Login | Register
Enrol now for our new online tutoring program. Learn from the best tutors. Get amazing results. Learn more.

Welcome, Guest. Please login or register.

May 13, 2021, 09:23:02 pm

Author Topic: SE HELP! Bad data for research question  (Read 285 times)

0 Members and 1 Guest are viewing this topic.

youcandothis

  • Adventurer
  • *
  • Posts: 10
  • Respect: 0
SE HELP! Bad data for research question
« on: April 29, 2021, 01:44:02 pm »
0
Hi all,

My bio student experiment draft is due tomorrow, and there are some things I'm worried about.

In the original experiment we used stratified sampling to collect and analyse primary biotic field data (height of tallest stratum and percentage cover) to classify an ecosystem (this was in a dry scherolphyll forest). Our class went on a field trip and collected a bunch of data, and then we had to develop a RQ from that. Our RQ is
What is the correlation between soil pH and percentage cover of trees, measured in a dry sclerophyll forest ecosystem and a mangrove ecosystem?
As you can see in the picture our data has virtually no correlation (A is forest and B is mangrove), but my teacher said we could still use this as our RQ. The optimum pH for plant growth is between 5-8%, hence the percentage cover should be higher when the pH is at this point. I am not sure what to write in the analysis, other than the fact that it doesn't correlate. Getting rid of outliers doesn't help.
In addition, I don't have that much research to back up my initial claim in the rationale as I feel like saying there should be a relationship between pH and percentage cover is a bit of a stretch as there isn't literature that directly supports this, and I've mainly only found gardening websites for optimum soil pH.


It's too late to change questions (plus changing a question wouldn't help as all the data we collected doesn't show a correlation). But I'm also concerned that this RQ doesn't really link to the content matter or our original experiment (which was looking at classifying the ecosystem, whereas we're looking at if there's a relationship between two variables), or the content matter.

Any help is appreciated! :)
« Last Edit: April 29, 2021, 02:15:04 pm by youcandothis »

Bri MT

  • VIC MVP - 2018
  • Administrator
  • ATAR Notes Legend
  • *****
  • Posts: 4584
  • invest in wellbeing so it can invest in you
  • Respect: +3542
Re: SE HELP! Bad data for research question
« Reply #1 on: April 29, 2021, 03:52:44 pm »
+2
Hi!

On mobile so I'll come back later and elaborate more but:

- it's fine if the data ends up meaning you support the null hypothesis
- have you tried fitting a non-linear curve?
- look at the literature which indirectly supports it,  which you can use to suggest that there may be this relationship
- you still have data you can use to address the idea of there being / not being a relationship so try not to stress about your data being "bad"


I hope this helps!

youcandothis

  • Adventurer
  • *
  • Posts: 10
  • Respect: 0
Re: SE HELP! Bad data for research question
« Reply #2 on: April 29, 2021, 05:15:20 pm »
0
What do you mean by a non linear curve? I tried putting into a polynomial curve, looks like this, but I don't have experience analysing this type of curve. Our teacher told us to use the R^2 value, which I think(?) is only for a linear curve.
« Last Edit: April 29, 2021, 05:16:53 pm by youcandothis »

Bri MT

  • VIC MVP - 2018
  • Administrator
  • ATAR Notes Legend
  • *****
  • Posts: 4584
  • invest in wellbeing so it can invest in you
  • Respect: +3542
Re: SE HELP! Bad data for research question
« Reply #3 on: April 30, 2021, 09:42:57 am »
0
What do you mean by a non linear curve? I tried putting into a polynomial curve, looks like this, but I don't have experience analysing this type of curve. Our teacher told us to use the R^2 value, which I think(?) is only for a linear curve.


Yeah a non-linear curve is some curve that isn't a straight line.

If you've been explicitly told to use R^2 it is better to stay away from a polynomial curve.