microsoft excel help

[neuro nephrons]

Advertisement - Members don't see this ad

I am currently working on a study that is looking at weight gain in college. I am working on the data analysis at this point, and I need some help running some correlations. For example, I am trying to see if weight gain is related to answers to questions on a survey. Example: a question that asks how often participants read the nutritional labels on the products they eat? Say I have 19 participants who say they never look at the nutritional facts, is it possible to run a correlation with the amount of weight they've gained? Is that possible?


I'm trying to relate specific questions to overall weight gain, but I don't know how to formally do that.


Any help would be excellent!

 

[911 Turbo]

idk, run the statistical tests and determine the confidence interval blah blah blah blah?



shouldnt you be asking your PI OP?

 

[EnginrTheFuture]

I am not sure I understand your question. Are you saying is it statistically possible to find the strength of a correlation between groups of thresholded ( never read, sometimes read, read often, always read) responses and weight gained? The answer is yes. Plot the data and see if there is a basic correlation. If there is, and it fits a functional form (linear, power, etc.) find the best fit line, power, etc function and it's coefficient of determination with respect to the actual data. The higher the R^2 value the better the correlation.

There are many ways of doing this but this way will be the easiest since you are probably new to these concepts and programs.

 

[neuro nephrons]

idk, run the statistical tests and determine the confidence interval blah blah blah blah?



shouldnt you be asking your PI OP?

I'm meeting with her next week. (I'm still on winter break in a different state). She wanted me to try it on my own first before we go over it.

I am not sure I understand your question. Are you saying is it statistically possible to find the strength of a correlation between groups of thresholded ( never read, sometimes read, read often, always read) responses and weight gained? The answer is yes. Plot the data and see if there is a basic correlation. If there is, and it fits a functional form (linear, power, etc.) find the best fit line, power, etc function and it's coefficient of determination with respect to the actual data. The higher the R^2 value the better the correlation.

There are many ways of doing this but this way will be the easiest since you are probably new to these concepts and programs.

This is what my next step was going to be. Thank you!

 

[kami333]

In addition to running some sort of regression analysis, I'd also run a paired t-test on each sample group to compare with baseline weight to see if weight gain is statistically significant.

 

[DAPI]

I wouldn't recommend doing this in excel, but...

You need to run a paired t-test to make sure there is significant weight gain (like kami333 said).

If there is, then you could take the average weight gain for each group (ie: did not read label, sometimes read label, always read label, etc) and use those values to run an ANOVA. This would tell you which groups do or do not have a significant difference in weight gain.

Seems like a reasonable place to start, but probably not the most robust test out there.

 

[neuro nephrons]

In addition to running some sort of regression analysis, I'd also run a paired t-test on each sample group to compare with baseline weight to see if weight gain is statistically significant.

Way ahead of you, weight gain is significant.