes.

Sources and scope

Sources inform the guidance. The worksheets and examples are No Vape editorial suggestions, not tested treatments.

  1. Megan E. Piper and colleagues / Addiction — Understanding contexts of smoking and vaping among dual users: analysis of ecological momentary assessment data
  2. StataCorp — Linear multilevel models
  3. No Vape — No Vape — app features and store links

Written with AI assistance and checked against the cited sources. No clinical review is claimed. This guide covers practical planning, not individual medical treatment. How we research and write