Psych Ch. 4

Front 1

Statistics in Psychological Research

Back 1

*Consistent, dependable, large effects are easy to interpret.

*But psychological effects are often small and fragile, or arise only under certain circumstances.

*Psychologists use statistics to help decide wether an observed effect is real or is due to chance.

*Psychologists generally use two types of statistics:
-Descriptive statistics
-Inferential statistics

*Why use stats? To determine if "difference" is significant.

Front 2

Descriptive Statistics

Back 2

*Mathematical summaries of results.

*Two broad categories:
-Measurements of Central Tendency: "The middle range."
-Measurements of Variation (also called variance or dispersion): How similar scores are (low variance), or how different they are (high variance).

Front 3

Central Tendency
(Descriptive Statistics)

Back 3

*Mean or Average, when normal distribution.

*Median, when distribution is skewed, (*must occur in the data set, i.e. real number*)

*Mode=Mean=Median, when normal distribution.

Front 4

Measurements of Variation
(Descriptive Statistics)

Back 4

*The range is a statement of the highest and lowest observed scores.

*The standard deviation (SD) is the average distance each score is away from the mean.

Front 5

Probability Values and Statistical Significance
(Inferential Statistics)

Back 5

*A probability value ("p" value) is a way to estimate the likelihood that a statistic from a sample is different from that parameter in the population, due to chance alone.

For example, if X and Y are correlated at 0.70 in our sample and p<0.05, there is less than a 5% chance that X and Y are reall not correlated in the population and that our 0.70 correlation is due to sampling error alone.

*A result from a sample that is unlikely to have occurred by chance is interpreted as being statistically reliable or statistically significant.

**Type II error: Saying there is no effect when there is one. (preferred)
**Type I error: Saying there is an effect when there is none.

Front 6

Confidence Intervals
(Inferential Statistics)

Back 6

*Sometimes we try to infer what some "true" value is in the population (such as the mean or a correlation), based on the value we observed in our sample.

*Confidence intervals state how certain we are that the true value (population parameter) lies within a certain range.

*Width of the confidence interval depends on sample size (ther larger the better) and the standard deviation (the smaller the better).

*Confidence intervals are typically reported at the 90%, 95%, 99% range.