Descritpive Statistics - Sem 1

Week 2 statistics Descriptive qualitative research Quantitative research - explaining a phenomena atta analysed using methods - primarily exploratory mathematical (statistics) - help us gain understand-ing of underlying & reasons investigates quantifiable information - - (ex Opinions motives) · methodologies : Surveys,asks what - does statistics Typical - , -Methodologies help us - want to we a Ideally can't generalize test sample from the sample characteristics This - focus groups Research Scenario Normally we population Since take intervie , a with ! our a - : is known of as an the entire population , population general population matches the a findings the representative sample we to statistics Descriptive Describes - tells Only - the took part us in data our sample sample (people who study) . Not necessarily the about our in our population (everyone else) Inferential - Uses the inferences - helps statistics sample us about the reach Needs to make - of the data sample Example study and helps us make population conclusions assumptions beyond about our data representativeness (things variables Independent - (IV) - It it - have - people or not left/right close Power Feelings of to Ability see fiscation symbol - the IV outcome the depends eye The (PV) input the on (IV) - based change can on have people power pose (DV) Dependent controls researcher The vary) that what DV is we measure Experimental Design Between subjects - When the sample is divided into and each groups group receives one performs - task - shorter participants statistical power. duration , receive every condition of same task - Good to number - a . ants are study in the treatment/perform I prevents carryovea and fatigue All study Requires larger number of participants for higher - one treatment subjects within when use of or participresources limited between-subjects design The - - - made Independent groups design-groups different people are Difference in groups be natural can Advantages : (groups) manipulated/allocated or not is influenced by do only Boredom/ Fatigue participants condition - of subjects between performance Performance up - one - No - No practice effects/experience effects from over carry conditions - previous don't the conditions each other Sometimes it's the - - only option within-subjects (repeated The Repeatedly affect the measure same measures) design people the on same DV Main - advantage control -To : ex . natural act (ppts - Two prouderes Advantages - - - More more : - : - - Same Different individual ability memory as time their : own One convenient to differences , control) session-all conditions times participants characteristics ppt compared powerful requires for to run fewer participants not themselves a problem- Power · probability The - that find actually will you significant effect mull hypothes correctly reject the - variation vs - · - reducing find 'real' a Low error Fewer part due to designs different diff individual makes it differences more likely you'll powerful more . Btw - The = - is result variance - - by variance error different caused statistically exists-i e experimental manipulation the variance variation - by caused error . when it a is - - sizes power demands subjects within = part zo noticeable subs within 40 = subs more Btw different sample require subs = = 100 20 . part . in . part part more complex research : Convenient · In - within - 1) a Between choose - subjects same choose - 2) Session-10 Each - - subjects-hesp Between a test category-memory word a - one part 10 diff . session people doing completing all 4 word within method Experiment , 3) Generate data 4) Analyse word category session design or one observation , tests , etc . categories levels Data of variables NOIR Nominal - categorical mostly - variables categories of things names - - measure frequencies through smoker/non-smoker - names , percentages , or counts female/male , Ordinal Data - - - the same we can'tonsume Likert scales - but intervals ordered is , position Interval - but true no cro in it can is smarter than distance - has - - so a is it data true true ratio weight There btw has It - order be not equal intervals negative necessarily 75 10 2x score (continuous data) Ratio - not Temperature - - at are 150 FO - than is race a meaning , cannot hold ratios - more data points between distance - anything point btw each a can , a height , points zero (20kg test , is equal intervals meaning it can't be negative are at 2 times scores ratio between the values hold ratios 10kg) - levels Data determine can perform the same - - - Describes tells only - - tests they that data. on require question the us data about in our sample descriptive stats are observed information types statistical type of data=observations 2 because different can you tests to answer stats Descriptive - important data levels different - - the are , sample population not the of participant numbers scores that on summarize : Measures of Measures of central tendency dispersion spread variables variation the Mean - Mode - They They may For - value are Ex use it values our data set statistical models are not (can single represent models - "typical" on variable is a (trimmed) are that - tendency Median - - a : Types - - what Describe - central of Measures be smaller one represent to if of version accurate reported our more modes multiple of data simpler particularly no . , be a score data of the 7 . data 57 , we they , simply Median used - - - (the middle) when observations ascending order (low to high) Central value ordered in - even odd . 2 with of no no · ordinal data obs--add up divide of obs--the 345067 median is or 5 slaved data 2 scores in by are the middle & 2 middle value 234567 median is 5 = 4 5 . Mean - - Add of up all Most can > - used be O I 1 it as 0 I All 0 · · median uses mean-may the - need we lot of information ! ? weaknesses on one value data points extreme values be affected only 1-2 by useful for not select to not the best measure mode is measure , - a 12 mode=fouces = but it's ignores have their measures · score or nominal for Which is the best - the number = value common problematic - by (Modal value Mode - divide values ) - and scores the measure categorical data that best describes data To do this you look athow need to the know the data level and data is distributed distributions Data > - distributed Normally extreme symmetric few , are close together majority of - Bell > - shaped curve skewed majority of at - values in middle the - ends on the scores The one end mean longer scores scores are is no central extreme affect the , mean To extreme scores Trimmed mean "trim' the data Idea : by removing on problem of solve the both extreme sides distribution Typical : (take off 5 5 % 5 of trimmed % of scores the mean lowest & % highest scores) mean being affected by Measures - measures or Range Interquartile Range - are (FOR) variance - standard driation - Range - - - The - between difference For Range also values (as it skewed date we highest the weakness is extreme - extent the data values spread out to what concentrated - spread of dispersion range of of - takes into & middle need diffe the upper 50 % & lower % of the values 3rd quartile and compute (the 2nd respectively) the range account Iow scores Remove 25 the lowest value strongly affected by is Interquartile Range (FOR) - & , measure only hight values) variance - - The the now - average squared : subtract the - find s the from scores from each score (xof these /divide by mean average 2(xi- *) 2 = n - = 2 = . 432 + 1 2 7 dividing by n-1 from measure statistics to - ) n-1) * mean Y 1010109842 S2 - of mean - if duiation . 432 10 = ....... . = 7 57 . 62 1 makes it possible our sample later make claims about to use in inferential the variance the variance in the population - The square the driations always - This is mean be is because from zero . the : No plug matter in mean [(Xi - E) called the balance for of numbers the Xi these , you summing up numbers will point characteristic of - - - standard duiation square useful the - - Easy root of measure the of variance the , SD spread of = 52 around values mean to interpret : SD has the I salary same is in E unit as instead If the data is normally distributed , people's scores fall within ISD of then the data of E2) 68 % of the mean NOTE - Although one group may differ from greater measure of untual tendency etc. ) that there is spread - - This a could just be tell us mean descriptive stats don't between sample - actually around the difference we , is difference large) these groups can be to the population need inferential stats & the other this (have doesn't (especially if mean the by chance whether the difference inferred beyond (p) values a our