Home / Summaries / ARMS / values-coefficients-assumption

Grasple lessen

Q: What can we check with the standardized residuals?

W hether there are outliers in the Y-space.

Q: What can be echecked with Cook's Distance?

W hether there are outliers within the XY-space.

56 important questions on Grasple lessen

What can we check with the standardized residuals?

Whether there are outliers in the Y-space.

What is the rule of thumb for standardized residuals?

Values must be between -3.3 and +3.3. Those smaller than -3.3, or greater than +3.3 indicate outliers.

What is the rule of thumb for an outlier in X-space?

Values of Mahalanobis distance must be lower than 10 + 2 • (number of Independent Variables). Values higher than this critical value indicate outliers.

What can be echecked with Cook's Distance?

Whether there are outliers within the XY-space.

What are the 3 consquences of multicollinearity?

- The regression coefficients (B) are unreliable
- It limits the magnitude of R (the correlation between Y and Ŷ)
- The importance of individual independent variables can hardly be determined, if at all.

How can you determine whether multicollinearity is an issue?

In the last two columns of the coefficients table. You can use the following rule of thumb:
- Values for the Tolerance smaller than .2 indicate a potential problem.
- Values for the Tolerance smaller than .1 indicate a problem.
The variance inflation factor (VIF) is equal to 1/Tolerance. So for the VIF, values greater than 10 indicate a problem.

What is the adjusted R square?

An estimate of the proportion of explained variance in the population. It adjusts the value of R2 on the basis of the sample size (n) and the number of predictors in the model (k).

What assumptions should you check before performing a multiple regression?

· Measurement levels
· Linearity
· Absence of outliers

What can be the measurement levels of variables that are involved in moderation?

- An interaction between two continuous variables
- An interaction between a categorical and a continuous variable
- An interaction between two categorical variables.

What is the required measurement level of a covariate?

Interval/ratio. Since the covariate correction is based on the linear relation between the covariate and y (within each group), both the covariate and y need to be of interval/ratio level (i.e. numerical/quantitative).

The ANCOVA test is an omnibus test. What does that mean?

Multiple comparisons are made with just one statistical test/p-value, and thus with only once a Type I error probability that will not exceed the intended alpha level

What is the rule of thumb for Cook's distance?

Values for Cook’s distance must be lower than 1. Values higher than 1 indicate influential respondents (influential cases).

What table contains information on multicollinearity?

The Coefficients table

What is the gain and price with an omnibus test?

Gain: Better control over your Type I error.
Price: If the null hypothesis is rejected, you only know that there are differences between some means, but you have not yet learned between which.

In multiple testing, the probability of making one or more Type 1 errors (that is, finding a significant test result while there is no effect in the population) can be computed with a simple equation. What is that equation?

1−(1−α) ^k, where k denotes the number of tests you do.

Why would we center continuous predictors in a moderation analysis?

Centering renders meaningful regression coefficients and it avoids multicollinearity.

On what principle is the Bonferroni correction based?

When you want to be certain that the (overall) Type I error (FWER) can not exceed 5% when you do k tests, then use for each single test α/k.

What do you know about the Bonferroni corrected p-value?

It's bigger than the formulated p-value. A correction makes the reported p-value bigger ('less easy significant' to protect against inflated type I error).

What is tested with Box's M-test?

The homogeneity assumption of MANOVA: the variance-covariance matrices should be approximately equal between the groups (homogeneity of covariances).

What are two reasons to run a MANOVA on a set of DV's instead of several ANOVAs (one per DV)?

1. To find truly multivariate differences between groups that are not picked up if you only examine one DV at a time (i.e., univariate).
2. Protection against inflated type I error.

What doe construct validity mean?

The extent to which the instrument (operationalization) succeeds in measuring the construct and thus fits the conceptual definition.

What does internal validity mean?

The extent to which the research method can eliminate alternative explanations for an effect/relationship.

If the assumption of sphericity is not met, there are several options. What are the options?

- Greenhouse-Geisser correction
- Huynh-Feldt correction
- MANOVA on the difference scores

What kind of variables are dummy variables?

Dichotomous variables that only take the values 0 and 1.

What are 3 requiremens for demonstrating a causal relationship?

- Two or more variables should be related
- The cause should precede the outcome over time
- No other explanations may be possible

'Unchangeable' characteristics (like age and gender) can only be ..., not ...

Moderators, mediators

The ... can be used to determine the most important predictor of satisfaction. The independent variable with the largest absolute beta is the most important predictor, where absolute means that the sign (+ or -) does not matter.

standardized coefficients, the Beta's

Missing data influences the ... of a study when the missing data pattern is systematic (not random), causing the research sample to differ from the population.

External validity

Consider a researcher performing an ANCOVA on a factor with 5 levels (i.e., comparing 5 groups). His interest is in evaluating if group 1 differs from group 2, 2 from 3, 3 from 4, and 4 from 5. What type of comparisons fits this question best?

Planned comparisons. He will not make all comparisons to find out where the differences are but instead formulated which groups to compare beforehand.

What information provide polynomial contrasts?

Information on the trend over time.

