Help with SPSS

[a***@gmail.com]

Hi Everyone,

I'm struggling with SPSS and would really appreciate a helping hand.

So...

My data consists of results from a test where I had
- 16x test blocks
- 4 points of interest on each test block through a test frame. ABCD.
- There are 6 different ways I have measured each point. 3D position, height, angle etc

I then measured the difference from the plan to the test position in each of the 4 spots in each of the 16 blocks

I repeated test block 1 10x with the same method to test the precision

I then repeated test block 1 10x with a different method to test the trueness.

I also have means of similar tests.

Initially I did the comparisons below of ALL the data in each test group. I used one-sample t-tests to compare to similar test data. I used a paired sample test to compare the main test data to the repeat and the repeat to the second repeat of block 1.

I've then been told that since the 4 positions are connected by a frame they are not independent so I have added a nominal value category to show which data entry relates to which position. And as such I have to adjust the p value (bonferone correction) by x6 there are 6 comparisons and if still less than 0.05 then the result is sig.
I understand that but correcting the analysis to fit this im finding difficult.

So I need to use SPSS to calculate the mean and standard deviation for each position. I'm competent with doing that with descriptive statistics but would really like help to;

1) Create a box plot where each of the variables are listed on the X axis with the 4x points of interest next to each other for each variable.

2) Calculate the systematic offset for the 4 positions ABCD

3) Compare a one sample t-test to the similar test means but differentiate the comparison by the position category OR exclude 3/4 of the position results.

4) Use the systematic error to compare the overall results of the main test data to the repeat tests.

5) Combine the separated test analysis by position to form an overall analysis.



Hopefully the above makes sense and you can help me please!

Thanks

[Ki]

Post by a***@gmail.com
Hi Everyone,
I'm struggling with SPSS and would really appreciate a helping hand.
So...
My data consists of results from a test where I had
- 16x test blocks
- 4 points of interest on each test block through a test frame. ABCD.
- There are 6 different ways I have measured each point. 3D position, height, angle etc
I then measured the difference from the plan to the test position in each of the 4 spots in each of the 16 blocks
I repeated test block 1 10x with the same method to test the precision
I then repeated test block 1 10x with a different method to test the trueness.
I also have means of similar tests.
Initially I did the comparisons below of ALL the data in each test group. I used one-sample t-tests to compare to similar test data. I used a paired sample test to compare the main test data to the repeat and the repeat to the second repeat of block 1.
I've then been told that since the 4 positions are connected by a frame they are not independent so I have added a nominal value category to show which data entry relates to which position. And as such I have to adjust the p value (bonferone correction) by x6 there are 6 comparisons and if still less than 0.05 then the result is sig.
I understand that but correcting the analysis to fit this im finding difficult.
So I need to use SPSS to calculate the mean and standard deviation for each position. I'm competent with doing that with descriptive statistics but would really like help to;
1) Create a box plot where each of the variables are listed on the X axis with the 4x points of interest next to each other for each variable.
2) Calculate the systematic offset for the 4 positions ABCD
3) Compare a one sample t-test to the similar test means but differentiate the comparison by the position category OR exclude 3/4 of the position results.
4) Use the systematic error to compare the overall results of the main test data to the repeat tests.
5) Combine the separated test analysis by position to form an overall analysis.
Hopefully the above makes sense and you can help me please!
Thanks

It will really help IF you can give an example of your data structure, and explain your measurements using that structure. It is hard to follow your description.

I imagine that your data is like:

B1_A1 B1_A2 B1_A3 B1_A4 B1_A5 B1_A6 B1_B1 ..

Where, "Bn_" indicate your blocks, "Am" indicates your "m" measures in the "A" frame (so, you have 4 frames and for each frame 6 measures) etc...

This is what I can "understand" reading your post, but there are quite a few jumps that make hard to make sense. Also, you should give us some sense of your data characteristics: like numerical with XX decimals, etc.

If you can post this data example, and re-explain what you have, you may get a better response.
Thanks.

[Rich Ulrich]

Post by a***@gmail.com
Hi Everyone,
I'm struggling with SPSS and would really appreciate a helping hand.
So...

I agree with Ki that this presentation is not clear enough.
I will show how much I understand of it.

Post by a***@gmail.com
My data consists of results from a test where I had
- 16x test blocks

16 IDs

Post by a***@gmail.com
- 4 points of interest on each test block through a test frame. ABCD.

4 Points, A-D, presumably mechanically connected. Are these always
the same?

Post by a***@gmail.com
- There are 6 different ways I have measured each point. 3D position, height, angle etc

6 variables, including 3 for "position"

Post by a***@gmail.com
I then measured the difference from the plan to the test position in each of the 4 spots in each of the 16 blocks

Comparing "position" variables to "plan" where "plan" is not
previously mentioned. It suggests to me that you intended to
achieve a certain position; and discrepancies replace the raw
numbers for position.

Post by a***@gmail.com
I repeated test block 1 10x with the same method to test the precision

This could be "reliability" of use of the yardstick, or it could be
"reliability" of a method that includes moving the "frame" to achieve
a position.

Post by a***@gmail.com
I then repeated test block 1 10x with a different method to test the trueness.

The term I use is "validity".

Post by a***@gmail.com
I also have means of similar tests.

I don't know what "similar tests" points to.

Post by a***@gmail.com
Initially I did the comparisons below of ALL the data in each test group. I used one-sample t-tests to compare to similar test data. I used a paired sample test to compare the main test data to the repeat and the repeat to the second repeat of block 1.

Are you distinguishing ID=1 from the rest, as "test groups"?

If ID=1 is subjected to paired-sample testing, that must be (I think)
across measurements 1-10, and that suggests that you expect systematic
bias as re-measures are taken. That's a bad characteristic for a
measurement. Otherwise, a two-sample test would be the test for
systematic difference in level.

Post by a***@gmail.com
I've then been told that since the 4 positions are connected by a frame they are not independent so I have added a nominal value category to show which data entry relates to which position. And as such I have to adjust the p value (bonferone correction) by x6 there are 6 comparisons and if still less than 0.05 then the result is sig.

If you label the whole thing as "exploratory", there is no need for
corrections.

Or. If you figure that one of the 4x6 measures /ought/ to be the most
sensitive, you should should focus your hypothesis-testing and
conclusions on that single measurement.

Post by a***@gmail.com
I understand that but correcting the analysis to fit this im finding difficult.
So I need to use SPSS to calculate the mean and standard deviation for each position. I'm competent with doing that with descriptive statistics but would really like help to;
1) Create a box plot where each of the variables are listed on the X axis with the 4x points of interest next to each other for each variable.
2) Calculate the systematic offset for the 4 positions ABCD
3) Compare a one sample t-test to the similar test means but differentiate the comparison by the position category OR exclude 3/4 of the position results.
4) Use the systematic error to compare the overall results of the main test data to the repeat tests.
5) Combine the separated test analysis by position to form an overall analysis.
Hopefully the above makes sense and you can help me please!
Thanks