$Proper Test Selection for Repeated Interval Data Collection?$

Question

$Proper Test Selection for Repeated Interval Data Collection?$

I am working on analyzing the results of some sensor data that I acquired through an experiment. The experiment involved subjecting a set of pipes to a repeated load. I collected sensor response data at various load intervals throughout the experiment (100 load cycles, 1,000 load cycles, etc..). Each data set contains 5 load cycles that I have pulled a maximum value from.

I am attempting to determine if the load cycle or run number (1-5) within each data set, effects the sensor response. I would also like to determine if the number of load cycles is effecting the response of the sensors.

I've been using a repeated measures anova with my various sensor values for each run as the dependent variable, my load cycles interval (100, 1000, etc) as my between-subject factor, and the load cycle number within each data set (1-5) as my covariate.

I'm not sure this is valid, as the conditions of the experiment are changing (soil/asphalt compaction) with each load application.

Statistics

Bjones13

8

Report

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.

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

Please see my detailed response below:

Repeated measures ANOVA is an appropriate choice if you have sensor data collected from the same subjects or experimental units under different scenarios. So, given that the conditions of your experiment change with each load application (e.g., soil/asphalt compaction), it's important that you consider the potential lack of independence between the different load cycles, which may violate the assumptions of traditional ANOVA or repeated measures ANOVA. In this case, I think you can consider altertnative approaches. Two main statistical approach/test alternatives comes to my mind. Mixed-Design ANOVA and Generalized Linear Mixed Model (GLMM).

Mixed-Design ANOVA approach can be used when you have both within-subject factors (like the load cycle number within each data set) and between-subject factors (like the different load cycle intervals and/or soil/asphalt compaction conditions). It can help account for the non-independence of data due to changing conditions.

GLMMs can handle correlated data as well as mixed categorical and continuous factors. They allow you to model the relationship between sensor responses and various factors, also taking into account the changing experimental conditions.

You can set up a Mixed-Design ANOVA with your data (use SPSS or other packages that support mixed-design ANOVA), as below:

Within-Subject Factor: Load cycle number within each data set (1-5).

Between-Subject Factor: Load cycle interval (e.g., 100, 1000, etc.) and the changing conditions like soil/asphalt compaction.

Make sure your data is organized in a way that allows you to distinguish between the within-subject and between-subject factors. You should have sensor response values for each load cycle number and condition (load cycle interval/compaction) for each run.

You'll find out if there's a significant difference in sensor response across different load cycle numbers. Also, find out the effect of load cycle interval/compaction telling you if there's a significant difference in sensor response across different load cycle intervals or compaction conditions. If you find significant effects, you may conduct post-hoc tests to determine which specific levels of factors are different from each other.

Kav10

1.9K

$Proper Test Selection for Repeated Interval Data Collection?$

Answer

Related Questions

Search