Research: Module 7 - Regression Flashcards by Chavez Girls

is a regression a type of correlation?

How well did you know this?

Not at all

Perfectly

regression

a statistical technique that predicts an outcome

How well did you know this?

Not at all

Perfectly

what type of data can regressions be run on?

the predictors can be interval/ratio, ordinal, or nominal

How well did you know this?

Not at all

Perfectly

predictors vs. outcome variables

predictor: the variable you use to predict or explain changes in another variable (IV)

outcome: the variable you are trying to predict or explain (DV)

How well did you know this?

Not at all

Perfectly

linear regression

one predictor (any level of data) and one I/R outcome

How well did you know this?

Not at all

Perfectly

multiple linear

multiple predictors (any level of data) and one I/R outcome

How well did you know this?

Not at all

Perfectly

logistic regression

**one or more predictors (any level of data) and one categorical outcome, 2 levels only

How well did you know this?

Not at all

Perfectly

multinomial logistic regression

one or more predictors (any level of data) and one categorical outcome, multiple levels

How well did you know this?

Not at all

Perfectly

ordinal logistic regression

one or more predictors (any level of data) and one ordinal outcome, multiple levels

How well did you know this?

Not at all

Perfectly

what type of regression is the following: Does GRE score significantly predict PT school GPA?

linear

one predictor (any level of data) and one I/R outcome

How well did you know this?

Not at all

Perfectly

what type of regression is the following: Does age predict days to clearance following a concussion in adolescent athletes?

linear

one predictor (any level of data) and one I/R outcome

How well did you know this?

Not at all

Perfectly

what type of regression is the following: Does disease category predict distance on the 6MWT?

linear

one predictor (any level of data) and one I/R outcome

How well did you know this?

Not at all

Perfectly

how would you convert this relationship hypothesis into a prediction hypothesis: There will be a significant relationship between time 1and time 2 at r >+.80 or r<-.80?

can time 1 ROM predict time 2 ROM?

what ROM value would be expected at time 2?

How well did you know this?

Not at all

Perfectly

linear regression equation

y=mx+b

m = regression coefficient
b = constant

How well did you know this?

Not at all

Perfectly

how would you calculate what time 2 would be?

for y=mx + b, you are solving for y

plug in ROM at time 1 for x

How well did you know this?

Not at all

Perfectly

what regression would you run for the following: Can sitting balance scores significantly predict
standing balance scores in patients following a CVA
in acute care?

linear regression

one predictor (any level of data) and one I/R outcome

How well did you know this?

Not at all

Perfectly

what are the 5 assumptions for regression?

1: data must be linear, not curvilinear (scatterplot)

2: data must have normality

3: homoscedasticty of variance

4: data must be free of influential outliers

5: data must be independent of each other

How well did you know this?

Not at all

Perfectly

what do you look at the assess if the data has normality?

histograms
skewness
kurtosis
boxplots
Shapiro-wilk test

How well did you know this?

Not at all

Perfectly

what is a major concern when meeting the assumptions for a regression?

Study These Flashcards

outliers

how do you tell if a data set is normally distributed when observing a box plot?

Study These Flashcards

The box is roughly symmetric around the median (the line inside the box).

The lengths of the whiskers on both sides are approximately equal.

few or no outliers

how do you tell if a data set is normally distributed when observing a histogram?

Study These Flashcards

Bell-Shaped Curve:
The bars form a peak in the middle and taper off symmetrically on both sides.

Symmetry Around the Mean:
The left and right sides of the peak are roughly mirror images.

Single Peak (Unimodal):
There is one clear central peak, not multiple peaks.

how do you determine if a data set is normally distributed based on kurtosis and skewness?

