Resources for different statistical models

[Hanis_Kadir]

Hi,

Does anyone have good resources on different statistical models, i.e. predictive model vs explanatory model vs descriptive model?

Thanks!

[f2harrell]

The introductory chapter of RMS covers the many similarities between models for estimation vs. prediction vs. hypothesis testing. I’m sure that others can tell us about more in-depth resources.

[R_cubed]

I remember this Bradley Efron paper from 2020 was helpful in clarifying the subtle distinctions:

Prediction, Estimation, and Attribution

[Hanis_Kadir]

Thank you, Prof Frank Harell.

I refer to the section 4.12 of RMS book, the different models:

1. Developing Predictive Models
2. Developing Models for Effect Estimation
3. Developing Models for Hypothesis Testing

Should univariable pre-screening be avoided for all 3 types of models?
In particular, performing a table 1 or univariable analysis, and selecting those that are significant; but at the same time, include those non-significant ones that may be considered clinically important.

I also refer to other questions here. My impression is that exclusively selecting variables that are significant, should not be done. But, if it’s coupled with variables that are not significant but are clinically important (based on prior knowledge), then is it okay?

How to determine the p value cutoff for unvariate regression analysis to be included in multivariate analysis

Hi All, I had a quick question about how to determine the cut off p value in order to include relevant variables from univariate analysis into multivariate regression analysis. For example, in my analysis- I am looking at the different clinical parameters that are associated with increased risk of mortality. So I first performed the univariate regression analysis and then variable whose p-value was <0.30 on univariate analysis, were included in the final multivariate regression analysis to dete…

Interpreting prognostic factor importance from univariate vs multivariable analysis

Some prognostic factor studies show results for both univariate and multivariable models. An example of this practice can be found in Bill-Axelson et al where they evaluate prognostic factors for prostate cancer. Table 3 shows results from univariate Cox models and an adjusted model with multiple factors [image] What is the usefulness of the univariate Cox results? It’s my understanding that univariate results don’t provide meaningful information due to confounding from other prognostic facto…

[f2harrell]

Since that is directly related to RMS chapter 4 please repost that to https://discourse.datamethods.org/t/rms-multivariable-modeling-strategies and I’ll remove it here. Thanks.