Transcript Statistics
Statistics • Sample: Descriptive Statistics • Population: Inferential Statistics Descriptive Statistics • Measures of central tendency – Mean – Median – Mode • Measures of variability – Standard deviation (SD) – Variance (SD squared) • Correlation – measure of relatedness Populations and Samples • Population: all the possible scores – All living humans – All possible rolls of a pair of dice • Sample: a subset of the population – Random sample – best kind; unbiased • Relationship of sample mean to population mean – example: number of pets Example of Comparing 2 Means • • • • • DV = # of minutes balancing on one foot IV = whether blindfolded or not Hypothesis: Lack of visual cues decreases balance Means of two samples: 2.6 yes, 5.1 no Confirmed hypothesis? – Is the difference due to chance? – How can you tell if a difference found in an experiment is due to chance? – How often do differences this big occur by chance? Tests of Significance • t • F • Tell us how often we would get the results we see in our sample just by chance Hypothesis Testing • H0 : the null hypotheis – Mean1 = Mean2 – Experimental Condition = Control Condition • H1 : the research hypothesis – Mean1 does not equal Mean2 – Experimental Condition does not equal Control Condition Hypothesis Testing: Logic • Assume H0 • Calculate a test statistic (such as t) to get p (p = how often a difference this large would occur by chance) • If p < .05, conclude that H0 is false (“reject H0”) – If you reject H0, you are left with H1 • Conclude that the means are really different • Say “the difference is statistically significant.” – If p > .05, you can not reject H0 • Say “the difference was not statistically significant.” Testing Hypotheses • Are the means for the two groups really different? (Blindfolded vs. not) • Steps to Follow: – Check for invalid data (analyze-> descriptive statistics -> explore) – Check for outliers; remove outliers – Compare means with t-test (analyze -> compare means) • Independent Samples: for 2 groups • Paired Samples: for repeated measures (one group)