Brainscape's Knowledge GenomeTM

Browse over 1 million classes created by top students, professors, publishers, and experts.


See full index

DClin Application > Stats Test > Flashcards

Stats Test Flashcards

Sharon's Flashcards

1
Q

What is an ANOVA

A

Analysis of Variance

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

When to use and ANOVA

A

When we are testing experiements that have 3 or more levels of independent variables (e.g., comparing a control vs caffeine in the morning vs caffeine at night)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

Why don’t we use multiple t-tests

A

Type 1 error will increase

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

What does ANOVA produce

A

F-ratio

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

What is an F-ratio

A

Compares systematic variance to unsystematic variance

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

What can / can’t an ANOVA tell us?

A

It can tell use there was an effect but it cannot tell us what the effect was

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

How do we find out what the effect was when doing ANOVA

A

Planned comparisons or post-hoc tests

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

What is the bonferroni correction

A

A wat to control type 1 error by dividing the alpha (0.05) by the number of tests
This then sets the new p-value for a test to be significant

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

What is planned comparisons

A

A set of comparisons between group means that are constructure before any data is collected
this is theory led
and there is more power to these than post hoc tests

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

What assumptions need to be met when doing ANOVA

A
  1. Normal distribution
  2. Homogentiy of variances
  3. Sphericity
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

Tests of homogeneity of variances for independent ANOVA’s

A

Levene’s test
significant Levene’s = assumption of homogeneity of variance has been violated

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

Test of Sphericity for dependent ANOVAs

A

Mauchly’s test
Significant Mauchly’s = assumption of sphericity has been violated

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

Define homogeneity of variance

A

Assumption that the variance of one variable is similar at all levels of another variable

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

Define Sphericity

A

The difference taken from the same participatn / entity are similar

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

What is a one-way ANOVA

A

One independent variable will be manipulated

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

What is one-way independent ANOVA

A

Experiments with 3+ levels of the independent variable and different participants in each group

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
17
Q

How to run a one-way ANOVA on SPSS

A
  1. Check Levene’s test - if significant then assumption of homogeneity of variances has been violated
  2. Between-group effects = SSm (variation due to the model aka experimental effect). To find the total experiment effect look at between-group sum of squares
  3. Within-group effects = SSr (unsystematic variation)
  4. To be able to compare between groups and within groups we look at the mean squares.
  5. Look at the F-ratio, if significant do post-hoc tests
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
18
Q

What post-hoc tests do you run after a significant ANOVA (you want to control for type 1 error)

A

Bonferroni correction

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
19
Q

What post-hoc tests to run after a significant ANOVA (you have very different sample sizes)?

A

Hochberg’s GT2

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
20
Q

What post-hoc tests to run after a significant ANOVA (you have slightly different sample sizes)

A

Gabriel’s procedure

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
21
Q

What post-hoc tests to run after a significant ANOVA (you have doubts about variance)

A

Games-Howell procedure (this one is a safe bet)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
22
Q

What is effect size

A

the magnitude of an effect
r

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
23
Q

How to calculate effect size

A

R squared = SSm / SSt
Square root this to get effect size (r)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
24
Q

What is SSt

A

Total sum of squares
total amount of variation within our data

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
25
Q

What is SSm

A

Model sum of squares
variation explained by our model

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
26
Q

What is SSr

A

Residual sum of squares
variation not accounted for in our model

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
27
Q

What is a two-way ANOVA

A

Two independent variables will be manipulated

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
28
Q

How to run a two-way ANOVA on SPSS

A
  1. Check Levene’s tests - if significant then assumption of homogeneity of variances has been violated. If violated then transform your data or use a non-parametric test or report inaccurate F value
  2. Summary table will include an effect for each independent variable (aka main effects) and the combined effect of the independent variables (aka interaction effects)
  3. Bold items are the SSm, Error = SSr
  4. Look at the F-ration, if significant then complete post hoc tests
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
29
Q

