19.Estad_ Pruebas paramétricas y no paramétricas _mayo2025

Understanding Parametric and Non-Parametric Tests

Overview of Statistical Significance Procedures

• Statistical significance tests are essential for procedures involving correlations, group comparisons, and level identification. These tests can be categorized as either parametric or non-parametric.

Determining Normality Conditions

• The choice between using parametric or non-parametric tests is based on evaluating whether a condition of normality is met. It’s important to note that this refers to the condition of normality rather than the normality of specific data sets.

Shapiro-Wilk Test for Normality

• The Shapiro-Wilk test is currently favored for assessing normality due to its effectiveness compared to other tests. This test is also available in popular free software like Jamovi.

Interpreting P-values from Shapiro-Wilk Test

• The results from the Shapiro-Wilk test yield p-values that must be interpreted against a significance level (alpha). If the p-value exceeds 0.050, it indicates a normal distribution; otherwise, it suggests a non-normal distribution.

Common Misunderstandings about Data Distribution

• A frequent misconception is equating the Shapiro-Wilk test with analyzing data distribution directly; however, it assesses how closely data approximates a theoretical normal distribution rather than confirming actual data distributions. This distinction is crucial for accurate statistical analysis.

Implications of Normality on Statistical Testing

Choosing Between Parametric and Non-Parametric Tests

• If the assumption of normality holds true, researchers should proceed with parametric tests; if not, they should opt for non-parametric alternatives. This decision impacts subsequent analyses significantly.

Software Utilization in Normality Testing

• In practical applications using software like Jamovi, users can easily conduct normality tests by selecting options within the interface to obtain p-values and draw conclusions regarding data distribution types efficiently.

Clarifying Misconceptions About Normal Distribution

Distinction Between Data Analysis and Goodness-of-Fit Tests

• It's vital to clarify that conducting a normality test does not analyze the actual dataset's characteristics but evaluates how well it fits into an expected theoretical model (normal distribution). Understanding this difference helps avoid common analytical errors among practitioners.

Historical Context and Evolution of Normality Testing Practices

• Historically, there was a practice where different tests were used based on sample size (Shapiro-Wilk for n<50 and Kolmogorov-Smirnov for n>50), but current practices have shifted towards exclusively using the Shapiro-Wilk test regardless of sample size due to advancements in software capabilities like Jamovi's testing algorithms over time.

Shapiro-Wilk Test for Normality

Overview of the Shapiro-Wilk Test

• The discussion begins with the mention of the Shapiro-Wilk test, highlighting its utility when options seem limited in statistical analysis.

• It is noted that the Shapiro-Wilk test is considered a robust method for assessing normality across various datasets.

• The speaker emphasizes that recent studies indicate the effectiveness of the Shapiro-Wilk test compared to other normality tests.

• A paradoxical observation is made regarding the performance of the Shapiro-Wilk test, suggesting it may yield unexpected results under certain conditions.