Test practice problems

When there is a perfect, direct relationship, what is the value of r?

1.00

“It is impossible for a relationship to be both inverse and strong” This statement is

False

What values of r represents the strongest relationship?

0.00

An r of -0.45 represents a stronger relationship than an r of 0.88"

False

It is appropriate to convert a Pearson r to a percentage by multiplying it by 100.

False

The symbol for the coefficient of determination is r

False — it is R²

For a given value of the Pearson r, the coefficient of determination is computed by:

squaring r

For an r of 0.60, the ability to predict 36% better than zero

True

When r = 0.20, the percentage of variance accounted for is 20%.

False

If participants are randomly assigned to experimental and control groups without matching, the resulting data are:

Independent

When reporting the results of a t-test, the values of the means and standard deviations should first be reported:

True

If a t-test yields p > 0.05, the null hypothesis normally would be rejected

False

If a researcher concludes that a difference is statistically significant, what else is true?

The null hypothesis should be rejected.

“statistical significance is synonymous with practical significance.”

False

For nominal data, a researcher normally reports means and standard deviations

False

“The purpose of effect size is to determine the statistical significance of the difference between two means.”:

False

You have been asked to study the use of music therapy to decrease anxiety in end-stage cancer patients.  You are interested in whether the number of hours of music played (x) predicts their decreased anxiety as represented by an anxiety scale (y).  You find the following regression equation: y = 3.0 - 0.4x.  If someone has played 10 hours of music, how much would you expect their anxiety to decrease?

-1 point because y = 3.0 - 4.0(10hrs), which is -1

What is the difference between sampling error and sampling bias?

Sampling error is a normal occurrence—random error. Sampling bias is a systematic problem of who is in your study and that means you are off from your focus population.

How are sample size and effect size related?

Large effect size does not require a large sample size. A small effect size requires a larger sample size.

Describe power and its component parts

Power is a combination of the effect size, chance of making a type 2 error, and sample size

What is beta?

Probability of making a type 2 error

Provide an example of Type I bias

Saying there is a significant association where there is nothing. (Telling a man he is pregnant).

Provide an example of Type 2 bias

Saying there is no association when there is one (telling pregnant woman she is not pregnant).

What is a case control study?

A study that starts with the outcome

A relative risk of 2.46 (95% CL 0.86-2.72) indicates what?

That there is no association between the exposure and the illness

Your study reports a relative risk of 0.24 with a p-value of 0.03, You know the exposure interest is:

a protective exposure

A higher absolute risk difference decreases the number needed to treat (NNT)

True

You enroll 45 women with ovarian cancer in your study and ask them about their smoking status over the last ten years. This data will allow you to calculate the relative risk of developing ovarian cancer if you smoke.

False

A pedestrian jaywalking at rush hour has a high ____ risk of being hit by a car.

Absolute

The prevalence of influenza decreases in your community. How would this affect the positive predictive value of nasal swabs to detect influenza? 

It decreases the positive predictive value

Pearson’s correlation coefficient is used for multiple comparison groups and can include ordinal variables

False

Logistic regression generates an odds ratio, or the odds or probability of the outcome occurring versus not occurring

True

The value of the coefficient of determination is that you can calculate the percentage of variance of one variable explained by another

True

Linear regression is conducted with a nominal dependent variable and an interval-ration independent variable

False

Multiple regression is used when you want to predict a dependent variable with more than one independent variable

True

A statistically significant correlation is equivalent to causation.

False

When a correlation coefficient is negative, the interpretation is:

The relationship between the variables is inverse or negative

A statistically significant regression coefficient represents clinical significance

False

The adjusted r² is used to reduce the chances of overestimating the amount of variance explained by the model.

True

Spearman correlation includes one ordinal variable and one interval/ratio variable from a single population

True

An F-statistic is calculated by

Dividing the mean squares between groups by the mean squares within groups

When you find a statistically difference in ANOVA, the results tell you immediately which groups have different mean values

False

What level of measurement are the variables in an ANOVA

Nominal or ordinal and interval

ANOVA is used for comparing ordinal and nominal variables

False

Repeated measures ANOVA can increase the power of the study and decrease Type II error when used appropriately

True

Difference in height between 4th grade and 5th grade students

Independent T-test

Comparing the use of e-cigarettes (yes/no) and drinking (none, sometimes, frequent, daily) among high school students.

Chi-square

Comparing weight gain among pregnant women in 3 age categories.

ANOVA

Mean A1c level over three data collection points for one group of adults.

Repeated Measures ANOVA