What is a repeated measures ANOVA

A

Three or more experimental groups with the same participants

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
30
Q

How to run a repeated measures ANOVA on SPSS

A
  1. Check sphericity (equal variances between treatment
    levels). If Mauchly’s test is significant then the assumption of
    sphericity has been violated.
    - If sphericity has been violated, we can look at either the
    Greenhouse-Geisser Estimate, the Huynh-Feldt estimate or
    the lowest possible estimate of sphericity (aka lower bound).
    - Use Greenhouse when Mauchly’s is LESS than 0.75, use
    Huynh when Mauchly’s is MORE than 0.75
  2. If the effect is significant, we need to look at ‘pairwise
    comparisons’ to see where the effect lies.
    - Look for significant values ie. less than 0.05
  3. Calculate effect size - use benchmarks of .10 / .30 / .50
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
31
Q

When to use greenhouse-geyser and when to use Mauchlys

A
  • Use Greenhouse when Mauchly’s is LESS than 0.75, use
    Huynh when Mauchly’s is MORE than 0.75
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
32
Q

What happens when we violate sphericity

A

violating sphericity = less power = increases type 2 error

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
33
Q

What is a mixed ANOVA

A

Independent variables are measured using both independent and repeated measures groups

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
34
Q

How to run mixed ANOVA on SPSS

A
  1. As mixed ANOVA uses both independent and repeated design
    we need to check if assumption of homogeneity of variances AND
    sphericity have been violated.
  2. Look at both output tables and find the main effects (one for
    each INDEPENDENT VARIABLE) and one interaction term. (words
    in CAPITALS are your INDEPENDENT VARIABLEs you need to look
    at these)
  3. Look at the F-ratios in both tables.
  4. If the effect is significant then we can run t-tests to see where
    the effect lies, make sure to use Bonferroni method (independent
    variable alpha 0.05 by the number of tests you will run)
    - Look at both ‘paired samples test’ tables. → this is known as a
    SIMPLE EFFECTS ANALYSIS.
  5. Calculate effect size - use benchmarks of .10 / .30 / .50
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
35
Q

What is ANCOVA

A

sometimes we conduct research we know some factors have influence on our DVs (from previous research e.g., age and memory)
These factors are called covariates and we can include them in our ANOVA

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
36
Q

Why do we use ANCOVA

A

to reduce the error variance (increase how much variance we can explain)
eliminate confounds (by including the covariates we remove the bias of these variables)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
37
Q

How to run ANCOVA on SPSS

A
  1. Check Levene’s test of homogeneity of variances.
    - If significant, transform the data of complete a non-
    parametric test.
  2. The output will look the same, it will just include the
    covariates.
  3. Look at the F-ratio for all the main effects and for
    the covariates.
    - If the covariate is significant, this means that it has
    a relationship with our main independent variable.
  4. Calculate effect size - use benchmarks of .10 / .30
    / .50
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
38
Q

What is MANOVA

A

Multivariate analysis of variance
ANOVA but when there are several dependent variables

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
39
Q

How to run MANOVA on SPSS

A
  1. Check for independence, random sampling, multindependent
    variableariate normality and homogeneity of covariances matrices.
    - If Box’s test is significant then the assumption of homogeneity of
    covariances matrices has been violated.
  2. Look at the multindependent variableariate test ‘group’ table. This is
    showing the effect of the Independent variable on the DV.
  3. When looking at the output Pillai-Bartlett test (Pillai’s trace) statistic
    is the most robust.
  4. If there is a significant F ratio then we need to look at the
    unindependent variableariate tests or run a discriminant analysis.
    How to interpret unindependent variableariate test statistics? -
    Levene’s should be non-significant
    - then look at ‘tests of between-subjects effects’ → corrected model
    and group row stats should be significant if there is an effect between
    IVs and DVs.
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
40
Q

How to interpret discriminant analysis

A

Look at the ‘covariance matrices’ to see the direction and
strength of the relationships
Eigenvaues percentage of variance = variance accounted
for, square the canonical correlation to use as an effect size.
Wilks’ Lambda table shows significance for all variables,
look for the significant ones.
Use the Standardised Canonical Discriminant Function
Coefficients table to see how the DVs have contributed.
Scores can range between -1 - 1, high scores = variable is
important for the variate. Look down the ‘function 1’ column,
if one value is positindependent variable and the other is
negatindependent variable then the variate (aka function)
has discriminated the two groups

