# help with dara analysis

Announcements
#1
Hi,
I have an assignment on data analysis I would like help on.

We are given a set of data and we have to choose a statistical test to use, explain why we used that (including whether it follows parametric criteria) and sketch a graph that would be used for that data...
If you can help, I would be eternally grateful!

Question-

As part of a research project on vertebrate sampling techniques, an ecologist left 24 camera traps of different sizes at random locations in a forest in central Africa. After 24 hours she checked to see if each trap had recorded any vertebrates or not. Her null hypothesis (H0) was that there is no association between trap size and whether the trap caught vertebrates. The data is summarised below:

Small traps- 6 traps invertebrates present, 2 traps invertebrates not present
Medium traps- 5 traps invertebrates present, 3 traps invertebrates not present
Large traps- 3 traps invertebrates present, 5 traps invertebrates not present

What statistical test would be used?
Why?
What graph would be used?

I have to follow the choosing chart attached.

Thanks!
0
5 years ago
#2
as the outcome variable (DV) is dichotmous, then i think a logistic (aka logit) regression should be used - generally report on the odds of catching a vertebrae (DV) from comparing the 3 IV groups of trap sizes + see if it statistically significant etc.

no need for graphs as would be reporting odds ratios i think or you could use descriptives as furthe evidence to make it clearer or comparitive bar charts if you really wanted to

Alternatiely you coulld use other complex analysis such as survival or mortality analysis although it might be aa lil too complex and unrelative
0
#3
(Original post by mao94)
as the outcome variable (DV) is dichotmous, then i think a logistic (aka logit) regression should be used - generally report on the odds of catching a vertebrae (DV) from comparing the 3 IV groups of trap sizes + see if it statistically significant etc.

no need for graphs as would be reporting odds ratios i think or you could use descriptives as furthe evidence to make it clearer or comparitive bar charts if you really wanted to

Alternatiely you coulld use other complex analysis such as survival or mortality analysis although it might be aa lil too complex and unrelative
Sorry I didn't state... they gave us a chart of choosing the correct statistical test and if I follow that, I get kruskall-wallis test but that doesnt work with nominal data?
I have attached the choosing chart i have to follow.
0
5 years ago
#4
(Original post by cgh112)
Sorry I didn't state... they gave us a chart of choosing the correct statistical test and if I follow that, I get kruskall-wallis test but that doesnt work with nominal data?
I have attached the choosing chart i have to follow.
i thought a logistic regression since you mentioned you are assessing an association between variables with independent + dependent variabes etc

although it mentions linear regression only - logistic regression is exactly similar except that it is used when the DV is dichotmous i.e. nominal data of fail or success and ten having the 3 conditions of the IV of type of trap as dummy ariables added in to the model.

usually paramteric measures have more statistical power etc. - kruskall-wallis is non-parametric alternative to 1-way aNOVA but you would need a ordinal/interval data at least for that
0
X

new posts
Back
to top
Latest

### Oops, nobody has postedin the last few hours.

Why not re-start the conversation?

see more

### Poll

Join the discussion

Talking to current university students (18)
18.95%
Talking to peers going through the same thing (33)
34.74%
Speaking to student ambassadors from the universities (5)
5.26%
Speaking to staff members from universities (2)
2.11%
Using the personal statement builder, library or helper service (9)
9.47%
18.95%
Learning about/speaking to Student Finance England (4)
4.21%
Something else (tell us in the thread) (6)
6.32%