Model Selection Tool

Disclaimer

The model selection tool provided below should guide you in selecting a statistical model that may be relevant to your research question. This tool is purely for illustrative purposes, Minerva statistical consulting LTD does not accept any kind of responsibility for any actions and consequences (including damages) deriving from its use and misuse. In particular, the tool is provided for your use at your own risk and “as is” without warranty of any kind.


*Choose one of three options of the given questions and submit.

Question 1: Is your dependent variable Qualitative (e.g., sex /marital status/ ethnicity) or Quantitative (e.g., salary in US $ / no. of years of education / weight in kg) or Count (e.g., no. of meals in a day/ no. of crimes in a month in NYC/ no. of doctor visits in a year)?





Question 2: Are your independent variables Qualitative or Quantitative or Mix?



What best describes your explanatory variable(s)?

Choose

(please find also below examples of the type of study you might be conducting)

One qualitative variable (factor) with two levels, e.g., effect of sex (male/female) on salary in US $.

One qualitative variable with multiple (j) levels, e.g., effect of ivy league attendance (Harvard, Yale, Brown, , etc) on salary in US $.

Several qualitative variables with several levels, e.g., effect of ivy league attendance (Harvard, Yale, Brown, Columbia, etc) and degree type (Undergraduate, Master, PhD) on salary in US $.


Answer

One quantitative variable, e.g., effects of no. years of work experience on salary in US $

Several quantitative variables, e.g., effects of no. years of work experience/ age / no. years of schooling on salary in US $


Answer

ANCOVA, e.g., effects of no. years of work experience and marital status on salary in US $.

Logistic regression (Binomial, Ordinal or Multinomial), e.g., effect of drug administration (in mg) and sex on patient recovery.

Log-linear regression (Poisson), e.g., number of weekly deaths due to “hit and runs” as a function of socioeconomic characteristics (sex, age, marital status, no. years of education, etc).