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
41
Q

What is power analysis

A

The ability for a test to find an effect is known as statistical power

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
42
Q

What is power of a test

A

Power of a test = the probability that a test will find an effect if there is one
We aim to achieve a power of 0.8

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
43
Q

Power of a statistical test depends on

A
  1. how big the effect is
  2. how strict we are with our alpha level (i.e., 0.05 or 0.01)
  3. How big the sample size is - the bigger the sample size, the stronger the power
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
44
Q

What are confidence intervals

A

A range of values that are believed to contain the true population value
eg. a 95% confidence interval means that if
we were to take 100 different samples and
compute a 95% confidence interval for each
sample, then approximately 95 of the 100
confidence intervals will contain the true
mean value

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
45
Q

How to interpret confidence intervals

A
  • If 95% CI do not overlap = means come from
    different populations.
  • CIs that have a gap between the upper and
    lower end of another - p <0.01
  • CIs that touch end to end - p = 0.01
  • CIs that overlap moderately - p = 0.05
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
46
Q

What are common effect sizes

A

Cohen’s D
Pearson’s correlation coefficient r
odds ratio

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
47
Q

What is Cohen’s d

A

The difference between two means divided by the SD of the mean of the control group, or a pooled estimate based on the SD of both groups

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
48
Q

What are the benchmarks for Cohen’s D

A

small d = 0.2
medium d = 0.5
large d = 0.8

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
49
Q

What are the benchmarks for Pearson’s correlation coefficient

A

small r = 0.1
medium r = 0.3
large r = 0.5
0 = no effect, 1 = perfect effect

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
50
Q

What does an odds ratio of 1 mean

A

the odds of an outcome are equal in both groups

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
51
Q

How to calculate the odds ratio

A

calculate by dividing the probability of the event happening by the probability of it not occurring

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
52
Q

What is categorical data

A

Data which can be divided into groups (e.g., gender, age group)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
53
Q

How to analyse categorical data

A

Pearson’s chi squared test
The likelihood ratio
Yates continuity correction
Log linear analysis

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
54
Q

when to use Pearson’s chi squared test

A

when we want to see if there is a relaitonship between two categorical variables
if the expected frequency is less than 5 then we need to use Fisher’s exact test

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
55
Q

When to use th likelihood ratio

A

to be used instead of chi squared test when samples are small

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
56
Q

When to use Yate’s continuity correction

A

When we have a 2x2 contingency table then type 1 error increases
Yate’s continuity correction fixes this by lowering the chi squared statistic

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
57
Q

What is a 2x2 contingency table

A

2 variables with two level e.g., males vs female / phone vs no phone

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
58
Q

When to use log linear analysis

A

When there are 3+ categorical vairbales

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
59
Q

What are the assumptions when analysing categorical data

A

independence of residuals (as such you cannot use chi squared on repeated measures)
expected values: should not be less than 5

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
60
Q

When to use chi-squared test

A

use a chi-squared test if you have nominal (categorical) data
the chi squared test can be used to see if these observed frequencies differ from those that would be expected by chance

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
61
Q

Types of chi squared test?

A

Chi squared goodness of fit test (one IV)
Chi squared as a test of association (Two IVs)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
62
Q

When to use chi-squared good ness of fit

A

Used to compare an observed frequency
distribution to an expected frequency
distribution.
- Eg. when picking fruit are people more
likely to pick an apple vs a banana.
- If significant then, some fruit get picked
more than we would expect by chance.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
63
Q

When to use Chi squared as
a ‘test of association’ (two
independent variables)?

A

Used to see if there is an association
between two independent variables.
- Eg. is there an association between
gender and choice of fruit.
- If significant then, there is an association
between the two variables.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
64
Q

What is additivity and linearity

A

the outcome variable is linearly related to predictors

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
65
Q

What are the parametric test assumptions

A

At least interval data
Additivity and linearity
Normally distributed
Homoscedasticity/homogeneity of variance
Independence

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
66
Q

What is homoscedasticity / homogeneity of variance

A

Variance of the outcome variable
should be stable at all levels of the
predictor variable.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
67
Q

What is independence

A

errors in the model should be dependent

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
68
Q

How to spot issues with assumption of normality

A
  • look at the histogram (it should look like a
    bell curve)
    -look at the p-p plot (dots should fall
    on/near the line)
  • Look at descriptive statistics (skewness
    and kurtosis should be near to 0)
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
69
Q

How to spot issues with
assumption of
linearity/homoscedasticity/
homogeneity of variances?

