• Home
  • Swinburne University of Technology
  • Foundations of statistics
  • Key Concepts and Techniques in Statistical Analysis

Key Concepts and Techniques in Statistical Analysis

StuDocu.com Exam notes sheet Foundations of statistics (Swinburne University of Technology) StuDocu is not sponsored or endorsed by any college or university Downloaded by Krystle Sky (krystle.sky@googlemail.com) ** ALL FIGURES TO 2 DECIMAL PLACES EXCEPT FOR P VALUE ** IF p < . 050 = significant. IF p >.050 = not significant. If p =. 000, report as p <. 001 ** DO NOT REPORT NEGATIVE FOR T VALUE ** ** PROBABILITY = P VALUE ** frequency PROPORTION = total number TOTAL OF ALL VALUES NUMBER OF NUMBERS MEAN X = MEAN DIFFERENCE = X a STANDARD DEVIATION = Sa Z SCORE Standardised value: number of standard deviations above the mean. Standardised Value=i data value - mean of pop Z= x-H standard deviation ? Below the mean = Negative score, Above mean = Positive score T SCORE Where standard deviation is unknown SIGNIFICANCE & SAMPLING In normal distribution, values fall within 95% (2.5 standard deviations of mean) Values which fall within this 2.5% are REGRESSION Y = a + b x X | X = I variable | Y = D variable | a = vert. intercept (constant). Describes average start size. I.e. new born guinea pig weights 50grams. a = 50. b = SLOPE /REGRESSION COEFFICIENT. Describes average change per period - ie, Guinea pigs weigh 70 grams more with each month of age. b = 70 Age 12 mth guinea pig = 50 + 70 x 12 (1440 grams) REGRESSION COEFFICIENT = Coefficients table [ IV] = X% OF THE VARIATION IN X CAN BE EXPLAINED BY THE LINEAR RELATIONSHIP BETWEEN X AND X NUSIANCE VARIABLES AND CONFOUNDING FACTORS Nuisance variables affect the dependent variable. Must design experiment to minimise the effects and/or evenly distribute the effects. Nuisance variables act as confounding factors. CONTROLLING NUISANCE VARIABLES Independent Groups Design: Random allocation to separate groups to undertake tests under different circumstances. I.e. Group 1 Driving with BAC of .00 and Group 2 BAC of .05. Does not minimise nuisance variables but distributes. Repeated Measures: Same participants undergo tests under different conditions several times. Introduces practice effects which sway results. Matched Pairs: Match participants with those very similar, then separate into groups to undertake tests under different circumstances. Time consuming and difficult to identify NV to match. OBSERVATIONAL = Natural and uncontrolled/uninfluenced observation EXPERIMENTAL = Manipulating the IV to observe effect on DV HISTOGRAM DISTRIBUTION Positively Skewed Symmetrical distribution Bimodal distribution Negatively skewed RELATIONSHIP STRENGTH [Pearsons Correlation] .75 and more = STRONG .45 - . 74 = MODERATE .25 - . 44 = WEAK .24 and less = EXTREMELEY WEAK IGNORE neg/pos, work from numerical value only COEFFICIENT OF DETERMINATION = r2. How much is result of correlation. PEARSONS R = r . Strength of relationship - can be distorted by outliers, should always check scatterplot. Can be used if scatterplot shows no curve. PVALUE FOR PEARSONS Correlations table[Sig] ** AVOID CAUSAL RELATIONSHIPS ** STRATIFIED RANDOM SAMPLING: Where one group outweighs the other and need to reflect appropriate proportions of