I need to use SPSS to see if the first 2 predictor variables correlate and by how much to OV of alcohol consumption.
Multiple regression? How do i input this data into SPSS, whats tests do i run and how would i explain the results/outputs please?
Any beginner help (SPSS for dummies) appreciated wholehearedly.
I can't help you with SPSS however I can with R. I have attached a helpful introduction. If you want something even simpler Google Rcommander which is a simple interface for R. The questions you asked are answered the information I sent you. Best wishes, David Booth
Hello Mandy,
As Bruce's question implies, how your variables are quantified makes a difference. If all three of your variables (DV = Alcohol consumption; IV1 = Marital problems; IV2 = Overall happiness) are quantified as continuous (or pseduo-continuous, as in a scale score) variables, then, yes: Multiple linear regression (MLR) is a possible option for data analysis.
If, however, a variable is categorical, the answer changes somewhat. If it's the DV, then possibly logistic regression (whether simple binomial or multinomial depends on the number of levels of your variate, Alcohol consumption) would be suitable. If just an IV (say, Marital problems was coded: None, some, lots, divorced), then you could recode the observed categories into C - 1 dummy variates (where C is the number of levels of the categorical variable), and proceed with MLR.
In SPSS, ordinary MLR can be accessed by the menu system as: Analyze/Regression/Linear... (specify the DV first, then specify the IVs in the box below). Doing so will give you the very basics: Model fit & significance, regression coefficient estimates (and significance).
Good luck with your work!
In all statistical analyzes, we need to identify the data, to identify the type of data, whether quantitative or nominal, then determine the type of correlation coefficient required,
Hi mandy. It depends on the variables you will be using to represent marital problems, happiness and alcohol consumption. if one variable is, let's say , marital problems is nominal (present/absent) and the other two are numerical ( happiness - measured in likert and alcohol consumption - frequency), then POINT BISERIAL correlation would be appropriate. If these three are numerical, then PEARSON would be the choice. I also suggest computing for its Rsquared (coefficient of determination) to measure the variation brought by the independent variable to the dependent.
- Badges
- Science topic
- Similar topics
- Psychology
- Emotional Psychology
- Happiness