Flashcards
Management Sc Session 5 & 6
Cards in this deck (24)
- 1What is confirmation bias?Confirmation bias: If the data is there, people often use it in a biased way to support whatever preconceived opinion they have
- 2"accept that the treatment worked if the number of 6’s is larger than a cut-off value c" Do we do this before, during or after we have analyzed the data?before. to avoid confirmation bias
- 3All statistical tests have two elements - name them1. A measure T, calculated from the data, that allows us to assess the strength of the evidence for the claim
2. An acceptance criterion specified by cut-off value(s) c . One-sided tests (covered in this class)
Accept the claim if T > c : the larger T the more evidence for the claim Accept the claim if T < c: the smaller T the more evidence for the claim - 4What is a false positive error?Testsaysthatthe managerial intervention worked but it actually didn’t
- 5What is a false negative error?Testsaystheintervention didn’t work but it actually did
- 6We should focus on avoiding what kind of error?false positive
- 7significance level = a test procedure’s probability of producing a______ errorfalse positive;
, i.e., of producing a positive test result when there is no effect - 8The significance level is chosen ______ we see the data – The p-value is calculated ____ we see the data – it is a summary statistic of the databefore; after
- 9A statistical test only controls the probability of a false ______ error The probability of a false ______ error can be very largepositive;
negative - 10The chance of a false negative can be as large as_________ of the test1- significance level
- 11For single variables, the most important graphs are: (name 3)Histogram
Percentile Curve
Line Graph - 12For multiple variables, the most important graphs are:scatter plots
- 13Since intercept and slope are statistics, they have ____sampling distributions
- 14A high correlation coefficient is always:an indication of a strong relationship
- 15A low correlation coefficient isnot an indication that there is no relationship
- 16The significance level tests the claim: “The slope is not zero”......against the null hypothesis
“The slope is zero” - 17The larger the unknown slope is, the more likely it is thatthe data will provide evidence that there is a slope
- 18A very small slope can be highly significant in one of the following two situations:A very small slope can be highly significant in one of the following two situations:
(a) you have lots of data (the standard error for the slope goes to zero as N goes to infinity)
(b) the population regression line fits the data exceptionally well (high R^2) - 19The confidence interval for the slope gives you information about:the size of the unknown slope of the population regression line
- 20—
- 21If two variables X and Y are correlated, information about X canimprove the prediction of the average of Y
- 22If an omitted variable is the reason for the correlation, then changing X:may not change the omitted variable and hence have no effect on Y
- 23R-square is often used as a “measure of fit” What does this actually mean?R-square can be interpreted as the percentage of variation in the response variable Y that is explained by the variation in the predictors X
- 24—
Information
- Description
- Flashcards on Management Sc Session 5 & 6, created by Georgia Tan on 15/10/2015.
- Created by
- Georgia Tan
- Views
- 8
- Tags
- Rating