A

Look at scatter plots
Look at Levene’s test - significant =
variances unequal = assumption of
homogeneity of variances has been
broken.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
70
Q

What does a scatterplot look like when data is normal

A

dots scattered evenly everywhere

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
71
Q

What does a scatter
plot look like when data
= heteroscedasticity?

A

funnel shape

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
72
Q

what does a scatter plot look like when data is non-linear

A

curve

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
73
Q

What does a scatter plot look like when data is non-linear and heteroscedasticity

A

curve and funnel (e.g., a boomerang)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
74
Q

Non-parametric alternatives to ANOVAs

A

kruskal-wallis
Friedman’s ANOVA

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
75
Q

Non parametric alternative to one-way independent ANOVA

A

Kruskal-Wallis

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
76
Q

Non parametric alternative to repeated measures ANOVA

A

Friedman’s ANOVA

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
77
Q

How to interpret the
Kruskal-Wallis test?

A
  1. Look at the ‘ranks’ table, the mean ranks tell us which
    condition had the highest ranks
  2. If the chi squared test is significant then there is a difference
    between groups (but we do not know what kind of difference)
  3. To see where the difference lies, look at the box-plot and
    compare the experimental group to the control group.
  4. OR we can do a Mann-whitney test and use Bonferroni
    correction (divide alpha by the number of tests), look to see
    which conditions are significant.
  5. Calculate the effect size by dividing the z score by the
    number of obvs square rooted.
    - use benchmarks of .10 / .30 / .50
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
78
Q

What is an alternative to
the one way repeated
measures ANOVA?

A

Friedman’s ANOVA

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
79
Q

How to interpret the
Friedman’s ANOVA?

A
  1. Look at the ‘ranks’ table, the mean ranks tell us which condition
    had the highest ranks.
  2. If the chi squared test is significant then there is a difference
    between groups (but we do not know what kind of difference)
  3. To see where the difference lies, look at the box-plot and
    compare the experimental group to the control group.
  4. OR we can do a Wilcoxen test and use Bonferroni correction
    (divide alpha by the number of tests), look to see which conditions
    are significant in the ‘test statistics’ box.
  5. Calculate the effect size by dividing the z score by the number of
    obvs square rooted.
    - use benchmarks of .10 / .30 / .50
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
80
Q

What are
correlations?

A

relationships between variables

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
81
Q

define covariate

A

If variables are related, then change in one variable will lead to similar change in another variable

82
Q

What is cross-product deviation

A

the similarities / differences of the deviation

83
Q

How to calculate cross product deviation

A

multiple the deviations of one variable but the deviations of the other variable

84
Q

How to calculate
covariance?

A

Calculate the cross product
deviation and divide by the number
of observations - 1.

85
Q

If covariance is positive, then the correlation will be…..

A

positive

86
Q

what is the standardised version of the covariance

A

correlation coefficient

87
Q

what is the correlation coefficient

A

the standardised version of the covariance
Pearsons correlation coefficient measures the strength of relationship between variables

88
Q

how to calculate a correlation coefficient

A

Divide covariance by standard
deviation.
Scores lie between -1 and +1
+1 perfect positive relationship

89
Q

What is bivariate correlation

A

correlation between 2 variables

90
Q

what is a partial correlation

A

quantifies relationship between two variables while controlling the effect of other variables

91
Q

How to run bivariate correlations on SPSS

A
  1. Have assumptions been violated? If they
    have use Kendall’s tau / Spearman’s rho
  2. Look at ‘correlations’ table and see if
    Pearson’s correlation are significant.
  3. Look at the confidence intervals (if the
    data is not normal, look at the bootstrap
    CI)
    - If the confidence interval crosses zero it
    suggests there could be NO effect.
92
Q

What is the coefficient of determination

A

The coefficient of determination is
represented by the term r2 (or R2) it is
the percentage of the total amount of
change in the dependent variable (y)
that can be explained by changes in the
iv (x).

