Assessment support

The P-value gives an indication of the Item difficulty. It shows the percentage of students who answered an item correctly.
A low P-value (e.g. below 0.30) indicates that for a many students this was a difficult item, a high P-value (e.g. above 0.90) indicates that almost all students found this question easy to answer. 

For open questions
The formula for calculating is:   P-Value = average score of the class / maximum possible score 
Example: Maximum score for question A was 20 points. Students on average got 14 points (sum of all scores / number of participants)
P-value = 14 / 20 = 0.70 

Optimal P-value for open questions
Theoretically a question that discriminates optimally, has a P-value of 0.50. It is not too easy and not too difficult. In practice you want to construct a test with questions that are a bit easier (e.g. 0.80 or higher), questions which will be a bit more difficult (e.g. below 0.40), and most questions will be in between and in the range around 0.50.  The exact distribution is difficult to indicate, but could be 20-60-20, for example.

For closed questions
The formula for calculating now is:  P-Value = total who answered the item correctly / total number of participants 
Example: 30 students gave a corret answer. Totale number of stduents: 50.
P-value = 30/50 = 0.60 

Optimal P-value for closed questions
Students may guess and you want to take that into account. Optimal P-value: slightly higher than midway between chance (1 divided by the number of answer options) and a perfect score (1) for the item. Formula: (1 + (1 / nr. alt.)) / 2 
Example: Say a MC item with 5 alternatives. Score: 0 or 1. Guessing chance = 1/5 = 0.20 (20%) Optimal P-value: (1 + 0.20) / 2 = 0.60. 
Optimal values:  2 answer options: 0.75    ||     3 answer options: 0.67     ||   4 answer options: 0.62

This applies only for closed questions.
When working with closed questions, beside looking at the P-value, it is also interesting to look at the A-value; the values for the aternatives or the incorrect answers. The alternatives are alos known as “distractors”. You can execute a distractor analysis.

You look at the number and percentage of students that choose one of the distractors.  
A-value =  total who choose a certain distractor / total of participants taking the test 

The distractors should be plausible but incorrect. How to interpret the values? Look at the outliers.
Look at the very low percentages (e.g.2%). This is not a good distractor. It needs revision or replacement for a next time (else it makes a question easier). 
Look at the values higher than the P-value. This can indicate that there may be something wrong with the question. E.g. are the answers mutually exclusive? Miskeyed? Or this points to a misconception shared by many students (evaluative information).

Item-total correlation (Rit)  [a.k.a. Point Biserial correlation]. 

The Rit-value provides an indication of the correlation between the item and the total score. This Item Discrimination Index shows the extent to which students with high overall test results also got a certain question correct. This is what you would expect. A high Rit value will indicate this. 
It becomes interesting when this will not be the case. When students who didn't perform well in general, answered a certain question correctly and those who did do well on the test overall, choose a incorrect answer. Then this is a signal that something can be the matter with this question.  

The RiT ranges from -1.00 to 1.00. 
As a general rule: the value should be higher than 0.20 

+ ) Positive (high) indicates that those scoring high on the total exam answered a test item correctly more frequently than low-scoring students.
- ) Negative indicates low scoring students on the total test did better on a test item than high-scoring students.

The Rir-value in general is used when a test has not many questions. It has the same function, but doesn't take the results of the question for which you calculate the value into account for the total. 

For thise interested, this is the Rit formula:   

For those who want to learn more, the following sites are informative:  
> Use point-biserial to discriminate high and low performers | GradeHub 
> To see how it is calculated: Item Statistics for Classroom Assessments Part 2: Computing P-Values and Point-Biserial Correlations - Educational Data Systems (eddata.com) 

Cronbach's alpha is a measure of internal consistency. It shows the extent to which the questions of a test provide consistent information regarding the students’ mastery of the course content. 
It takes a lot of calculation, but if you administer the test ivia Remindo or Contest, you will get the value automatically. 

The range is from 0 to 1.     As a rule of thumb:  < 0.70  bad / moderate    > 0.80  sufficient / good

The value will be influenced by homo/heterogeneous of items (topics) and group and the number of participants and items.
It won’t be reliable to use for small or very heterogeneous groups or a test with just a few questions or content that is not expected to be coherent (like for instance two subject-contents combined in one test).  
So yes it can be used to get an impression of the test in total. But be careful with the interpretation. 

For those who want to read more about this:
> In this guide the formula is explained: An Instructor’s Guide to Understanding Test Reliability. Craig S. Wells &  James A. Wollack, 2003.