Data Analysis in Quant Research: Understanding Stats PART 2

Question 1

Which type of analyses:

*one DV and one IV

Answer 1

univariate

EX: calculating prevalence OR calculating incidence

Question 2

Which type of analyses:

*one DV and two IVs OR two DVs and 1 IV

Answer 2

bivariate

Question 3

Which type of analyses:

*one DV and multiple IV

Answer 3

multivariable

EX: multiple regression analysis examines a number of IVs and their combined relationship to a single DV

Question 4

Which type of analyses:

*multiple DVs

Answer 4

multivariate

multiple outcomes

Question 5

Between Group OR Within Group?

*Intervention group compared to control group

Answer 5

Between Group

AKA Between Subject OR Independent Samples

Question 6

Between Group OR Within Group?
*One group, but looking at pre- and post-;
same group but at different times

Answer 6

Within Group

AKA Within Subject OR Correlated Samples

Question 7

Nonparametric Statistics:
Select one: Chi-Square, Mann-Whitney, Kruskal-Wallis, Friedman
* Compares actual number of frequency in each group with expected number; between group design, uses NOMINAL level data

Answer 7

chi-square

EX: Research question: Is there a difference between 2 groups related gender?

Question 8

Nonparametric Statistics:
Select one: Chi-Square, Mann-Whitney, Kruskal-Wallis, Friedman
* Compares two groups; between group designs; uses ORDINAL level data

Answer 8

Mann-Whitney

EX: Research question: Is there a difference in breast self-examination based on educational level?

Question 9

Nonparametric Statistics:
Select one: Chi-Square, Mann-Whitney, Kruskal-Wallis, Friedman
* Compare greater than two groups; between-group designs; uses ORDINAL level data

Answer 9

Kruskal-Wallis
(EX: Research question: Is there a difference in the medians for change in the number of days of having cold symptoms among those who take placebo pill, those who take low doses of vitamin C, and those who take high doses of vitamin C?)

Question 10

Nonparametric Statistics:
Select one: Chi-Square, Mann-Whitney, Kruskal-Wallis, Friedman
* within-subject design; 1 group measured on 3 or more different occasions; uses ORDINAL level data

Answer 10

Friedman
(EX: Research question: Is there a statistically significant difference in perceived sense of well-being survey scores for individuals with clinical depression more from city to rural areas (collected prior to move, 3 months after move, and 6 months after move?)
(EX: our TBI research review)

Question 11

Parametric Statistics:

There are different forms of this test, used for different purposes. OT does not really use this type.

Answer 11

t-test

Question 12

Parametric Statistics:
Select one: Unpaired t-test OR Paired t-test
* uses continuous level data; for examining the difference between 2 groups (e.g control/treatment); comparing the mean of one group against the mean of another group on one or more variables

Answer 12

Unpaired t-test

EX: Research question: Is there a difference between treatment and control group in relation to age?

Question 13

Parametric Statistics:
Select one: Unpaired t-test OR Paired t-test
* also known as “t-test for independent samples” or “t-test for group means”

Answer 13

Unpaired t-test

Question 14

Parametric Statistics:
Select one: Unpaired t-test OR Paired t-test
* uses continuous level data; generally used when measurements are taken from the SAME SUBJECT before and after some manipulation such as a therapeutic intervention

Answer 14

Paired t-test
(EX: Research question: Is there a difference in mean score of knowledge of breast examination from pre-test (mammogram) to post-test?)

