QNT 275 Week 4 Practice: Week 4 Discussion

Review the Discussion FAQs Module.

Reference the Week 3 Case Study.

Respond to one of the following:

• Regarding requirement #2: What does the mean and standard deviation mean in terms of expected sales?
• Regarding requirement #2: How can this information be used to benefit the business?
• Regarding requirement #3: Can you conclude whether the data is normally distributed?
• Regarding requirement #4: What can be interpreted by that percentage?
• One-sided test vs. two-sided test
• Hi Class,
• Let me talk a little bit about one-sided and two-sided tests (note I use “side” and “tail” interchangeably sometimes – they mean the same thing).   Whether the test or hypothesis should be one-sided (left or right) or two-sided depend on what we are interested in, i.e the research question.  It is against the statistical principal if one looks (peeks) at the data and then determine the hypothesis, or the test statistics.
• For example, we are interested in if replacement with new appliances in the house can increase the house price (this is the research question), then the hypotheses are:
• H0:  Replacement of new appliances will not increase house selling price
•         Or price increment<=0
• H1:  Replacement of new appliances will increase house selling price
•         Or price increment >0
• This is a right-sided test, because in our research question, we don’t think (or care) that replacement of new appliances would have negative impact on the house price.
• For another example, we are interested in seeing if boys and girls have similar performance in the spelling test in a school.  Some people might think girls are better, some might think the other way.  So it’s better to have a two-sided test:
• H0:  boys’ average score = girls’ average score
• H1:  boys’ average score <> girls’ average score
• As a result we found out girls did better, but to plan a study in the beginning, it’s probably more appropriate to use a two-sided hypothesis.
• How do we know it’s one-sided or two sided, the cue is in the ALTERNATIVE.
• If there is a “>” sign, it’s one-sided and right-tailed test
• If there is a “<” sign, it’s one-sided and left-tailed test
• If there is an unequal sign “<>”, it’s two-sided.

Hypothesis Testing

Class, this week, we are jumping into one of the two things the inferential statistics is about :  hypothesis testing.  (the other one is estimation)

There are some new terms that we need to get familiar with.  Let’s start with null and alternative hypotheses:

The null and alternative hypotheses put the research/business question in a form of statement so that they can be either “rejected” or “not rejected” with a statistical hypothesis testing.  The null and the alternative

should be mutually exclusive.  Sometimes people get confused about which statement they should put in the null hypothesis (H0) and the “opposite” in the alternative hypothesis (H1).  In researches and business, it is the common practice that we put what we hope to prove/see in the alternative hypothesis statement.  For example:

If the research question is “Is the drug effective?”  then we use

H0:  The drug is not effective

H1: The drug is effective

“Would advertising increase retail sales?” then

H0: Advertising does not increase retail sales

H1: Advertising increases retail sales

“Does the new teaching method improve Spanish score in students?”

Then

H0:  The new method does not improve Spanish score in students.

H1:  The new method improves Spanish score in students

It’s sort of like we would like to look for alternatives – “Change is always good” ?

Class, please feel free to chime in your thoughts!