# Thread: What statistical test should I use to analyse this data?

1. ## What statistical test should I use to analyse this data?

Sorry to bother you guys again, I'm just having some real trouble trying to tackle this data.

The experiment I carried out was testing people's response to a oral glucose load when their bodies are in a fasting state and also to see if the person's BMI (body mass index) affected this response. To do this, I used the Oral Glucose Tolerance Test.

This test involved starving a bunch of volunteers overnight and giving them 50g of glucose in solution to be drank all at once. Measurements of their blood glucose concentration was just before the drink at 0 minutes and then 30, 60, 90 and 120 minutes after the drink. The plot of the group means are attached...

Rather than assess whether a person's BMI value has any effect on the value of their peak glucose, I thought it was more appropriate to test if BMI had any affect on the rate of increase and the rate of decrease in blood glucose.

Here is the processed data of: subject's BMI, rate of blood glucose increase (to peak), and the rate of blood glucose decrease (using glucose concentration 30 minutes after peak) <-- in this order.

Code:
19.20	0.07222223	0.02666667
20.30	0.09333333	0.090000
21.30	0.2466667	0.040000
20.40	0.1416667	0.003333333
19.60	0.2766667	0.020000
18.44	0.2633333	0.05666667
23.22	0.3566667	0.150000
22.64	0.210000	0.060000
20.25	0.310000	0.010000
16.93	0.290000	0.01666667
24.60	0.2433333	0.050000
19.33	0.1766667	0.1133333
24.00	0.290000	0.02333333
I'm not exactly sure HOW to ask you what I want to assess but its something like this...

What statistical test should I use to test whether BMI has an effect on a person's rate of increase AND decrease?
ie. if someone has a high BMI is it likely that their rate of increase will be very high and their rate of decrease (of blood glucose conc) will be slow?

Can I do this with a single test?

Thanks guys!

2. Originally Posted by anthmoo
Sorry to bother you guys again, I'm just having some real trouble trying to tackle this data.

The experiment I carried out was testing people's response to a oral glucose load when their bodies are in a fasting state and also to see if the person's BMI (body mass index) affected this response. To do this, I used the Oral Glucose Tolerance Test.

This test involved starving a bunch of volunteers overnight and giving them 50g of glucose in solution to be drank all at once. Measurements of their blood glucose concentration was just before the drink at 0 minutes and then 30, 60, 90 and 120 minutes after the drink. The plot of the group means are attached...

Rather than assess whether a person's BMI value has any effect on the value of their peak glucose, I thought it was more appropriate to test if BMI had any affect on the rate of increase and the rate of decrease in blood glucose.

Here is the processed data of: subject's BMI, rate of blood glucose increase (to peak), and the rate of blood glucose decrease (using glucose concentration 30 minutes after peak) <-- in this order.

Code:
19.20    0.07222223    0.02666667
20.30    0.09333333    0.090000
21.30    0.2466667    0.040000
20.40    0.1416667    0.003333333
19.60    0.2766667    0.020000
18.44    0.2633333    0.05666667
23.22    0.3566667    0.150000
22.64    0.210000    0.060000
20.25    0.310000    0.010000
16.93    0.290000    0.01666667
24.60    0.2433333    0.050000
19.33    0.1766667    0.1133333
24.00    0.290000    0.02333333
I'm not exactly sure HOW to ask you what I want to assess but its something like this...

What statistical test should I use to test whether BMI has an effect on a person's rate of increase AND decrease?
ie. if someone has a high BMI is it likely that their rate of increase will be very high and their rate of decrease (of blood glucose conc) will be slow?

Can I do this with a single test?

Thanks guys!
First you need to decide what the null hypothesis and alternative hypothesis are. What tests have you been taught for comparing data sets? What are the criteria for their use?

You are advised to go back to your class notes and textbook to review hypothesis testing.

3. Originally Posted by mr fantastic
First you need to decide what the null hypothesis and alternative hypothesis are. What tests have you been taught for comparing data sets? What are the criteria for their use?

You are advised to go back to your class notes and textbook to review hypothesis testing.
We have been taught a lot of different tests over the 2 years at university (this is a Biomedical Sciences course) but as this forms part of the content of the terminal assessment, it basically means we have to choose our own. I'm really not sure which one to choose though as I'm really not great at stats and the fact that the rate of increase and decrease are from the same individuals in the same data set!

As far as a hypothesis is concerned...would this work?

As BMI increases, the rate of glucose concentration rising increases and the rate of glucose concentration falling decreases.

Is there a test out there to analyse both at the same time?