These are thoughts that I have included here from AP Statistics Gurus that I admire and respect..

1.  When answering a question about a test, for instance a two sample confidence interval, make sure you are specific. 

I think reading the rubrics it's clear that something like "This is a two-sample confidence interval for the difference of means with

independent random samples with large samples (n's>30).  Using my calculator I find the interval (a, b), df = whatever.  I am 95%

confident that the true difference inmeans [context here] is between a and b." is perfectly OK.  

P. Flanagan-Hyde, Phoenix Country Day School

2. When looking at a boxplot to help determine normality, what are we looking for when it comes to outliers or skewness, long whiskers or a median that isn't in the middle of the box?

Hopefully no outliers AND ... no really long whisker at one end with no (or dinky) whisker at the other end.  Better yet, look at some REAL frequency display ... dotplot, histogram, tally, etc. ...

D. Roberts, PSU.edu

3.  When doing a chi-square test, do they have to show the individual components that make up the sum?  

I tell my students to show the computations (of the expected count, and the component of the chi-square) for just ONE cell in

order to give evidence to the reader that they know how to do the computations, and then let their calculator take care of the

rest. 

J. Zucker, Gunn High School

4.  Confidence Interval or Hypothesis test??

"Doing inference" seems to me to be a vague term that could include testing a hypothesis or estimating a parameter.
To my way of thinking, render hypothesis testing unto testing a hypothesis, and confidence intervals to estimating parameters.
C. Olsen, Cedar Rapids IA

estimation --> confidence interval (aka interval estimation)
    evidence -->  hypothesis test (aka test of significance)

D. Bee,

5. On proving "normal"

 Students must SHOW that they have checked the conditions. If they refer to a graph (histogram, residuals plot, etc.), they must sketch it - nothing fancy, just enough to let the Reader know what they saw. Similarly it's not enough to say "np and nq were greater than 10) - they must show "120*0.70 = 84 and 120*0.30 = 36; both are at least 10". For chi-sq, saying all expected counts are at least 5 doesn't earn credit unless accompanied by those expected counts. And so on. Tell your students they must show enough detail to convince the Reader that they actually did check the conditions, not just write down a memorized list of catchphrases.
D. Bock