In this case outcome is inevitable. There would not even be one case where the outcome doesn't occur. The intention is to develop a model that predict when that outcome occurs based on a set of influencing parameters. What would be the appropriate statistical analysis to be employed together with estimation of sample size?
Sample size: The most important parameter to be considered for estimation or else can calculate for each parameter and take the maximum
Need some more information on the nature of the data, but a Cox proportional hazard model would probably be appropriate to model the time-to-event outcome. To determine a sample size, there is a lot of information/assumptions needed as well as specifying your hypothesis. What type of effect do you want/expect to see? Do you have any background data on what the variability of the data will be? How many predictors do you plan to include in the model? etc.
@ Joe Stanek. Thanks for the response. The intention is to develop a prediction model of when a rotary instrument breaks based on four parameters of stress undergone by the instrument, cracks, surface micro changes, and roughness of the surface of the instrument. The inevitable thing is all the rotary instruments in the study would eventually break at some point. The point of emphasis is what combination of the influencing parameters is suggestive of probable breakage of the instrument.
Just occurred if the conventional rule of a sample size of thirty be applied considering this as failure time analysis?
Viswa Chandu for multivariable models, a general rule of thumb is 30 per predictor/event. So with 4 predictors, I would think you'd definitely want more than 30 to obtain a saturated sample.
Hi. For a multivariate models, the ideal sample size n must be 200 and above.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Sampling Design
- Sample Size