# Help Me!!!!

#### chanel

##### New Member
I have a homework mini assignment and im finding statistics very hard!
Can somebody please help me figure out what to use and why for the following questions....

1. Estimate the proprtion of soft drinks that are diet. (Out of 40 soft drinks, 25 are diet)

Would this be 62.5% ????

2.Use an appropriate hypothesis test to answer the research question.(Question 1)

#### JohnM

##### TS Contributor
chanel said:
I have a homework mini assignment and im finding statistics very hard!
Can somebody please help me figure out what to use and why for the following questions....

1. Estimate the proprtion of soft drinks that are diet. (Out of 40 soft drinks, 25 are diet)

Would this be 62.5% ????

2.Use an appropriate hypothesis test to answer the research question.(Question 1)
1. Yes, 62.5% is correct.

2. You need to provide us with the entire question....I think the "research question" may be hiding in there.....

#### chanel

##### New Member
Question 2

Thanks John,

All it says is;

2. Use an appropriate hypothesis test (no study information required) to answer the research question above; Estimate the proportion of soft drinks that are diet.
(Out of 40 soft drinks, 25 are diet)

Would I use a hypothesis test that tests small sampkes and proportions? I dont understand hypothesis testing....

#### JohnM

##### TS Contributor
What does it say above number 1? Is there an opening paragraph or short statement?

Sorry to ask a lot of questions, but #2 references a "research question." I don't see one. All #1 asks is that you estimate something.

Hypothesis testing is about stating a situation, then collecting data to see if there is enough evidence to refute the hypothesis in favor of an alternative hypothesis.

An example- say you think that, on average, girls score higher than boys on a particular test. The "null" hyptohesis would be worded something like: Girls' and boys' average scores are equal on the test. The "alternative" would be worded like: Girls' average score is higher than boys' average score.

After collecting data on scores, you would then compute the averages for each gender. If the girls' average is "significantly" higher than the boys' average, then you would conclude that there is sufficient evidence to reject the null hypothesis in favor of the alternative. If the averages are pretty close, then you would fail to reject the null hypothesis...

#### quark

Hi chanel,

The problem you posted is probably incomplete. There should be an opening paragraph. Also, hypothesis testing has multiple steps, which part do you have difficulty with? You can post an example from your textbook, we'll help you understand.

#### chanel

##### New Member
Research question

I found the research question! Yay! It was on another slip of paper.

It says;

Are half of all the soft drinks diet?

b) Use an appropriate hypothesis test (no study info required) to answer...

#### JohnM

##### TS Contributor
You could set it up this way (known as the confidence interval approach):

Null Hypothesis: Proportion of diet soft drinks is < 0.5
Alternative Hypothesis: Proportion of diet soft drinks is >= 0.5

Then I would estimate the proportion of diet soft drinks by taking a sample, then computing the confidence limits around p (% of diet). If the lower confidence limit is >= 0.5, then you could reject the null hypothesis and conclude that at least half of all soft drinks are diet.

Another way would be a significance test, but the hypotheses would be worded the same way.

Null Hypothesis: Proportion of diet soft drinks is < 0.5
Alternative Hypothesis: Proportion of diet soft drinks is >= 0.5

The test statistic would be a z-statistic (using the normal distribution, a "close enough" approximation to the binomial distribution when p is around 0.5).

Basically you would test to see if the z statistic is large enough to be significant (i.e., is the actual % of diet soft drinks far enough away from and larger than 0.5 in order to reject the null hypothesis).

Your textbook should have an example of testing the difference of two proportions (percentages), using the normal approximation.....

Last edited: