Comparing samples, siginificance and experiements

Variation Sampling Error means…

• the natural statistical discrepancy between the sample statistic and the population statistic

How could we better approximate the population mean?

Take more samples and measure the mean of the new samples

Would the sampling distribution of means become narrower or wider as more samples are drawn?

• narrower

Reminder - sampling error is…

Deviation of the samples mean from the populations mean

However,

… it can be estimated!

Standard error meaning…

• estimates how much the mean might vary if the study was repeated

  • smaller = estimate more precise, more likely to be close to the true population mean

  • Large = estimate is less precise

Confidence intervals…

Range of values that likely contains the true population parameter, based on a random sample from that population

(Realistic buffer zone)

Null hypothesis significance testing…

• what we use to decide wether there is a difference, and whether we can generalise the finding obtained from a sample to the wider population

The test relies on two hypotheses =

1. Null hypothesis

2. Alternative hypothesis

1. Predicts no difference between the populations OR no Trend

2. The alternative hypothesis there IS a difference between the populations of there is a trend

When is the Null hypothesis rejected?

When the probability of the finding occurring under the null hypothesis is less than 5%

We fail to reject the Null hypothesis means there is …. Significant difference/trend

We do reject means there …. A significant trend

What % is the confidence threshold in social science statistics?

5% (convention not universal rule)