analysis of composite scores I've created a task comprising 24 task items. Each item asks a 'right/wrong' type binary answer. For data analysis, I calculated the percent of the right answers (e.g. 20 out of 24) and treated the percent of the right answer as a continuous variable for data analysis. 
But my journal reviewer says that I should use 'logistic analysis' because the nature of my data is 'binary'. As far as I understand, we use logistic analysis when the outcome variable is 'binary'. In my case, I calculated the composite scores of the task..so I'm not certain that I should use 'logistic analysis' in this case. Maybe I'm wrong. Please help me with this issue. 
 A: You can use binomial logistic regression for your data if you define your dependent variable as the number of right answers out of 24 (and assume that the answers given by a subject to each of the 24 task items are independent of each other). 
This will enable you to model the probability of a correct answer for any task item as a function of the independent variable(s) of interest. 
See, for example, https://www.theanalysisfactor.com/when-to-use-logistic-regression-for-percentages-and-counts/. 
If you only had one task item in your study, then you would use binary logistic regression - but you have 24, so your reviewer's suggestion is only partly correct. 
Addendum:
If you use R to fit a binomial logistic regression, you first need to set up your data to include the following variables: 


*

*Right: Number of correct answers out of 24 for a study subject;

*Wrong: Number of incorrect answers out of 24 for a study subject.


(Note that Wrong = 24 - Right.)
Then you can use the following R syntax to fit a binomial logistic regression model with a logit link and a single predictor variable X, say:
model <- glm(formula = cbind(Right, Wrong) ~ X, 
             family = binomial(link="logit"), 
             data = yourdataset)

Notice the special formulation of the model that allows you to specify both the number of right answers and the number of wrong answers.
With binomial logistic dispersion you may need to worry about overdispersion. 
