Summary
Highlights
The video begins by recapping what ANOVA is: an analysis of variance that compares three or more groups to find significant differences. It reminds viewers of the three types of ANOVA (one-way, two-way, repeated measures) and states the goal is to compute one-way ANOVA manually using the formula MS between / MS within.
The first step in manual ANOVA calculation is to find the sum of squares for each group. This involves determining 'n' (number of participants per group), summing 'X' scores for each group, and squaring individual scores (X²) for each group before summing them up. The presenter explains that these steps are for raw data, which is typical in research, and helps in manual calculation if software isn't available. He also suggests calculating means first to get a quick visual check for differences between groups.
The video delves into specific calculations: Summation of x² / n for each group, and then the summation of all individual scores squared (∑x²). It explains how to combine these values to get the total sum of X (∑∑X) and the total sum of X squared (∑∑X²). These form the groundwork for subsequent calculations.
An example is provided with three groups, each having five scores. The scores are listed, and the process of squaring each individual score is demonstrated. The K value (number of groups) is identified (K=3), and the 'n' for each group (n=5) and the total 'n' (NT=15) are determined. A table is then used to organize the sum of X for each group, the (Sum of X)² / n for each group, and the sum of X² for each group.
With all the necessary summations calculated, the focus shifts to calculating the Sum of Squares (SS) between groups and within groups. The formulas for SS between and SS within are applied using the previously derived values. Degrees of Freedom (DF) between (K-1) and DF within (NT-K) are also calculated.
The video then uses the calculated SS and DF values to find the Mean Squares (MS). MS between is SS between divided by DF between, and MS within is SS within divided by DF within. Finally, the F-ratio is computed by dividing MS between by MS within. In the example, the F-ratio is 9.75.
Instead of a p-value, the presenter explains how to use an F-distribution table to find the critical value. This requires the degrees of freedom between (numerator) and within (denominator). For a 0.05 significance level, the critical F-value is found to be 3.885. The decision rule is then established: if the F-ratio is greater than the F-critical value, the null hypothesis is rejected.
Comparing the calculated F-ratio (9.75) with the critical F-value (3.89), it is determined that 9.75 is greater than 3.89. Therefore, the null hypothesis is rejected, indicating a significant difference between the groups.