Study These Flashcards

the values should be inbetween -2 and +2

Shapiro-Wilk test

Study These Flashcards

statistical test used to assess whether a dataset is normally distributed.

we DO NOT want there to be a significant difference

p-value > 0.05 → Fail to reject H₀ → Data appears normal

p-value ≤ 0.05 → Reject H₀
→ Data is not normally distributed

Shapiro Wilk test calculates a _____ statistic

Study These Flashcards

small W value → data not normally distributed

Levene's test

a statistical test used to assess whether two or more groups have equal variances (homogeneity of variance) want this test to NOT be significant (p>0.05) used in ANOVAs, t-tests

homoscedasticty of variance

in relationship designs, the variance (the spread) of the outcome variable should be approximately the **same at all levels** of the predictor variable used in regression analysis

homogeneity vs. homoscedasticty of variance

**Homogeneity**: Equality of variances across groups (comparing two or more groups.) -ex: Comparing knee ROM between two therapy groups: both have similar variability **Homoscedasticity**: Variance stays consistent across all values of an independent variable (continuous predictor in regression.) -ex: Predicting knee ROM from weeks of rehab: ROM variance is steady across weeks.

heteroscadasticity

the spread of the data around the best fit line is NOT even

Cook's distance

used in regression analysis to identify influential data points, outliers that have a large impact on the fitted regression model. >1: outlier is potentially influential → eliminate participant

Durbin-watson test

way to check for independence of observations...checks for possible correlations between the participants which would violate out assumption ranges: 0-4 (2 is perfect - no correlation)

adjusted R-squared

a modified version of R-squared that adjusts for the number of IVs in the model it penalizes the addition of unnecessary variables

what regression would you run for the following: Does UG GPA and GRE score significantly predict PT school GPA?

multiple linear regression **multiple predictors** (any level of data) and **one I/R outcome**

what regression would you run for the following: Does age and sport predict days to clearance following a concussion in adolescent athletes?

multiple linear regression **multiple predictors** (any level of data) and **one I/R outcome**

what regression would you run for the following: Does disease category and quadriceps strength predict distance on the 6MWT?

multiple linear regression **multiple predictors** (any level of data) and **one I/R outcome**

multicollinearity

occurs in multiple regression when two or more predictor (independent) variables are highly correlated with each other. This means they provide redundant information DO NOT want this

what is considered a high correlation amongst predictors?

>0.9

what 2 variables are used to test multicollinearity?

variance inflation factor - should be <10 tolerance - should be >0.1

what regression would you run for the following scenario: We are interested in predicting CI scores from the final clinical internship....this outcome score ranges from 0 (poor)-100 (excellent). We are interested in two predictors....entered into a model together - The number of classes that the student received a grade of “A” – max is 35. - The average ratings from their past three CIs at the end of each clinical affiliation. (0=poor – 100=excellent)

multiple linear regression **multiple predictors** (any level of data) and **one I/R outcome**

what would the regression equation look if you have multiple predictor variables?

y = mx + mx + b x = predictor variables

what type of regression should you run for the following?: Does age and type of cardiac procedure (value repair vs bypass) predict discharge location to home vs. other?

logistic regression one or more predictors (any level of data) and **one categorical outcome**, 2 levels only

what type of regression should you run for the following?: Does TUG score predict fallers and non- fallers?

logistic regression one or more predictors (any level of data) and **one categorical outcome**, 2 levels only

what type of regression should you run for the following?: Does a score on the VOM (Vestibular Ocular Motor) Screening test predict return to sport in the next 30 days? (yes/no)

logistic regression one or more predictors (any level of data) and **one categorical outcome**, 2 levels only

a logistic regression uses a _____________ instead of a straight line

logistic curve

what type of regression should you run for the following?: predictors: lateral trunk sway during gait, family history, cervical protraction/retraction during gait, birth history outcome: spastic CP, hereditary spastic paraplegia

logistic regression one or more predictors (any level of data) and **one categorical outcome**, 2 levels only

multinominal logistic regression one or more predictors (any level of data) and **one categorical outcome**, multiple levels

what type of regression should you run for the following?: predictors: age, FIM, balance score, MMSE score, gait velocity outcome: live independently at home, assisted living, requires 24 hour care

ordinal logistic regression one or more predictors (any level of data) and **one ordinal outcome**, multiple levels

Research: Module 7 - Regression Flashcards

(46 cards)