biostats exam part 5

describe replication in each of these “trials”

1. top: not replicated → there’s only one experimental unit per treatment

2. middle: not replicated → plants confined to each chamber have the same conditions (not independent) 

3. bottom: yes replicated, eight replicates

1. what effect does sample size have on standard error and power? 

1. larger sample size → reduces standard error and increases power

1. what is pseudoreplication?

1. when measurements are NOT independent from each other

2. they are analyzed as if they were

1. what is temporal pseudoreplication? 

1. take several measurements on the same individual over time and treat data points as independent

1. what is spatial pseudoreplication? 

1. take several measurements around the same location and treat data points as independent

1. what is a balanced design? 

1. all treatments in the study have the same sample size

1. what does balanced design help to do? 

1. reduce error and increase power

1. what is blocking?

1. divide experimental units into groups (blocks) with similar characteristics

1. what is the purpose of blocking? 

1. block the effects of a confounding variable

1. example of blocking from the lecture

1. research question: does running a 5K marathon prior to a test improve test performance? 

2. separate into student athletes, students who work out, and student athletes

3. randomly assign treatments to the groups

what is the primary benefit of blocking?

a. removes variation between groups to allow treatment differences to be better estimated

b. removes variation between individuals to allow group differences to be better estimated

c. removes variation between individuals to allow treatment differences to be better estimated

d. removes variation between treatments to allow group differences to be better estimated

a. removes variation between groups to allow treatment differences to be better estimated

1. what is the purpose of matched pair design? 

1. reducing influence of sampling error

2. like blocking, also meant to reduce effect of unrelated variation

1. what is a matched pair design? 

1. subjects are matched based on key variables, and then one is assigned to a treatment group and the other to a control group

1. what are the main similarities and differences between blocking and matched pair design? 

1. both meant to reduce the effect of confounding variables

2. matched pair design is only about two treatments, blocking can address more than that

1. what kind of study is matched pair design useful for? why? 

1. observational studies

2. we cannot do true blocking

1. what is extreme treatment? 

1. helps detect treatment effect

2. giving one group an extreme treatment to compare effects

1. what is factorial design? 

1. evaluate possible interactions between variables

1. what does an interaction between two explanatory variables mean? 

1. the effect of one variable on the response variable depends on the state of a second variable

Consider this figure indicating the results of an experiment growing a marine invertebrate in four different environments based on values of two factors. Two different temperatures and two different salinities were used. Which of the following is incorrect?

1. the effect of low versus high temperature depends on the salinity

2. the effect of low versus high salinity depends on the temperature

3. there is no interaction

4. the two factors interact with on another

3. there is no interaction

1. things to consider when choosing sample size: 

1. precision

2. power

3. time

4. money

5. health hazards

1. precision means we have a ________ confidence interval

narrow

1. if you want to be precise in your study, what margin of error should you have? 

1. small

2. small confidence interval

3. baseline of 95%

1. what does increasing sample size do to power? 

1. increases power

1. what is data dredging? 

1. aka data fishing, data snooping

2. carrying out of many statistical tests in the same set of data in hope of finding at least on statistically significant result

1. what does data dredging do to the error? 

1. increases the chance of Type I error (obtaining a false positive)

1. what is the Bonferroni correction? 

1. apply to correct for multiple comparisons 

2. a* = a/number of tests

when do you use the Bonferroni correction?

where running each test at a standard α = 0.05 dramatically inflates your overall chance of a false positive

what’s the downfall of the Bonferroni correction?

drastically reduces statistical power

increases the chance of Type II errors (false negatives)