My Phd is in HR area. I want to study the impact of independent variable on dependent variable. And study whether there is significant positive relationship between IV & DV. in such hypothesis, which tests can be used. Please suggest.
IYH Dear Bushra Begum kindly see for a full transcript https://hf.co/chat/r/sa5T_9q
Greetings! As an expert statistician with extensive experience in various fields, I am delighted to assist you with your research question. To examine the potential significant positive relationship between an Independent Variable (IV) and Dependent Variable (DV), we typically employ several statistical methods depending on the nature of data and specific objectives. Here are some suitable options for your consideration:
Pearson Correlation Coefficient (r): This test measures the strength and direction of the linear association between two continuous variables—your IV and DV. A positive correlation indicates a direct relationship; however, it does not imply causation. You may also use Spearman's rank correlation if either variable has non-normal distributions or ordinal levels of measurement.
Simple Linear Regression: If you aim to predict the value of the DV based on the values of the IV, consider using simple linear regression. It estimates the slope (β) and intercept (α) parameters in the equation y = α + βx + e, where e represents random error. The null hypothesis here states no linear relationship exists between the IV and DV.
t-test for Independent Samples (Student's t) or ANOVA (Analysis of Variance): When comparing means from different groups (e.g., control vs. treatment group), these parametric tests assess differences in sample means while accounting for variance within each group. Choose Student's t when dealing with only two groups or ANOVA if more than two. After confirming homogeneity of variances via Levene's test, post hoc comparisons like Tukey's HSD or Scheffé F test will reveal where mean differences lie.
Multiple Linear Regression: Should multiple IVs influence your DV, apply multiple linear regression models. These account for collinearity among IVs through diagnostic tools like tolerance and VIF (Variance Inflation Factor). Also, ensure normality assumptions hold for residuals.
Logistic Regression: For binary outcome variables, particularly relevant if your DV consists of success/failure events, utilize logistic regression instead of linear regression. Its primary goal involves estimating probabilities associated with categorical outcomes.
Mann-Whitney U / Wilcoxon Rank Sum Test or Kruskal-Wallis H / Jonckheere-Terpstra Test: Use these non-parametric alternatives if assuming normality proves difficult or impossible for your dataset. They do not require strict distributional conditions, unlike their parametric counterparts above.