$Choosing the right statistical tests and how to organize the data accourdingly (student research project)$

Question

$Choosing the right statistical tests and how to organize the data accourdingly (student research project)$

Hi all!

This is my first time using this website so I apologize in advance for any mistakes. Accordingly, I have no idea if I have offered the right amount of money. Let me know if it’s too low (or too high). I would really love some help with choosing the right statistics for my research project for my masters degree. It has been a couple of years since our last statistics course… I program in Matlab and I use SPSS for the statistics. I have attached a (sample) data set to this question (as a screenshot, I couldn’t upload an excel file). For the sake of this question assume normality and sphericity of data.

I will now explain the relevant parts of my experiment. I had one group of (17) persons. Each person performed a bench press protocol. The bench press protocol consisted of 3 intervention (‘INT’) reps and 3 control( ‘CON’) reps. During the repetitions we measured the (peak) power and the muscle activity.

I want to compare the peak power values and the muscle activity of the intervention reps against the control reps. I want to compare it at group level but I also want to be able to say something about each individual (to be able to classify responders and non-responders, which is a phenomena for this specific area of research).

I thought of using a paired samples T-test. I would then put all INT_peakpower in 1 column and all CON_peakpower in another column and compare them (and vice versa for muscle activity). I also thought about averaging the values for each person (so each person gets 2 values, 1 INT_peakpower and 1 CON_peakpower) and then performing a paired samples T-test. However, I thought that both methods undermine the power of me having done 3 measurements per person…

I didn’t think a repeated measures ANOVA would work because both repetition types have 3 measurements. Maybe a 2x3 ANOVA (2 types of repetition and 3 repetitions) but I’m not sure how that works.

All in all, how can I best analyze my data set and how do I need to arrange the data accordingly?

Thanks you so much in advance!

Statistics

Wvdw94

9

Matchmaticians Choosing the right statistical tests and how to organize the data accourdingly (student research project) File #1

File #1 (png)

Report

Savionf

+3

This is not my area of expertise, but the offered bounty seems low. Think about how much time one may need to spend to write a good solution and what a fair hourly rate for such individual would be. That would give you a pretty good idea to decide on a fair bounty.
Wvdw94

+1

Thanks for the info. I doubled the bounty.

Answer

The answer is accepted.

Join Matchmaticians Affiliate Marketing Program to earn up to a 50% commission on every question that your affiliated users ask or answer.

Savionf

+3

This is not my area of expertise, but the offered bounty seems low. Think about how much time one may need to spend to write a good solution and what a fair hourly rate for such individual would be. That would give you a pretty good idea to decide on a fair bounty.

Answer 1

Answers can only be viewed under the following conditions:

The questioner was satisfied with and accepted the answer, or
The answer was evaluated as being 100% correct by the judge.

View the answer

It would be wrong to run a paired t-test between columns (INT_peakpower and CON_peakpower) because you would then ignore the subject-wise intrinsic variability of the repeated measures (per individual).

The second approach is what you want to do in practice, and it's basically a mixed model.

The main assumption of this model is that measurement $\mu_{ij}$ (index $i = 1,\dots, 3$ represents a measurement on subject $j = 1, \dots, n)$, obeys the equation:

$\mu_{ij} = \beta_0 + \beta_1 \cdot INT_i + \omega_j + \epsilon_{ij}$,

where $\omega_j =^d N(0,\sigma^2_p)$ are iid and represent subject-wise intrinsic variability and $\epsilon_{ij} =^d N(0,\sigma^2_t)$ are also iid and represent treatment variability.

You could run this test in R like this: (you would need to change your data into a vertical form with a variable indicating TREATMENT: 1 or 0).

library(lme4)
model1 <- lmer(INT_peakpower~ TREATMENT+ (1 | PP)
summary(mod)

Here is a brief introduction to linear mixed models:

https://stats.oarc.ucla.edu/other/mult-pkg/introduction-to-linear-mixed-models/

Mathe

3.5K

Wvdw94

0

Thank you for your answer. Any idea how to do that in SPSS?
- Mathe
  
  0
  
  This is a very thorough explanation for SPSS: https://csass.ucsc.edu/self-study%20courses/SPSSMixed.pdf

$Choosing the right statistical tests and how to organize the data accourdingly (student research project)$

Answer

Related Questions

Search