I am doing a 5-point Likert scale survey using SurveyMonkey on my topic for my dissertation. 18 5-point Likert scale questions are in the survey and I am sending it out via email through SurveyMonkey. Before I can use it for my main or primary study, I am sending it out to similar groups via email that are not being used in the main study to use Cronbach's alpha on to test reliability and validity. My question is how many respondents do I need before I do Cronbach's alpha? Again 18 5-point Likert scale questions total or I could break them down into factors/factor analysis which would consist of one factor of 6 5-point Likert scale questions, one of 5 questions, and one of 7 questions. Would this affect the numbers needed by breaking the survey questions for Cronbach's alpha down into factors? Would the required number of respondents change