Cross Validated is a question and answer site for people interested in statistics, machine learning, data analysis, data mining, and data visualization. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I have data from groups of professionals with in the same medical field. They are grouped according to years in practice. Each group going up in 5 year 0-5 yrs experience 6-11 etc. Individually all were asked a set of questions from which I was only interested in one response based on my hypothesis. I'm wanting to see, if the more experienced in practice an individual is, the better they are at spotting a disease. What is the best statistical test to use when comparing the nine age groupings against each other in relation to the frequency each group has for the 1 question I zoomed in on?

share|improve this question
Why group individuals according to 5-year increments when you could simply have years of experience as a continuous variable? – Nick Adams Nov 15 '12 at 1:10

I would set up an ANOVA test, grouping the individuals by whether they got your question right or wrong. Your null hypothesis is that the means of the years of experience between the two groups is not significantly different. Then you can answer the question "Do the people that answer the question right tend to have more years experience?".

share|improve this answer
I understand the question to refer to just a single question, so the dependent variable is yes/no. ANOVA will not really work with yes/no dependent variables. Conversely, if we use the yes/no status as the _in_dependent variable and experience as the dependent one, this will require some mental calisthenics from readers of the analysis, as it would be akin to "answering this question correctly causes you to have more experience"... Although in principle I agree that a case could be made for this approach. – Stephan Kolassa Nov 15 '12 at 13:51
For the OP's edification, the case for using an ANOVA in this situation (that it is essentially a special case of regression) is made here. – Nick Adams Nov 16 '12 at 23:41

You could use a logistic regression on the years of experience and see whether the resulting coefficient is significantly different from zero.

I agree with @NickAdams that you should definitely not discretize your continuous years-of-experience variable. This is very bad practice. See, e.g., here ("Much is lost and nothing gained by a median split.") or here or here. Everything that is bad about dichotomizing continuous variables also applies to splitting continuous variables into more than two bins.

share|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.