Variables are things that we measure, control, or manipulate in research. Aria!les" are things that are measured, controlled, or manipulated. 'Ependent variables are the values that we measured e.g. (arks.
Variables are things that we measure, control, or manipulate in research. Aria!les" are things that are measured, controlled, or manipulated. 'Ependent variables are the values that we measured e.g. (arks.
Variables are things that we measure, control, or manipulate in research. Aria!les" are things that are measured, controlled, or manipulated. 'Ependent variables are the values that we measured e.g. (arks.
MODULE 1: Elementary Concepts in Statistics (Dr See Kin Hai)
1. What are aria!les"
Variables are things that we measure, control, or manipulate in research. #. What are the $i%%erences !et&een Correlational an$ E'perimental (esearch" 1. Empirical research consists of Correlational and Experimental research 2. In Correlational research we just measure the values of variables and look for their relation 3. in Experimental research, we manipulate some variables then we measure the effects of the treatment on other variables. ). What are Depen$ent an$ *n$epen$ent aria!les" Independent variables are those that are manipulated eg. Gender !o"s or Girls#, $%perimental and &ontrol Groups. 'ependent variables are the values that we measured e.g. (arks. +. Which are the $i%%erent meas,rin- scales" )he" are a# nominal, b# ordinal, c# interval or d# ratio scales. .. What are the properties o% /ominal 0aria!les" *se onl" for +ualitative classification and cannot +uantif" or rank the variables $%amples are race, gender, colour, cit" etc. 1. What are the properties o% Or$inal 0aria!les" ,llow us to rank order the items we measure. $%ample of -$- *pper, (iddle and .ower class 2. What are the properties o% *nter0al 0aria!les" ,llow us to rank order and also to +uantif" and compare the si/es of differences between them. $%ample of temperature scales of o & or o 0 3. What are the properties o% (atio 0aria!les" -imilar to interval variables but the" have absolute /ero point. $%amples are 1elvin temperature scale, time and space. 4. What are the !asic %eat,res o% e0ery (elation !et&een 0aria!les" )he most elementar" properties are the relation2s a# magnitude si/e# and b# its reliabilit" 1. Magnitude (size) is easier to understand and measure. 0or e%ample, if ever" bo" in our sample was found to have a higher score in -cience than the girl, we can sa" that the magnitude of the relation between the two variables Gender and -cores# is ver" high in our sample. 2. Reliability (truthfulness) of a relation is ver" important as it pertains to the representativeness of the result found in our sample for the entire population. 3e use a standard measure called p4 value of the significance level. 1 15. What is statistical si-ni%icance (p60al,e)" 1. It is the probabilit" that the observed relationship between variables or a difference between means in a sample occurred b" pure chance and that no such relationship or differences e%ist in the population. 2. 5r we can sa" that the statistical significance of a result tells us something about the degree to which the result is true or representative of the population. 3. )he higher the p4value, the less reliable is the observed relation between variables in the sample. 6. p 7.89 indicates that there is a 9: 1;28# probabilit" that the relation between the variables found in the sample is a fluke. (eaning if the e%periment is repeated 28 times there would be 1 that is having stronger or e+ual relationship then the 1< cases. 9. p .89 level is treated as a border4line acceptable error level. p.81 level is considered statisticall" significant p.889 or p.881 levels are often called highl" significant 11. Stren-th 0s. (elia!ility o% a (elation !et&een aria!les 1. 0or a sample of a particular si/e, the larger the magnitude of the relation between variables, the more reliable the relation. 2. )he relation of a given strength could be either highl" significant or not significant depending on the sample si/e. If there are few observations then the probabilit" of obtaining b" chance a combination of those values indicative of a strong relation is relativel" high. $%ample if the sample si/e is 6 2 bo"s and 2 girls#, then we can find 188: b" chance the relation between Gender and -core where both bo"s have high score and both girls having low score. 3. -ignificance depends on sample si/e. 0or a large sample si/e even a ver" small relations between variables will be significant. 3hereas for a small sample si/e even a ver" large relations we cannot consider significant. 1#. Why the /ormal Distri!,tion is important. 1. , characteristic propert" of normal distribution is as follows= i# >?: of all its observations fall within the range of 1 -' from the mean ii# <9: within the range of 2 -' iii# <<.<<: within the range of 6 -' or the p @ .8881 2. (ost statistical tests are based on the normalit" of the distribution. $%amples are t, 0 or &hi4 s+uare tests. 3. If the variables are not normall" distributed then we can use nonparametric test distribution4 free test#. Aowever these tests are less powerful and fle%ible in terms of the t"pes of conclusions that the" can provide. In most cases we can still use the normal distribution4based test if the sample si/e is large n 738# then the distribution approaches a normal shape 0isher, 1<2?#.)his is called the central limit theorem. 2 6. 3e can use Monte-Carlo experiments to check the normalit" assumption or b" visual e%amination using a histogram as below. 3