LECTURE ONE/EDX MODULE ONE: Sources of variability: · Error variability: o Natural variability - due to chance or random variability (ie, different lengths of forearms). o Measurement variability - due to measurement errors (ie, different understandings of where a forearm starts and finishes). · Group variability (ie, difference between females and males). · Sampling variability Different types of variables: · Data - observations and measurements which have been collected in some way (often through research). Data are usually organised by variables and observational units. . Variable - a characteristic or an attribute that you are observing, measuring and recording data (ex. height, weight, eye colour, blood pressure, age, sex, etc.) Variable Quantitative (or numerical) (represent measurements - e.g. Weight) 0 Discrete Continuous Categorical (represents groups of objects with a particular characteristic - e.g Gender) Nominal Ordinal Continuous variable - variable that is measurable, can have any value over some range, includes numerical values with decimal places and can be counting numbers e.g: Height - 5.2", 5.55", 5", 6", 5.11", 5.9" ... Discrete variable - variable that can have only whole counting numbers such as 0, 1 2, 45, 907 and so on e.g: Number of phone calls - 1, 5, 8, 2, 15 ... Nominal variable - the groups or categories do not have an order. e.g. marital status - Never married, married, divorced Ordinal variable - categories or groups have an order. e.g. Grades for STAT1201 - HD, D, C, P, F; Satisfaction level Observational Study The researcher observes part of population and measures the characteristics of interest. Make conclusions based on the observations but does not influence to change the existing conditions or does not try to affect them. Example: Examine the effect of smoking on lung cancer Language of hypothesis testing: · Two types of hypotheses: Experimental Study The researcher assigns subjects to groups and apply some treatment(s) to group(s) and the other group does not receive the treatment. Can be designed as a blind or a double-blind study. When an experiment involves both comparison and randomization then we call it as a randomized comparative experiment. Example: Examine the effect of caffeinated drinks on blood pressure
Null hypothesis (Hg) Usually a statement of "no effect". Also, refer to the status quo (no change from the past, the old standard still correct). Either reject or do not reject Ho In our caffeinated drink example, the null hypothesis is as follows. Ho: the mean increase in pulse rate is the same for caffeinated and decaffeinated drinkers among young adults (or caffeinated drinks has no effect on pulse rate among young adults) Alternative hypothesis (H1) Usually a statement of "an effect". Also refers challenges to the status quo (something new is now occurring compared to the past). If we reject Ho we conclude there is sufficient evidence to accept the alternative hypothesis. In our caffeinated drink example, the alternative hypothesis is as follows. H1: the population mean increase in pulse rate is higher for caffeinated drinkers (or caffeinated drinks increase