Clear IV and DV Estimating population parameter · When standard error is known, the population value is estimated to be: For 68% confidence: sample statistic +/- 1 standard error For 95% confidence: sample statistic +/- 2 standard error For 99% confidence: sample statistic +/- 3 standard error What if you don't know the standard error? Population parameter= sample statistic +/- 0.5 When you're only given this info: Sample value = 30 hours of watching tv per week You want to be 90% confident about the population value · Sample value +/- 1-confidence level/2 · 30 +/- 100%- 90%/2 = 10/2 = 5 (90% is our confidence, so 100-90=10) · 30 +/- 5 · 25-35 · Estimating that based on our sample of 30 hours, the population value is between 35 and 35 hours Estimating population parameter A researcher found that 65% of UBC students in a sample undergo stress during the midterm season. What can she infer about the UBC student population with 95% confidence? Statistic- 65 Population 95% = 1-c/2 =100-95/2. = 5/2 Sample +/- 2.5 65+2.5 65-2.5 = 67.5 - 62.5 TESTS OF STATISTICAL SIGNIFICANCE . . If lambda error is 1.3 or -5, that's an error because lambda only varies between 0 and 1 2 variables cant move within more than 100% of each other Regression analysis · All we can do is look at interval or ratio variables. Numerical variables Number of hours spend studying and grades in Soci 217- 2 numerical varibles · Saying one causes the other . ID number of hours studying (x) . DV grades in soci 217 (y) . Formula for straight line- y= a+bx . Assuming even if you don't study at all, nobody will get less than 5%, so "a" (constant) is 5% Y(DV)=a(constant)+b(coefficient)x(IV) · +/- 0.5x is going to casue y to either increase or decrease · Always +e because will never capture perfectly Multiple regression
. There are lots of things not on the straight line, never perfect, will always have errors · Null hypothesis- no relationship, generally want to prove wrong and find relationship ON THE FINAL* State the null hypothesis 100% of the time- the answer is always going to be there is no relationship between IV and DV, that cannot change · Is the variable statistically significant? Only if there is an (*) - meaning we do not know the chance that the relationship is due to errors. If we don't know- its not statistically significant. We will always know based on the *. If the question is what is the relationship between a and b and there is no * the answer is there is no relationship Regression equation · On worksheet · R square R2- variance explained OR coefficient of determination R2 tells us how good all the variables we have in the model, how much of the DV they explain. How good are they at explaining the DV. How much of the DV is explained by all our IV's together ·