Yes you can. The RHS of your model will be the same as if you had only one DV because you are using the same experimental design for each DV you are interested in studying. Because you have a longitudinal design I would look at these links as well:

Davidian Notes

https://www4.stat.ncsu.edu/~davidian/st732/notes.html

Multivar. Long. Models:

https://www.researchgate.net/post/What_statistical_test_is_appropriate_for_pre_and_post_trails_with_1_before_treatment_value_and_4_after_treatment_values_assessed_weekly

On the question about coding your categorical IVs this link may help:

https://stats.idre.ucla.edu/spss/faq/coding-systems-for-categorical-variables-in-regression-analysis-2/

Best wishes, David Booth

Hi David Eugene Booth thank you for the reply. Would this be considered a true longitudinal study, and if not would this still apply? For the study design it was impossible to do repeated measures because some of the DV we are looking at required animal sacrifice. So while there are 3 timepoints, each timepoint is made up of a different population of mice.

Yes, you can use multiple regression analysis that combines continuous and count (or Categorical) explanatory or independent variables. However, If the response variable is continuous (from a continuous function) it's simply a Multiple Regression Analysis, and if the response variable is binary, it's Possion Regression Analysis. You can check further for the assumptions, if the dataset would also satisfy it. Your N trials, will it be small? (since you don't like much animal casualities, nor repeated measure).

With the dummy(binary) , you can as well put the time levels as Categorical at three time points (no repeated measure) so far the animals were distributed by a random process.

1. Diet/Dummy (row), time (column) and alternatively,

2.Three Diets (row), binary dummy(column), on time(t1,t2,t3) layer. i.e., {time, dummy}/{diet} .

*Only proposing pls.

Geoff Kerr Assuming that the animals at each each time point were given the same treatments throughout the experiment, I would probably treat it as a repeated measure design with the different populations mentioned prominently in the limitations. The reason is that I have no idea how to handle a distinct experiment with that design other than what I said. Two points:

1. ALWAYS decide on the statistical methods PRIOR to collecting data ESPECIALLY if sacrifice is required. At Rothhamsted Experimental Station they are said to have had a song, "and Fisher can always allow for it". Most of us aren't of Fisher's class however.

2. When designing an experiment I ALWAYS suggest trying to use a standard design because then we are sure that we can deal with the results. My Fav reference is: https://www.bing.com/search?q=montgomery+d+c+design+and+and+analysis+of+experiments&qs=NW&pq=montgomery+d+c+design+and+and+analysis+of+ex&sk=NW2&sc=5-44&cvid=CCA605986A21463B896ECAD4C6DBFEE0&FORM=QBLH&sp=3

If someone else knows a better answer, I hope s/he will comment.

Finally, I am told by a colleague that your university has a fine statistics department. You may wish to consult with a statistician there.

Best wishes, David Booth One further comment:

As Abiodun Christian Ibiloye suggests the regression depends on the type of DV you have, see: https://www.bing.com/search?FORM=U528DF&PC=U528&q=What+type+of+Regression+do+I+use

If you have all continuous DVs the use OLS based multivariate regression is OK, see:https://www.bing.com/search?q=multivariate+regression&qs=LS&pq=multivariate&sk=LS2&sc=8-12&cvid=4E07EB70F0314B1AA5E38E81D3400C43&FORM=QBRE&sp=3

Otherwise I would separate DVs and use the regression appropriate for each. See and use the appropriate regression type as described in a previous link.

Hope this helps, Best wishes, David Booth

I believe this is not a longitudinal study since it does not involve repeated measure of the same animal -subjects. Also, it fits Multiple Regression Model. The response variable is continuous but the independent variables can be both norminal/counts or dummy[eg sex](x1) and continuous(x2, x3). Yes!

See https://www.healthknowledge.org.uk/public-health-textbook/research-methods/1b-statistical-methods/multiple-logistic-regression

However, if the response variable is binary, then logistic Regression can be used, and if it is a count, the poisson regression.

*Logistic regression is used when the outcome/response variable is binary, being either an event (e.g. death or survived or not cured). The input/independent variables can be either binary or continuous!

{The simplest case is that of one independent variable and which is binary, then it gives the same result as a chi-squared test.}

The logistic regression model for q Vs:

logit(p)=a + b1X1 + b2X2 + … + bqXq and X={X1, X2, ...Xq}

If you use a GLM (General Linear Model) procedure, the categorical independent variables will be automatically re-coded as k-1 binary (0;1) variables (where k is the number of categories). Continuous independent variables are called covariates and nominal independent variables are called factors. GLM includes ANOVA, Regression Analysis and Covariance Analysis as special cases.

Some reading this thread may wish to look at the different coding schemes for categorical variables see

https://stats.idre.ucla.edu/spss/faq/coding-systems-for-categorical-variables-in-regression-analysis-2/

Best, David Booth

Yes you can use these as independent variables. But you have to be careful in interpreting the estimates of parameters corresponding to (i) categorical variables. Because each software defines dummy variables (one less than the number of labels in categorical variable) differently.