91Ó°ÊÓ

When we talk about the "difference in proportions," we're interested in comparing two groups' percentage of an event or characteristic. In this exercise, you see that the focus is on the percentage of American adults visiting Facebook daily in 2010 and 2013. Proportions are seen as ratios. They can be expressed as percentages or fractions.

• The proportion in 2010 was 51%.
• The proportion in 2013 was 64%.

The difference in proportions is simply the subtraction of these two numbers: 64% - 51%, which equals 13%.
This 13% difference is what we're analyzing statistically to understand if it's significant or just due to random chance. It's not just about the numbers, rather what they mean in a larger context that’s important.

A statistical test is a formal method that helps us decide whether a significant change or effect exists between two groups. When comparing the proportions of Facebook users between 2010 and 2013, we would utilize a statistical test to tell us if the observed change is meaningful enough to conclude an actual trend.
Statistical tests hinge on the idea of null and alternative hypotheses:

• The null hypothesis assumes no real change or difference (e.g., the increase in Facebook usage might just be random).
• The alternative hypothesis suggests a real difference (e.g., the increase signifies a genuine rise in Facebook usage).

By performing a statistical test, we assess which hypothesis is more plausible based on probability. If the test shows that the probability of the observed difference as a random event is low, we might reject the null hypothesis and consider the change significant.

In statistics, sampling refers to selecting a smaller segment from a larger population to make conclusions about the whole. For the Pew Internet survey, they likely took a sample of American adults to represent the entire population. Sampling is crucial because surveying everyone is usually impractical or even impossible.
There are key concepts related to sampling that ensure reliability:

• Random Sampling: Every individual in the population has an equal chance of being selected, minimizing bias.
• Sample Size: Larger samples tend to provide more reliable estimates of the population.

Proper sampling is essential because it ensures the survey's findings are reflective of the broader population rather than being skewed by outliers or peculiarities in a smaller, non-representative group.

Random fluctuation refers to variations in data that occur by chance rather than being caused by a specific, identifiable factor. For instance, if we were to flip a fair coin 100 times, we expect 50 heads on average. However, some random fluctuation might mean getting slightly more or fewer heads. With survey data, some variability is expected even without any actual change.
In the context of the exercise, we must distinguish between changes in Facebook usage that are merely random fluctuations and those that indicate a substantial and genuine trend. Statistical tests help differentiate random fluctuation from meaningful change. If the observed increase in usage from 2010 to 2013 is deemed significant, it's likely more than just random noise, reflecting an actual shift in user behavior.