statistics help
Watch
Announcements
Page 1 of 1
Skip to page:
Hello, i am really struggling to figure out what statistical tests to do on my dissertation data.
my data consists of two groups which total only 34 participants. each group has many factors on a continuous scale measured about them and are all normally distributed.
i've done a multiple regression which came out with no significance. However, i wanted to expand the tests conducted. any ideas of how i can test my data?
my data consists of two groups which total only 34 participants. each group has many factors on a continuous scale measured about them and are all normally distributed.
i've done a multiple regression which came out with no significance. However, i wanted to expand the tests conducted. any ideas of how i can test my data?
0
reply
Report
#2
(Original post by jlee4)
Hello, i am really struggling to figure out what statistical tests to do on my dissertation data.
my data consists of two groups which total only 34 participants. each group has many factors on a continuous scale measured about them and are all normally distributed.
i've done a multiple regression which came out with no significance. However, i wanted to expand the tests conducted. any ideas of how i can test my data?
Hello, i am really struggling to figure out what statistical tests to do on my dissertation data.
my data consists of two groups which total only 34 participants. each group has many factors on a continuous scale measured about them and are all normally distributed.
i've done a multiple regression which came out with no significance. However, i wanted to expand the tests conducted. any ideas of how i can test my data?
0
reply
(Original post by Gregorius)
If you tell us what the scientific question is that you're trying to answer with your data, we may be able to help you!
If you tell us what the scientific question is that you're trying to answer with your data, we may be able to help you!
0
reply
Report
#4
(Original post by jlee4)
I'm trying to see if there is a relationship between the hatching success (%) of nests and the parameters of the nest e.g temp, distance to sea, depth..
I'm trying to see if there is a relationship between the hatching success (%) of nests and the parameters of the nest e.g temp, distance to sea, depth..
Next question is: how many parameters do you have recorded for each nest? For a sample size of 34, regression models will only support 2, maybe 3 at a stretch, parameters as covariates.
If you have many more than this then we're going to have to get crafty.
Technique A: For each nest parameter in turn, do a univariate regression. Select the two or three parameters that have the largest effect size that is also statistically significant (i.e. don't choose on the smallest p-value, but on the largest effect size that has p < 0.05). Stick these two or three parameters into your regression.
Technique B: Select those two or three parameters that the scientific literature suggest should be associated with the outcome into your regression model.
Technique C: enter all of your parameters into a LASSO regression and let it find the two or three parameters that survive shrinkage the longest. Put these two or three parameters into your regression equation. (This is a technique called the "relaxed lasso"

For your situation, with a small sample size, B is the best. C is next best, but is using "advanced" techniques that you might need to get assistance with. A is often used (badly) and will result in very imprecise p-values for the regression coefficients.
1
reply
Thank you!
I've got both percentage and actual number of successful eggs and clutch number.
I have several parameters but there are 2 which are most important so I'll focus on those if there's significance for them!
I'll give these a try and hopefully one will work.
Thank you again for helping! 😊
I've got both percentage and actual number of successful eggs and clutch number.
I have several parameters but there are 2 which are most important so I'll focus on those if there's significance for them!
I'll give these a try and hopefully one will work.
Thank you again for helping! 😊
0
reply
X
Page 1 of 1
Skip to page:
Quick Reply
Back
to top
to top