I'm conducting a hypothetical study on sugar consumption and the effects it has on metabolic risk factors.
The study will contain 2 experimental groups different in sugar consumption and one control group consuming their regular diet. Primary outcome measures are blood pressure, triglyceride levels, HDL-C levels, plasma blood glucose and waist circumference. Secondary outcome measures will be BMI and body fat% (DEXA).
I'm wondering what statistical tests to apply to get the best possible results. I'm thinking of using an ANOVA (but which one?) amongst others.
Any help would be truly appreciated.
Kind Regards,
Matt
You can use ANOVA If the researcher wants to study the impact of one or more factors or categorical variables on more than one variable then MANOVA or Multivariate Analysis of Variance can be used. It can be used to measure whether there are any differences between independent groups on more than one continuous dependent variable. Like any other ANOVA method the dependent variable should be on metric scale whereas the independent variable may be on non-metric scale .
Dear Matt Moore , it sounds like you have one categorical independent variable and many dependent variables. The appropriate test in these circumstances is MANOVA. However, if your risk factors lead to a binary outcome, such as contracting/not contracting an illness or survival/non-survival you should try a binary logistic regression.
You haven't mentioned the sample size. Run normality tests first for each of these outcome measures. I guess Kruskal Wallis ANOVA with post hoc Mann Whitney tests will do.
Hi Peter Samuels , it does seem like the MANOVA is the test I should use from reading about it. Although I will be looking at metabolic risk factors increasing, I am not looking to see if they result in metabolic syndrome (binary outcome).
Many thanks,
Matt
The ANOVA to be used separately for Blood pressure, Tg, HDL-C level, ........
Also use Post-Hoc comparison to see how Controls are differing with experimental group 1 and experimental group 2. Sometimes, even you may use some transformation like log transformation.
- Badges
- Science topic
- Similar topics
- Social Science
- Sociology