93
Q

How to calculate R^2

A

Square the correlation coefficient (R)

94
Q

What correlation coefficient test should you use if the data is non-parametric

A

spearman’s rho

95
Q

How to interpret spearman’s rho on SPSS

A
  1. Look at ‘correlations’ table and see if
    correlation coefficient is significant.
  2. Look at the confidence intervals (if the
    data is not normal, look at the bootstrap CI)
  3. If the confidence interval crosses zero it
    suggests there could be NO effect.
96
Q

What correlation coefficient test should you use if the data is non-parametric and sample is small

A

Kendall’s tau

97
Q

how to interpret Kendalls tau on spss

A
  1. Look at ‘correlations’ table and see if
    correlation coefficient is significant.
  2. Look at the confidence intervals (if the
    data is not normal, look at the bootstrap
    CI)
  3. If the confidence interval crosses zero
    it suggests there could be NO effect.
98
Q

What type of correlation is used when one of the two variables is dichotomous

A

point-biserial correlation

99
Q

when is point biserial correlation used

A

when one variable is discrete dichotomy (e.g., pregnancy)

100
Q

when is biserial correlation used

A

used when one variable is a continuous dichotomy (e.g., passing or failing an exam)

101
Q

what is a semi-partial correlation

A

we control for the effect that the third variable has on only one of the variables in the correlation

102
Q

what is the equation of the simple linear model

A

Outcome = (b0 + b1X) + error.

103
Q

what is the simple linear model used for

A

we can predict an outcome for a person using the model (the bit in brackets) and some error associated with this model

104
Q

in the linear model what is b0

A

intercept

105
Q

in the linear model what is b1

A

slope / gradient

106
Q

what are b0 and b1 in the simple linear model

A

parameters
regression coefficients

107
Q

in the linear model, what does a positive b1 mean

A

positive relatiomship

108
Q

what are the assumptions of the linear model

A

Normally distributed errors
Independent errors
Additivity and linearity
Homoscedasticity

109
Q

What is additivity and linearity

A

outcome variable and predictors combined effect is best described by addition effects together

110
Q

how can we check independent error

A

durbin-watson value should be between 1 and 3

111
Q

how big should our sample be when using the linear model

A

10 or 15 cases of data per predictor

112
Q

What is cross validation

A

assessing the accuracy of a model over different samples
look at the adjusted R squared

113
Q

how to inter-reset simple linear regression in SPSS

A
  1. Look at the ‘model summary’ R represents
    correlation
    - R squared represents the amount of variance
    accounted for by the model.
  2. Look at the ‘ANOVA’ table, if the F ratio is
    significant then our model is a better predictor in
    comparison to using the mean.
    3.
    ‘B’ in ‘coefficient’ table tells us the gradient and
    the strength of the relationship between a predictor
    and outcome variable. Significant means the
    predictor significantly predicts the outcome variable.
114
Q

what does R squared represent

A

the amount of variance accounted for by the model

115
Q

what is multiple regression

A

a model with several predictors

116
Q

what are the different methods of regression

A

hierarchical regression
forced entry
stepwise methods

117
Q

what is hierarchical regression

A

predictors are based on past work and the researcher decides which order to enter the variables
known predictors should go first, followed by any new ones that we suspect will be important

118
Q

what is forced entry regression

A

all predictors are forced into the model at the same time
we have to have good theoretical support to include the predictors we have included

119
Q

what is stepwise methods regression

A

Generally frowned upon because the
researcher is not in control.
The decision is based on mathematical criterion
that SPSS decides.
It will see how much variance is a accounted
for by one predictor, if it is sufficient then it will
keep it and move on to find another predictor
which may explain more variance.

