This post has been de-listed
It is no longer included in search results and normal feeds (front page, hot posts, subreddit posts, etc). It remains visible only via the author's post history.
Hi!
I'm working on a social sciences study where I surveyed a group of people to know wether they knew how to act in case of an earthquake. First, I asked some sociodemographical questions, (age, highest degree of studies, gender, suburb, occupation), afterwards, I made an "exam" where I evaluated two parameters: knowldege of earthquake origin and knowledge on how to act in case of an earthquake. I'm looking for a statitistical test that can tell me how much does each independet variable contribuite to explaining a pattern in the dependant variables, and if the combination of two independent variables explain a dependent vairable better than both of them by separate.
I was looking at implementing a MANOVA, but I have no experience with it, and I have no idea wether it makes sense to use it.
Any help would be much appreciated.
Thank you very much in advance.
Please, pardon my English, it's not my first language.
This is hard to answer without looking at your dataset.
How are your dependent variables coded? Is it continuous (many numerical values) or categorical (e.g., binary, yes/no)?
First, I suggest looking at your univariate distributions (establish normality, look for outliers, etc.). Then, I would suggest looking at bivariate associations of sociodemographic characteristics by each of your 2 outcomes (separately). This will tell you the significance of the association between each of your independent variables and your 2 dependent variables.
To look at the significance of multiple independent predictors on each dependent variable, multivariable linear or logistic regression seems appropriate, depending on your distributions.
Subreddit
Post Details
- Posted
- 2 years ago
- Reddit URL
- View post on reddit.com
- External URL
- reddit.com/r/statistics/...