70 likes | 193 Views
Parametric hypotheses tests. Marek Majdan. Training in essential biostatistics for Public Health Professionals in BiH , Marek Majdan, PhD; marekmajdan@gmail.com. Principle of hypothesis tests. Null and alternative hypothesis
E N D
Parametric hypotheses tests Marek Majdan Training in essential biostatistics for Public Health Professionals in BiH, Marek Majdan, PhD; marekmajdan@gmail.com
Principle of hypothesis tests • Null and alternative hypothesis • Statistical tests used to reject or accept these hypotheses and infere results from a sample to the population • Every test has his own hypothesis • Choice of test depends on data Training in essential biostatistics for Public Health Professionals in BiH, Marek Majdan, PhD; marekmajdan@gmail.com
T test • One sample (compares sample mean with population mean) • Two independent samples (compares means of two samples) • Paired (compares means of repeated measurements in the same sample) • Null hypothesis: compared means are equal Training in essential biostatistics for Public Health Professionals in BiH, Marek Majdan, PhD; marekmajdan@gmail.com
T test in R • 2 independent groups t-testt.test(y~x) -where y is numeric and x is a binary factor • 2 independent groups t-testt.test(y1,y2)- where y1 and y2 are numeric • Paired t-testt.test(y1,y2,paired=TRUE)- where y1 & y2 are numeric • One samle t-testt.test(y,mu=3)– where mu is the population mean Training in essential biostatistics for Public Health Professionals in BiH, Marek Majdan, PhD; marekmajdan@gmail.com
ANOVA test • To compare means of three or more groups • Null hypothesis: the compared means are equal • In R: summary(aov(variable~grouping variable, data=database name)) Training in essential biostatistics for Public Health Professionals in BiH, Marek Majdan, PhD; marekmajdan@gmail.com
Post hoc test • Anova only tells us that at least two of the compared means are equal • Post hoc test compares mean of each group with each • Tukey honest significance test • results=TukeyHSD(aov(ar_lv~sampleNO, data=zoltan));results • Other tests: paired t test with corrections Training in essential biostatistics for Public Health Professionals in BiH, Marek Majdan, PhD; marekmajdan@gmail.com
Comparing proportions • Chi squared test to compare proportions between categories of a 2x2 table, 2xN table or MxN table • In R: chisq.test (table) prop.test (table) Training in essential biostatistics for Public Health Professionals in BiH, Marek Majdan, PhD; marekmajdan@gmail.com