Is his the right stats to use?

B

Bulbemsaur

New Member
#1
#1
For my dissertation I need to analyse my data, my supervisor recommended chi squared but I’m not sure if i can use it with my data. I’m looking at occurrences of cetaceans between 2014 and 2017 so the data I have collected is for 4 years and I’m looking at occurrences and time effort put in so they can be compared. So they are comparable I have done averages to see how frequent the sightings are for each year (however these are not whole numbers). i can identify the actual differences with these figures but I need to check if they’re accurate and not just a chance occurrence/number. All the chi squared examples I’ve found are using whole numbers and mine are all values between 0.01 and 1.5 and this doesn’t seem to work.
Another question, if I’m looking at 3 different species throughout the years but they’re not related, so I put them in the same test? i will be comparing their differences so Im not sure.
Any advice would be greatly appreciated!

https://1drv.ms/x/s!Ap7CVJENszLq43X0G6Vhb2sK-pr0 A link to my data so you can get an idea


• To identify how the occurrences of cetaceans differ between 2014 and 2017.
• To determine whether there is a difference in the occurrences from day ferry routes and cruise ferry routes.
• To find which, if any, species occurs the most frequently.
• To determine whether there are more sightings on average on the day ferry routes or the cruise ferry routes.
These are my questions I’m answering
 
Karabiner

Karabiner

TS Contributor
#2
#2
I’m looking at occurrences of cetaceans between 2014 and 2017 so the data I have collected is for 4 years and I’m looking at occurrences and time effort put in so they can be compared.
Click to expand...
If I understand you correctly, for each year you have a value which is something
like "number of occurences per hour of search"?

Is it possible to use the original numbers (like "in 2014: 38 occurences; in 2015: 72 occurences etc.)?
You can then perform a Chi² analysis which takes into account that search time was different
in each year. Mind that Chi² compares the empirical frequencies (or proportions) with expected
frequencies/porportions (those under the null hypothesis). So for example, if you had only 2 years,
and in the first year searching time was 50 hours, and in the other year it was 150 hours, then you
would expect that also 3/4 of all occurences in your study were in the first year and 1/4 in the second,
if occurences were independent of year.

Another question, if I’m looking at 3 different species throughout the years but they’re not related, so I put them in the same test? i will be comparing their differences so Im not sure.
Click to expand...
I must admit that I do not unerstand what you mean.

With kind regards

Karabiner
 
B

Bulbemsaur

New Member
#3
#3
Karabiner said:
Is it possible to use the original numbers (like "in 2014: 38 occurences; in 2015: 72 occurences etc.)?
You can then perform a Chi² analysis which takes into account that search time was different
in each year.
Click to expand...
Thank you for your response it really helps!

Yes I do have the original values that I can put in instead. Does the test just know that my search time is different or do I need to input something other than my numbers for that?

Ignore that last question I think I was just confused
 
Karabiner

Karabiner

TS Contributor
#4
#4
You have to insert expected proportions. The expected distribution of occurences across years reflects the distribution of search time. E.g. if total search time was distributed 20% /10% / 50% / 20% between years, then total occurences should be distributed accordingly, if occurences and year are not associated.
 
You must log in or register to reply here.