2
$\begingroup$

I'm currently writing my bachelor's thesis and I'm facing a small problem: I don't know how to properly discuss my results.

I conducted a 2×2 factorial experiment with the factors Vole and Mulch. In total, there are 16 lysimeters filled with sandy soil. Eight plots have been subjected to vole activity (voles digging for one week per year for the past 11 years), and for the past 5 years, eight plots have received a mulch layer (the cut vegetation is left on the plot after mowing in autumn).

My aim is to measure the steady-state infiltration rate (SIR) using a hood infiltrometer and to determine whether there are differences between the treatment groups. I measured SIR 9 months after mulch application (no vissible mulch was left on the plots).

My resulting plot looks like this: enter image description here

I ran an ANOVA and found no statistically significant effects. However, visually it seems that:

  1. mulching might have a slightly negative effect on SIR,
  2. vole activity has no clear effect on SIR,
  3. there is no interaction between the two factors.

Does anyone know of studies that explain why not mulching could slightly reduce SIR? Or do you see any other patterns in the figure? I have trouble finding studies specifically on SIR. Some papers use metal cores in the lab to measure saturated hydraulic conductivity under constant head conditions, but that's not equivalent to SIR.

Additional information about the experiment:

  • Soil: medium-granied sandy, loose soil; some earthworm castings present
  • Vegetation: mainly grasses and legumes; some moss
  • cliamte: temperate-continental; annual precipitation: ca. 540mm, annual temp.: ca. 9,7°C
  • Vole treatment: 1 week of vole activity per year for 11 years (in autumn)
  • Mulch treatment: mowing in autumn and leaving the cut vegetation on top of the plots
  • No-vole treatment: no vole manipulation
  • No-mulch treatment: vegetation removed after mowing
  • No human or animal has stepped on the plots; no additional compaction occurred.

I'm grateful for any advice. If you need additional inforamtion let me know :)

Improve this question
$\endgroup$
3
  • $\begingroup$ I'm certainly no expert in the field (or statistics for that matter), but wondering: doesn't no statistically significant effects thereby mean, while visually a certain trend may appear, there isn't huge confidence it's real? At least not to the (guessing 95%?) confidence level. Maybe there's a 60% or even 90% chance this is showing a real difference, but there's real reasonably potential for it just to be noise? $\endgroup$ Commented Nov 25 at 9:14
  • $\begingroup$ Guessing the confidence level is quite low just because of the fairly low sample size (if I understand right, you only did infiltrometer after the 5th year, so only 16 data points?) Funny though, my BS capstone relied on similarly uncertain statistics [looking at small diffs in dew points around city obs] (many more data points, but also a lot of other factors!) & posed a bunch of thoughts, so I guess it may not be awful at your level, especially if you can properly discuss the confidence level and meaning? And maybe there's more exotic stat tools you can apply to get some better power still? $\endgroup$ Commented Nov 25 at 9:23
  • $\begingroup$ your sample size is small and all of your boxes overlap so I would not expect a significant difference. Remeber we expect some diffrence in reslts just due to chance thats why Anova exists, t calculate if difference is significant. Remeber also you ASSUME no difference then try to disprove it. $\endgroup$ Commented Nov 27 at 14:27

0

Reset to default

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.