which statistical test do i use?

Question

i have undergone an experiment and do not know which test to use. i want to see if there is a relationship between age, gender and height with grip strength.

Answer 1

Hi Shanise,

Your question is very interesting because it goes directly to a real-world application of statistics. To provide a clear answer, we first need to define the roles of your variables. I am assuming your setup is as follows:

VARIABLE TO PREDICT (Dependent Variable):

1. Grip Strength: Measured with continuous quantitative values (e.g., kilograms or pounds).

PREDICTOR VARIABLES (Independent Variables):

1. Gender: Categorical (e.g., Male and Female).
2. Age: Continuous (e.g., years).
3. Height: Continuous (e.g., centimeters or inches).

Under this scenario, I recommend using Multiple Linear Regression. Here is why:

1. Versatility: Linear regression is designed to predict a quantitative variable from a set of predictors that can be either quantitative or categorical (using dummy variables). For Gender, you would simply assign a 0 for Women and 1 for Men.
2. Statistical Significance (p-values): The model will provide a p-value for each predictor. For example, if the p-value for Height is less than .05, we can conclude that Height is a significant predictor of Grip Strength in your sample.
3. Accounting for Overlap (Control): This is the most powerful part. Height and Gender are often correlated (on average, men are taller). Multiple regression "disentangles" these effects. It tells you if being taller actually predicts strength, or if it just appears that way because of gender differences. It attributes the effect to the most likely predictor while holding the others constant. (Note: This suggests a relationship but is not absolute proof of causality).
4. Strength of Relationship (B Coefficients): The model provides B values. For instance, if the B value for Height is 0.10, it means that for every additional centimeter in height, we expect a 0.10 kg increase in grip strength, assuming age and gender remain the same.

Final Tip: Multiple regression works well in most cases, but to be 100% sure, we should check that the data doesn't violate key assumptions like Normality and Homoscedasticity of the residuals.

I'd be happy to help you run this analysis in R, SPSS, or Excel and help you interpret the output for your experiment.

Good luck!