Hypothesis testing.

• Apr 20th 2010, 05:15 AM
MickQ
Hypothesis testing.
Anyone got any hints on how to do the questions below?

A study was conduced in Scotland in 2005 to determine if there was evidence to suggest that folic acid should be added to bread, or if there may be an alternative solution.

Data were recorded on the use of folic acid supplements in pregnancy from 233 women. The results are shown in Table 1 and displayed in Figure 1. The data shown in Figure 1 are obviously categorical, with each woman having taken folic acid periconceptually (i.e. correct use - prior to conception), incorrectly (i.e. only after conception) or not at all.

Based on data from the US it was thought that the proportions of women using folic acid supplements correctly would be about 35%, incorrectly about 25% and not at all 40%.

http://img714.imageshack.us/img714/182/statspics.jpg

1. Perform an appropriate statistical test to determine if the data from this
study followed the hypothesis from the US experience.
(3 marks)

2. State and verify any check required on the data before conducting this test.
(1 mark)

I'm just having a bit of trouble getting started, i.e which type of test to use etc.
• Apr 21st 2010, 02:18 PM
Kiki
assuming you dont need an answer anymore?

If you still need an answer let me know tho
• Apr 21st 2010, 03:18 PM
MickQ
Unfortunately, I still haven't got any further. My problem with stats always seems to be starting. Any help would be greatly appreciated.

Also, (for another question), after performing a test, I ended up with z = 20.204 with N(0,0.000339) for my (mean,S.D). Would that be my final answer?

• Apr 21st 2010, 04:09 PM
Kiki
Quote:

Originally Posted by MickQ
Also, (for another question), after performing a test, I ended up with z = 20.204 with N(0,0.000339) for my (mean,S.D). Would that be my final answer?

I suppose you have to comment the result according to what the problem is questioning you.

What critical area did you find?
• Apr 22nd 2010, 03:09 AM
MickQ
I checked it out with a few other people, and apparently that is more than enough to get the question finished. Do you have any answers for my original post?

Thanks
• Apr 22nd 2010, 04:40 AM
MickQ
For Question 1, I tried a hypothesis test using the \$\displaystyle X^2\$ distribution.

I worked out that the expected value for Correct, Incorrect & Not Used was 77.67.

Using that, I done this:

\$\displaystyle X^2 = (53-77.67)^2/77.67 + (111-77.67)^2/77.67 + (69-77.67)^2/77.67\$ giving:

\$\displaystyle X^2 = 23.1063\$ I then used \$\displaystyle X^2(2)(0.05)=5.99\$, showing that the test statistic lies in the rejection region.

However, I'm not sure if that is an appropriate method for this question, and what my Null Hypothesis should be. Am I way off, or is this a step in the right direction?
• Apr 22nd 2010, 04:57 AM
MickQ
I had the wrong expected values in my last post, as the question states:

"Based on data from the US it was thought that the proportions of women using folic acid supplements correctly would be about 35%, incorrectly about 25% and not at all 40%."

Therefore my expected values were 81.55, 58.25 and 93.2

For my N.H, I chose "Data follows the hypothesis from the US experience" and the opposite for my A.H.

My \$\displaystyle X^2\$ value lay in the rejection region, so I rejected the N.H. Also, Figure 1 and Table 1 have evidence to support this rejection.

If that's correct, I just need to do the other questions now!

EDIT: For question 2, is the required check that the expected values are greater than 5?