# Hypothesis Test "Statistics"

#### killivolt

Joined Jan 10, 2010
805
My Daughter wants to check her answers to this question. She believes it is 0.05

Thank you,

kv

#### Attachments

• 205.4 KB Views: 11

#### Papabravo

Joined Feb 24, 2006
17,570
My Daughter wants to check her answers to this question. She believes it is 0.05

Thank you,

kv
I don't think so. The p-value represents the chance of the null hypothesis being wrong. I would say that a p-value of 1 represents a certainty that the null hypothesis is wrong.

#### killivolt

Joined Jan 10, 2010
805
I don't think so. The p-value represents the chance of the null hypothesis being wrong. I would say that a p-value of 1 represents a certainty that the null hypothesis is wrong.
"the null hypothesis is a general statement or default position that there is nothing new happening, like there is no association among groups, or no relationship between two measured phenomena"

This didn't make any sense to me, your answer clarified it for me.

Thank you,
kv

#### Papabravo

Joined Feb 24, 2006
17,570
"the null hypothesis is a general statement or default position that there is nothing new happening, like there is no association among groups, or no relationship between two measured phenomena"

This didn't make any sense to me, your answer clarified it for me.

Thank you,
kv
Right. When you do a statistical test it does not prove the proposition which constitutes the "null" hypothesis. It says that, at 0.05 level for example, the chances of the result being wrong are less than 5% or one chance in 20. with p=1 the chances that the "null" is wrong are certain. With a result greater than the confidence level you are looking for you would reject the "null".

#### jpanhalt

Joined Jan 18, 2008
11,088
Whatever the "null hypothesis" is, the p-value represents the probability that the result is different from the null hypothesis (e.g., no difference) merely by chance.

So, I vote for p=0.001

From another perspective (as a test taker), the choices are 0.95, 0.50, 1 (significant figures?), 0.05, and 0.001. Now, you know the most gleeful answer is either large or small. On the large scale is the distractor of 1. That can't be right, unless the database is extremely large. On the other hand, a p=0.001 is possible (with a large database).

In reality, I don't think either p value would be obtained, but 1 is virtually impossible to get on any human-involving experiment. The question could have been made harder by this sequence: 0.95. 0.50. 0.98, 0.95 and 0.01.

#### killivolt

Joined Jan 10, 2010
805
Right. When you do a statistical test it does not prove the proposition which constitutes the "null" hypothesis. It says that, at 0.05 level for example, the chances of the result being wrong are less than 5% or one chance in 20. with p=1 the chances that the "null" is wrong are certain. With a result greater than the confidence level you are looking for you would reject the "null".
She had another one, she thinks it's the last answer and from what I have read it seems to me she's correct.

#### Attachments

• 260.8 KB Views: 2

#### Papabravo

Joined Feb 24, 2006
17,570
She had another one, she thinks it's the last answer and from what I have read it seems to me she's correct.
I think it is the last answer. A random boy has a 95% chance of being within the interval. The confidence interval is roughly ±2σ.

#### killivolt

Joined Jan 10, 2010
805
Whatever the "null hypothesis" is, the p-value represents the probability that the result is different from the null hypothesis (e.g., no difference) merely by chance.

So, I vote for p=0.001

From another perspective (as a test taker), the choices are 0.95, 0.50, 1 (significant figures?), 0.05, and 0.001. Now, you know the most gleeful answer is either large or small. On the large scale is the distractor of 1. That can't be right, unless the database is extremely large. On the other hand, a p=0.001 is possible (with a large database).

In reality, I don't think either p value would be obtained, but 1 is virtually impossible to get on any human-involving experiment. The question could have been made harder by this sequence: 0.95. 0.50. 0.98, 0.95 and 0.01.
She felt strongly about 1, she missed that one but still got 95/100.

I think it is the last answer. A random boy has a 95% chance of being within the interval. The confidence interval is roughly ±2σ.
She got that one correct, she chose the last one.

kv

#### killivolt

Joined Jan 10, 2010
805
She felt strongly about 1, she missed that one but still got 95/100.

She got that one correct, she chose the last one.

kv
By the way a shout out to all who have helped. @jpanhalt @Papabravo and others. She has completed stats, now she has 3 more semesters to go to get her Bachelors, she has already received her Cert as a nurse, she's working currently in a hospital Phsyc ward.

Thanks again,

kv