Question 15

Parametric Statistics:
Select one: Unpaired t-test OR Paired t-test
* also known as “dependent t-test” or “t-test for correlated samples”

Answer 15

Paired t-test

Question 16

Parametric Statistics:
Select one: ANOVA, One-way ANOVA, or Two way ANOVA
* Examines the difference between group means; three or more groups when participants are being tested only once; there is 1 categorical IV with 3 or more levels or conditions (or 3 groups) and 1 continuous DV

Answer 16

One-Way ANOVA
(EX: Research question: What is the effect of therapist counseling, compared to nicotine patch and no treatment on number of cigarettes smoked daily? DV = number of cigarettes, IV = interventions/control group)

Question 17

Parametric Statistics:
Select one: ANOVA, One-way ANOVA, or Two way ANOVA
* AKA Simple ANOVA

Answer 17

One-way ANOVA

Question 18

Parametric Statistics:
Select one: ANOVA, One-way ANOVA, or Two way ANOVA
* used when comparing 3 or more group means (e.g. the average performance of 3 or more groups); statistic computed is F-ratio

Answer 18

ANOVA: Analysis of Variance

Question 19

Parametric Statistics:
Select one: ANOVA, One-way ANOVA, or Two way ANOVA
* Examines the difference between two IV (2 or more levels) and one continuous DV; comparison of each IV with the DV; also allows for a test of the impact of the interaction between the 2 IVs on the DV (the “interaction”)

Answer 19

Two way ANOVA

Question 20

Parametric Statistics:
Select one: ANOVA, One-way ANOVA, or Two way ANOVA
* AKA Factorial ANOVA or multifactorial ANOVA

Answer 20

Two way ANOVA
(EX: Research question: Are counseling, compared to nicotine patch and no treatment equally effective on number of cigarettes smoked daily for women and men?)

Question 21

Parametric Statistics:
Select one: Repeated Measures ANOVA (within-subject), ANCOVA, MANOVA, or MANCOVA
* measurement taken at more than 2 time points; 3 or more measures for 1 DV; DV must be continuous

Answer 21

Repeated Measure ANOVA (within subject)
(EX: Research question: What is the difference in depression level of clients receiving mental health services by participating in group therapy sessions at pre-test, 6 months, and 1 year?)

Question 22

Parametric Statistics:
Select one: Repeated Measure ANOVA, ANCOVA, MANOVA, or MANCOVA
* purpose is to equalize differences between groups by controlling for a potentially confounding variable (the covariate)

Answer 22

ANCOVA: Analysis of Co-variance
(EX: Research question: What is effect of training on the running speed of adolescents who do or do not work out regularly? IV = training; DV = running speed; covariate = gender; To compare response to training vs. no training, one may need to control for known differences between males and females; ANCOVA removes the gender factor, so differences between training and no training are not corrupted by it)

Question 23

Parametric Statistics:
Select one: Repeated Measure ANOVA, ANCOVA, MANOVA, or MANCOVA
*used when two or more dependent variables (DV) that correlate with each other are compared to one or more independent variables

Answer 23

MANOVA: Multivariate Analysis of Variance

ex. examine effect of two methods of exercise on diastolic and systolic blood pressure

Question 24

Parametric Statistics:
Select one: Repeated Measure ANOVA, ANCOVA, MANOVA, or MANCOVA
*used when two or more dependent variables (DV) that correlate with each other are compared to one or more independent variables EXCEPT also controls for confounding variables (covariates)

Answer 24

MANCOVA: Multivariate analysis of Co-variance
(Ex. Research question: what is the effect of different interventions (desensitization, relaxation training, wait-list control) on several types of anxiety (measured with 4 sub scales of anxiety measure) after controlling for age and marital status)

Question 25

Statistical procedure to predict the outcome of a dependent variable based on modeling of an independent variable or variables

Answer 25

Regression

Question 26

Creates a regression line (i.e. line of best fit) that is the best predictor of the relationship between the IV and the DV; one IV used to predict a DV

Answer 26

Simple linear regression

Question 27

Statistical procedure designed to test the relationship or prediction between two or more IVs and one DV.

Answer 27

Multiple regression

Question 28

Provides measure of overall fit that is the best predictor of the relationship between the IVs and the DV

Answer 28

Multiple linear regression
(*Research question: how well does # of smoking cigarettes a day, # exercise per week, and cholesterol predict mortality from CHD?)