Using Spearman's rho to analyse results of an experiment I have just gathered some data for an experiment I am doing for my research. There are two groups in my experiment. Each of them is shown a different prompt, and subsequently asked two questions. The first question has a yes'no answer, the second a yes/perhaps/no answer. I am trying to see if seeing Prompt #1 can induce people to choose 'Yes' in the follow-up questions.
I am working in SPSS. The data that I have includes: the size of both groups, and the number of respondents in each group who picked each response. Prompt 1 is coded as 1.00, Prompt 2 is coded as 0. Similarly, Yes is coded as 1.00 and No is coded as 0.00. I thought about using Spearman's rho to look for a correlation between each prompt and people's responses. Is this reasonable? 
My apologies if this is a stupid question, but I have pretty much had to 'learn' stats overnight, so I could really use some help.
 A: Spearman's rho is note a good idea here, it is more for looking at the correlation of two continuous variables. 
Assuming that assignment to the different prompts was done in a randomized manner, your example would seem like a classic case for using logistic regression (or if there is no further things to be taken into account Fisher's exact test, which is essentially exact conditional logistic regression). Logistic regression provides you both an effect measure (the odds ratio), measures of uncertainty around the effect measure (e.g. standard error, confidence interval) and a hypothesis test.
If the assignment was not randomized (e.g. by self-selection), things become much more complicated and you may have to look e.g. into propensity score methods (e.g. logistic regression stratified by deciles of propensity scores).
A: I wouldn't use Spearman's rho as it assumes ordinality, that is, you can order your data in increasing order. Given that you converted your data to 0/1 you could do it, but it doesn't really make sense, as yes in not "more" than no.
To me this sounds like data that would be best analysed with a McNemar test given that the data is paired. You can then look at the contingency table to identify the direction of the effect. Read more on a McNemar test here 