The ... presented provide homogeneity tests per DV.

Levene's tests

What is simple slope analysis?

Follow-up tests after significant interaction in multiple regression: is there an effect of age for children with high IQ?

Suppose there are 4 groups. An alpha value of .05 is used. The experimenter sees that the means of the groups equal 2, 4, 9 and 7, then the same test will have a type 1 error rate of?
a. <.05
b. .05
c. >.05

C. You picked the hypothesis based on the results. And did not correct for actually looking at all the possible orderings. You have an inflated alpha level.

In what 2 situations is something HARKing?

1. If you have a hypothesis and you change it because of the results.
2. If you did not have a hypothesis, but created one after the results

A ... factor is a factor separating measurements within the same person

within-subjects

Does a non-significant interaction mean that the assumption of homogeneity of regression coefficients is met or not?

Assumption is met. If the interaction term is not significant, it means that the lines are not significantly different in steepness (slope of regression line). We conclude that the relationship between the covariate and the dependent variable is not different between the groups. In that case the condition of homogeneity of regression coefficients is therefore met.

An analysis with only within-subjects factors is called a

Repeated Measures Analysis.

Why is HARKing a problem?

This finding could be the one that is incorrect

What is the problem with selective reporting?

Literature seems to only show that the effect is present

What are 3 advantages of Registered Reports for the scientific community?

1. Rigorous review of theory and methods
2. Eliminates publication bias and reporting bias
3. Increases the reproducibility of science

To protect against the inflated Type I error risk, one can use ...

Alpha correction

At which value should we look to assess how severe the assumption violation is?

- Epsilon – this is a measure for the severity of the assumption violation.

What are 3 advantages of Registered Reports for scientists?

1. Keeping track of what you did and why
2. Peer review when it is most helpful
3. Publication guaranteed regardless results

A researcher hypothesis that in chldren the relationship between age and linguistic proficiency is dependent on birth order. Older children are more proficient, but the relationship is stronger for firstborns than laterborns.

Which conclusion is correct if the hypothesis of the researcher is true?

a. Age is a moderator
b. Age is a mediator
c. Birth order is a moderator
d. Birth order is a mediator

C. It shows that for different birth orders, the relationship between age and linguistic proficiency is different. That makes birth order a moderator.

Below is a quotation from Kanazawa (2012) about measuring social class:

"In addition to birth order and number of siblings, both measured at 16, I control for the following variables in my multiple regression analyses below: social class at birth measured by father's occupational class: 0 = umemployed, dead, retired, or no father present; 1 = unskilled; 2 = semi-skilled; 3 = skilled; 4 = white collar; 5 = professional"

Complete the following sentence: If Kanazawa's research were to be replicated in the year 2021, due to the way social class was measured, the ... Validity would be at stake.

Construct. Because social class cannot be measured based on the father's profession at present (in 1958 the father's profession was mainly decisive for the social class). You do not measure the social economic status in this way, thus the variable is not well operationalized.

Is the following statement true or false?

In cae of a significant indirect effect, zero lies inside the bootstrapped confidence interval

False

The assumption of homogeneity of variance does apply to a mixed design, but it does not for repeated measures analysis.

True.

What is not an assumption for multiple regression analysis?

a. The measurement level of the dependent variable should be at least interval or nominal with the use of dummies
b. The relationship between the dependent and the independent variable should be linear
c. The relationships between the predictors should not be too strong
d. The variance in the dependent variable should be the same for all values of the independent variable.

A. With regression, the dependent variable should be at least interval. Dummies are only relevant in case of categorical independent variables.

c is about multicollinearity, which is an important assumption for multiple regression

b is about linearity, which is an imortant assumption for multiple regression

d is about homoscedascity, which is an important assumption for multiple regression.

Sphericity is a model assumption for repeated measures analysis

True

A linear contrast test in a repeated measures analysis tests if the means at all timepoints are significantly different

False

A ... factor is a factor separating different persons

between-subjects

... Are influenced by the scale on which the variable is measured so should not be used to evaluate the importance/relevance of an effect

B-values

What is an adjusted mean / corrected average?

The predicted scores for the different groups when they have the overall average score on the covariate.

Why is the resampling size always the same as the original sample size with bootstrapping?

Because otherwise you get inaccurate standard error, because the standard error is based on the sample size. You are then oversure

Epsilon takes on values between .. and .. The lower the value, the worse the violation.

0, 1

What are the 7 steps of pre-registration?

1. Authors submit stage 1 manuscript (Introduction, Methods & Analysis)
2. Stage 1 peer review
3. If reviews are positive, then the journal offers in-principe acceptance (IPA), regardless of study outcome (protocol archived)
4. Research
5. Authors resubmit completed stage 2 manuscript (Introduction and method (unchanged), results (new), discussion (new), data and materials deposited in a public archive)
6. Stage 2 peer review
7. Manuscript published!

The question on the page originate from the summary of the following study material:

ARMS

View summary

A unique study and practice tool
Never study anything twice again
Get the grades you hope for
100% sure, 100% understanding

Remember faster, study better. Scientifically proven.