
1. Re: Formula for R2 and P values for Trend Lines
lenaic.ridinger Nov 4, 2016 8:50 AM (in response to Jordi Buch)The Force wasn't with you
Let me call the Tableau Jedis one more time!
Any Guardian able to help Jordi Buch with his formula?

2. Re: Formula for R2 and P values for Trend Lines
Patrick A Van Der Hyde Dec 14, 2016 1:15 PM (in response to lenaic.ridinger)Bora may chime in but he mentioned that correlation coeffiient's can be done with LOD calcs. R makes things easier but it sounds possible in Tableau.

3. Re: Formula for R2 and P values for Trend Lines
Bora Beran Dec 15, 2016 1:23 AM (in response to Jordi Buch)2 of 2 people found this helpful 
4. Re: Formula for R2 and P values for Trend Lines
Jordi Buch Dec 15, 2016 1:27 AM (in response to Bora Beran)Hi Bora,
You left me speechless... Exactly what I needed!!!!!
Thank you so much!!!

5. Re: Formula for R2 and P values for Trend Lines
Heather Lewis Mar 22, 2017 11:07 AM (in response to Bora Beran)While this is incredibly helpful  I am stuck on the Critical Values (per Category) piece. I have a few correlations where the p value < .05 and my "Is Significant" calculated field is saying "False". Do you think this could be due to the Critical Values (per Category) calculated field only includes rows 122 from the ttable? Plus, how did you calculate the numbers such as ".000024" in the Critical Values (per Category) calculated field?
I am very close to my goal though.
Many thanks!
Heather

6. Re: Formula for R2 and P values for Trend Lines
Bora Beran Mar 22, 2017 12:36 PM (in response to Heather Lewis)Hi Heather,
Those coefficients are used to interpolate between points in the table so the slope and the intercept used for interpolation. Can you share how many points you have in the pane and what the critical value calculation returning is?
Thanks,
Bora

7. Re: Formula for R2 and P values for Trend Lines
Heather Lewis Mar 22, 2017 1:25 PM (in response to Bora Beran)Hi Bora,
Thank you so much for the quick response.
Attached is the .csv file that shows my Critical Values (per Category) calculated field. In my case "(per Category)" translates to "(per Question ID)" for my data. Did I do this correctly?
For your example there are 3 categories, I have about 50+ Question IDs.
The scatter plots are the students' responses to a survey item correlated with their first fall term GPA. Each dot is a student. Each scatter plot has about 300 students represented.
The "Is Significant? (per Question ID)" calculated field is > and is returning a "TRUE" or "FALSE".
Now I will say this  the series of calculated fields are working for 3 scatter plots. They are the only scatter plots with a pvalue of <.0001 (see jpeg that is attached).
Your feedback is greatly appreciated!
Heather

tabulaTTable.csv.zip 819 bytes

8. Re: Formula for R2 and P values for Trend Lines
Heather Lewis Mar 22, 2017 1:49 PM (in response to Bora Beran)Hi Bora  also attached are the slope, intercept, tstatistic, and critical value for each Question ID. Also, I have a very tall data set given that this is survey data and I need to work with both Labels and Values of the survey responses. You think that might be affecting the SUMS within the calculated fields?

9. Re: Formula for R2 and P values for Trend Lines
Heather Lewis Mar 24, 2017 5:57 AM (in response to Bora Beran)I tinkered all yesterday with the calculated fields. I think part of the problem may be the fact that my X variable (Values) is coming from a tall data set and my Y variable (Term GPA) is coming from a wide data set. Is there some SQL that I can include in the calculated fields? Some sort of COUNTD for Values? Really want to get this accomplished before my meeting with leadership next week to present this data. If necessary, I can deidentify the Student IDs and send you twbx?
Many thanks!
Heather

10. Re: Formula for R2 and P values for Trend Lines
Toby Erkson Mar 24, 2017 6:01 AM (in response to Heather Lewis)Heather,
It's ALWAYS best to include a .twbx when asking a question. Yes, please anonymize your data and attach it to your post. This way others can pop in and help.

11. Re: Formula for R2 and P values for Trend Lines
Heather Lewis Nov 12, 2018 5:57 AM (in response to Bora Beran)Attached is the twbx of my problem. Please help!!!!
< author removed object >
Many many many thanks!
Heather

12. Re: Formula for R2 and P values for Trend Lines
Bora Beran Mar 24, 2017 1:08 PM (in response to Heather Lewis)Hi Heather,
This is a twotailed test so to mimic the comparison the table is doing given the significance level (the reference ttable link is provided in the calculated field so you can see what significance the test assumes by looking at the table and corresponding upper tail probability) you have to multiply your pvalue by two.
0.02514 * 2 = 0.0528
0.0528 > 0.05
If you would like this to be considered significant you can change the confidence level the test is running.
Thanks,
Bora

13. Re: Formula for R2 and P values for Trend Lines
Heather Lewis Mar 24, 2017 2:14 PM (in response to Bora Beran)I'm sorry Bora Beran ! I'm trying to follow the table at this link https://www.stat.tamu.edu/~lzhou/stat302/TTable.pdf , but where did the .02514 come from? It's not in the table? If anyone can please help me with resolving my pvalue issue I would GREATLY appreciate it!!

14. Re: Formula for R2 and P values for Trend Lines
Bora Beran Mar 24, 2017 2:36 PM (in response to Heather Lewis)That is the value you circled in your screenshot as the pvalue in the tooltip for the trendline. The table is for tcritical values. The relevant piece you will see in the table is that 0.95 is associated with 0.025 not 0.05.