1.8 Preparing for analysis

Question 1

Before conducting a statistical analysis you need to check your data for eight things:

Answer 1

1. Accuracy of data entry,
2. Missing data,
3. Outliers,
4. Normality,
5. Linearity, homoscedasticity, and homogeneity of variance,
6. Independence,
7. Multicollinearity and singularity (MANOVA and multiple regression).
8. Other assumptions

Question 2

Missing data may be addressed through a range of approaches such as

Answer 2

list-wise deletion, mean substitution, expectation-maximization, multiple imputation

Question 3

As defined by Tabachnick and Fidell (2019, p. 63), an outlier is

Answer 3

“a case with such an extreme value on one variable (a univariate outlier) or such a strange combination of scores on two or more variables (multivariate outlier) that it distorts statistics”

Question 4

If not identified and processed, outliers can lead to

Answer 4

Both Type I and Type II errors.

Question 5

There are several ways that outliers can be addressed that include

Answer 5

• ignoring (non-influential) data points (univariate, multivariate),
• deleting individual data points, if sample size can accommodate for this (univariate, multivariate),
• running the analysis with and without the outlier/s to justify keeping the outlier/s (univariate, multivariate),
• modification to reduce the bias of the data through winsorizing or trimming data (univariate), and
• transforming data for large data sets (univariate, can be extremely complex for multivariate).

Question 6

Occasionally, new multivariate outliers may have been identified following deletions or original outliers. This happens because once you remove a single outlier, the data set becomes more consistent and new data points will become

Answer 6

extreme points

Question 7

Distributional information, such as skewness and kurtosis values, can provide indicators of

Answer 7

symmetry and peakedness of a variable’s distribution

Question 8

Skewness relates to the

Answer 8

symmetry of the distribution

Question 9

Positive skew is depicted when most scores are clustered at the

Answer 9

lower end of the distribution,

Question 10

Kurtosis refers to the

Answer 10

peakedness of the distribution

Question 11

A positive skew is described as ________ and a negative skew is described as:

Answer 11

leptokurtic; platykurtic

Question 12

Screening the residuals for normality is common practice when conducting data analyses for

Answer 12

ungrouped data

Question 13

Linearity (straight-line relationships between variables) can be observed graphically through

Answer 13

bivariate scatterplots

Question 14

For ungrouped data, the assumption of homoscedasticity refers to

Answer 14

assumption in regression analysis that the residuals on the continuum of scores for the predictor variable are fairly consistent and as such have similar variances

Question 15

For grouped data, Homogeneity of variance is

Answer 15

the assumption that the variance of one variable is stable (i.e. relatively similar) at all levels of another variable

Question 16

There are often two types of assumptions of independence often referred to in statistics which are

Answer 16

Independence of Observations and Independence of Residuals/Errors

Question 17

Independence of Observations requires each participant to

Answer 17

participate only once in the research and as such only contribute one set of data

Question 18

the assumption of Independence of Residuals/Errors is the assumption that

Answer 18

errors in your model are not related to each other

Question 19

the Durbin-Watson test statistic is used to

Answer 19

assess for serial correlations (autocorrelation) of errors

Question 20

Multicollinearity and singularity are

Answer 20

problems with a correlation matrix that occurs when variables are too highly correlated. With multicollinearity, the variables are very highly correlated (say, above .80); with singularity, the variables are redundant; one of the variables is a combination of two or more of the other variables

Question 21

Investigation of Tolerance and Variance Inflation Factors can help determine

Answer 21

whether multicollinearity is a problem within your sample

Question 22

the assumption of sphericity relates to

Answer 22

repeated measures ANOVA and mixed model ANOVA designs

Question 23

Sphericity assumes that

Answer 23

variances of the differences between data taken from the same participant are equal

Question 24

Field (2018, p. 283) suggests that nonparametric statistics based on ranks are not affected by

Answer 24

small sample sizes, extreme scores, and outliers, and they do not require a normally distributed sample

Question 25

Allen, Bennett & Heritage (2018) suggest that non-parametric tests should be used with

Answer 25

ordinal data, and/or where the sample is not normally distributed