IQR interquartile range

Discuss statistics related things

by ste » Sat Sep 01, 2018 6:54 pm

Hello!

I am new to statistics and I was wondering if you could help me.

I have collected weight and height of 80 people, those have performed a test twice. This mean, I have two final scores for each person (160 scores).
With jamovi I got the IQR of weight and height, and I need to understand if these variables could impact the scores. This mean I have to find evidence against the null hypothesis by percentile (25th, 50th, 75th) comparing the scores by height and weight and BMI.

How do I proceed?

Thanks for your support ;)
ste
 
Posts: 3
Joined: Sat Sep 01, 2018 6:40 pm

by Ravi » Tue Sep 04, 2018 6:17 am

Hi,

I'm not completely sure what you want to do with the IQR. Could you elaborate on this sentence:
This mean I have to find evidence against the null hypothesis by percentile (25th, 50th, 75th) comparing the scores by height and weight and BMI.


Cheers,
Ravi
User avatar
Ravi
 
Posts: 140
Joined: Sat Jan 28, 2017 11:18 am

by ste » Tue Sep 04, 2018 8:19 am

Ravi wrote:Hi,

I'm not completely sure what you want to do with the IQR. Could you elaborate on this sentence:
This mean I have to find evidence against the null hypothesis by percentile (25th, 50th, 75th) comparing the scores by height and weight and BMI.


Cheers,
Ravi



Yes, sure!

80 people performed resuscitation manoeuvres on a mannequin. The mannequin elaborates some data such as mean rate (frequency) and mean depth of chest compressions.

In different publications, weight, height and BMI are characteristics able to impact the performance. This is why it's necessary to compare the results (such as mean depth and mean rate) by weight, height and BMI. Weight and height (as BMI) are grouped into IQR and than compared with scores.


You can check out this article: Contri E, Cornare S "Complete chest recoil during laypersons' CPR: Is it a matter of weight?" American Journal of Emergency Medicine, 2017
Attachments
Screen Shot 2018-09-04 at 10.17.15.png
Screen Shot 2018-09-04 at 10.17.15.png (75.23 KiB) Viewed 1862 times
ste
 
Posts: 3
Joined: Sat Sep 01, 2018 6:40 pm

by Ravi » Tue Sep 04, 2018 5:43 pm

Hmm, I wonder why you would want to split your continuous variables into 4 quartile groups and then do separate ANOVA's for each predictor. I might be missing something, but this seems more like a multiple regression to me with weight, height, and BMI as covariates.
User avatar
Ravi
 
Posts: 140
Joined: Sat Jan 28, 2017 11:18 am

by Ravi » Tue Sep 04, 2018 6:18 pm

Nonetheless, if you want to do this you have to use computed variables at the moment (we are finishing up an easier way to recode variables but it's not completely finished yet). Here's a blog post about computed variables: https://blog.jamovi.org/2017/11/28/jamovi-formulas.html.

You need to use a series of IF statements to recode the continuous variables into quartile labels as follows:
Screenshot from 2018-09-04 11-14-13.png
Screenshot from 2018-09-04 11-14-13.png (41.49 KiB) Viewed 1849 times


So you have to use the values you get from the descriptives to recode your variable. Here's the example code:
Code: Select all
IF(x1 < 4.17, 'Q1',
IF(x1 < 5.00, 'Q2',
IF(x1 < 5.67, 'Q3', 'Q4')))


Hope this will help you.
User avatar
Ravi
 
Posts: 140
Joined: Sat Jan 28, 2017 11:18 am

by ste » Tue Sep 04, 2018 6:27 pm

Ravi wrote:Nonetheless, if you want to do this you have to use computed variables at the moment (we are finishing up an easier way to recode variables but it's not completely finished yet). Here's a blog post about computed variables: https://blog.jamovi.org/2017/11/28/jamovi-formulas.html.

You need to use a series of IF statements to recode the continuous variables into quartile labels as follows:
Screenshot from 2018-09-04 11-14-13.png


So you have to use the values you get from the descriptives to recode your variable. Here's the example code:
Code: Select all
IF(x1 < 4.17, 'Q1',
IF(x1 < 5.00, 'Q2',
IF(x1 < 5.67, 'Q3', 'Q4')))


Hope this will help you.



I chose jamovi over R for this reason :thinking:

anyway, I'll try! thanks! ;)
ste
 
Posts: 3
Joined: Sat Sep 01, 2018 6:40 pm


Return to Statistics

cron