Sampling Distribution of the Sample Proportion

Professor Dan Kernler introduces the concept of the distribution of the sample proportion. Using a national survey example, he compares the proportion of registered voters who are Democrats (32%) to a sample proportion from a database of children of immigrants (0.339 or 33.9%). The core question is whether this difference is statistically significant or merely due to chance, highlighting the need to understand the distribution of sample proportions.

The video explains that if we consider individuals as either Democrats or not, this forms a binomial distribution. By visualizing binomial distributions for increasing sample sizes (N=10, 20, 50, and 3238), it's observed that as N increases, the distribution becomes increasingly bell-shaped, resembling a normal distribution. A key condition for this approximation is that N * P * (1 - P) must be at least 10.

For a binomial distribution that is approximately normal, the mean is N * P, and the standard deviation is the square root of (N * P * (1 - P)). These formulas are crucial for understanding the properties of the distribution of counts.

The video then transitions from counts to proportions. The mean of the sample proportions (P-hat) is the population proportion (P). Through algebraic derivation, the standard deviation of the sample proportions (P-hat) is shown to be the square root of (P * (1 - P) / N). These formulas are fundamental for analyzing sample proportions.

A summary of the key concepts is provided: if the condition N * P * (1 - P) ≥ 10 is met, the sample proportion (P-hat) will be approximately normally distributed with a mean equal to the population proportion (P) and a standard deviation given by the derived formula.

The video revisits the initial example, checking if the condition for normal approximation is met (N*P*(1-P) = 704.6, which is > 10). It then calculates the mean (0.32) and standard deviation (0.008198) for the sample proportion distribution based on the population proportion of 32%. The importance of retaining many decimal places for the standard deviation in calculations is emphasized.

The video demonstrates how to calculate the probability of observing a sample proportion of 0.339 or higher, given the established normal distribution. Using StatCrunch, the probability is found to be 0.01. This low probability suggests that it is quite rare to get such a high sample proportion if the true population proportion is 0.32. The conclusion drawn is that the proportion of Democrats among children of immigrants is likely slightly higher than the national average, despite a small numerical difference.

The video concludes by acknowledging the challenging nature of sampling distribution sections and encourages viewers to rewatch the video or seek additional resources. It thanks Elgin Community College for supporting the creation of these educational videos.