Independent t-test Analysis Guide

Independent t-test

• Step 0: Set the working directory. Example: setwd("C:/Users/W11/Downloads/Master UM/Data")
• Step 1: Activate library. Example: library("psych") (used for checking normality)
• Step 2: Set significance level (alpha). Example: alpha = 0.05
• Step 3: Check data requirements.
  • a. Load data: Data <- read.csv("KSM3403 Data.csv")
  • b. Independent variable can be numerical or text.
  • c. Check assumptions:
    • i. Normality: describeBy(Data$Score, Data$Group)
      • Descriptive statistics by group calculated. Example: for group 0, mean, standard deviation; median etc are listed. Skewness values should be between -2 and +2 for normal distribution.
    • ii. Variance equality: var.test(Data$Score ~ Data$Group)
      • Levene's test result for variance equality and relevant p-value are reported. If p-value > 0.05, variances are equal.
• Step 4: Run Independent t-test.
  • t.test(Data$Score ~ Data$Group, var.equal = TRUE) (If variances are equal)
  • t.test(Data$Score ~ Data$Group, var.equal = FALSE) (If variances are not equal)
    • Calculate t-value, Degrees of freedom(df), p-value, confidence interval and sample estimates for the groups.

Analysis of Variance

• Step 1: Set working directory: Example: setwd("C:/Users/W11/Downloads/Master UM/Data")
• Step 2: Load libraries: Example to run t-test, descriptive analysis and Levene's test library("stats"), library("psych"), and library("car")
• Step 3: Import datafile into RStudio. Example Data <-read.csv("Data for 1 and 2 way.csv")
• Step 4: Test assumptions.
  • a. Test for normality: describeBy(Data$Inquiry, Data$Type) (example) for normality test of inquiry by type, to see whether the distribution of inquiry is a normal distribution.
• Step 5: Run analysis.
  • One-way Analysis of Variance: one.way <-aov(Inquiry~ as.factor(Type), data = Data)
  • Two-way ANOVA example: two.way <-aov (Inquiry~ as.factor (Type) * as.factor (Day), data = Data)
• Step 6: Post hoc analysis.
  • Run post hoc test if the factor in ANOVA step 5 is significant.
    • Example: post_type <- TukeyHSD (one.way)

Correlation

• Step 1: Set working directory
• Step 2: Load libraries: e.g. library("stats"), library("psych")
• Step 3: Import data
• Step 4: Assumption test.
  • Normality: check if the variables normally distributed. Example describe (CorData$Work_Motivation). Descriptive statistics, skewness are useful for checking.
  • Independence: check for independence of observation
• Step 5: Run tests.
  • Pearson correlation test: cor.test(CorData$Work_Motivation, CorData$Job_Satisfaction) to test the correlation between work motivation and job satisfaction
  • Point biserial correlation: cor.test(CorData$Status, CorData$Work_Motivation) to test the correlation between employment status and work motivation. Correlation value, p-value and confidence interval are reported.

Non-parametric test

• Purpose: to examine differences between two or more groups when the dependent variable is not normally distributed. For example, Mann-Whitney U (or Wilcoxon Rank Sum) test.
• Steps:
  • Set working directory.
  • Import data.
  • Conduct descriptive analysis for each group (describeBy).
  • Run the wilcox.test to examine difference in each group.

Multiple linear regression

• Step 1: Set working directory
• Step 2: Load libraries: library("stats"), library("psych"), library("car")
• Step 3: Import data
• Step 4: Fit the linear model: MLR.fit <- lm(Job_Satisfaction ~ Work_Experience + Job_Performance, data=MLRData)
• Step 5: Testing assumptions. Includes
  • Test of independence: check if observations are independent.
  • Linearity: check if the relationship between independent (IVs) and dependent variables (DV) is linear
  • Homoscedasticity: Are the errors homoscedastic? Plot of predicted vs residual values, and check.
  • Normality: Are the errors normally distributed? Use Q-Q plot
  • Multicollinearity: Are predictors highly correlated? Check variance inflation factor (VIF)
• Step 6: Interpretation
  • Interpret results of the fitted line (summary(MLR.fit)) (R-squared, coefficients table, p-value, predictors, confidence interval).

This quiz provides a comprehensive step-by-step guide for conducting an independent t-test using R. It covers setting up your working environment, checking data requirements, and running the t-test while addressing key assumptions such as normality and variance equality. Enhance your statistical analysis skills with this practical resource.