120
Q

what are the concerns when including more than one predictor in a model

A

multicollinearity

121
Q

what is multicollinearity

A

exists when there is a strong correlation between 2+ of our predictor variables

122
Q

why is having 2+ variables with perfect collinearity problematic

A

the values of b for each variable are interchangeable

123
Q

what happens when collinearity increases

A

standard errors of the b coefficient increase
the size of t is limited
it is difficult to assess the individual importance of predictor when they are highly correlated

124
Q

what is r

A

the correlation between predicted values of the outcome and the observed values

125
Q

how can we check to see if multicollinearity is a problem

A

check is the variable inflation factor (VIF) and tolerance statistics

126
Q

how to interpret VIF

A

VIF greater than 10 = cause for concern
VIF greater than 1 = regression may be biased

127
Q

how to interpret tolerance statistic

A

tolerance below 0.1 = serious problem
tolerance below 0.2 = potential problem

128
Q

what’s cooks distance

A

quantifies the impact of an outlier on a model
if cooks distance is above 1, then that case may be influencing the model

129
Q

what is factor analysis

A

a statistical procedure that identifies clusters of related items on a test
do several facets reflects one variable (e.g., burnout (variable) - stress levels, motivation (facets))

130
Q

what do factors represent

A

clusters of variables that correlate highly with each other

131
Q

what cane we use to decide which factors to extract

A

scree plot
where the inflexion is where you should cut off

132
Q

what is rotation used for in factor analysis

A

to discriminate factors

133
Q

what is ethics comprised of

A

informed consent
deception
debriefing
confidentiality
protection from physical and psychological harm

134
Q

what is informed consent

A

participants should understand what the experiment involves and understand their rights
the ability to withdraw at any point

135
Q

when is it ok to not gain informed consent

A

in observational studies only if the person being observed is in a situation where they would be in public view anyway (e.g., shopping centre)

136
Q

what are the levels of measurement

A

Nominal
ordinal
interval
ration

137
Q

which levels of measurement use non-parametric tests

A

nominal
ordinal

138
Q

which levels of measurements use parametric tests

A

interval
ratio

139
Q

what is nominal data

A

the numbers act as a name
data from a nominal scale should not be used for arithmetic
nominal data can be used for frequencies

140
Q

what is ordinal data

A

tell us the frequencies and in what order they occured
does not tell use the differences between values
most self report questionnaires are ordinal data

141
Q

what is interval data

A

differences between values on a scale are equal
tested with parametric statistics

142
Q

what is ratio data

A

differences between values on a scale are equal
distances along the scale are divisible
there is a true zero point (i.e., no minus numbers, e.g., reaction time)

143
Q

types of variables

A

discrete
continuous

144
Q

what are discrete variables

A

non-overlapping categories
eg being pregnant - you either are or are not

145
Q

what are continuous variables

A

runs along a continuum
e.g, agression

146
Q

what is validity

A

whether an instrument measures what it sets out to measure

147
Q

what is criterion validity

A

whether you can establish if a measurement is measuring what it is meant to through comparison to an objective criteria
we assess this by relating scores on your measure to real-world observation

148
Q

what is concurrent validity

A

evidence that scores from an instrument correspond to external measures
eg. nurses are assessed for knowledge
via a written & practical test. If they
score well on the test and then well on the
practical = concurrent validity.

149
Q

what is predictive validity

A

when data from the new instrument are used to predict observations later in time

150
Q

what is content validity

A

with questionnaires, we can assess how well individual items represent the construct being measured

151
Q

what is factorial validity

A

when making questionnaire and using factor analysis
if your factors are made up of items that seem to go together meaningfully = factorial validity

152
Q

what is reliability

A

whether an instrument can be interpreted consistently across different situations

153
Q

what is test-retest reliability

A

the ability of a measure to produce consistent results when the same entities are tested at two different points in time

154
Q

how can we test reliability

A

split-half method
cronbach’s alpha

155
Q

what is split-half method

A

Splitting a test into two and having the
same participant do both.
The results are then correlated, and if
they are similar then there is high
internal reliability.

156
Q

how can we infer reliability through Cronbach’s alpha

A

if the correlation is above 0.8 = reliable

157
Q

what is measurement error

A

The difference between the score we
get using our measurement and the level
of the construct we are measuring.
Eg. I actually weigh 47kg but the scales
show 57kg

158
Q

what is a histogram

A

Used for frequency distribution.
Plots a single variable (x-axis)
against the frequency scores (y-axis)

159
Q

what is a box plot

A

Used to show important characteristics of a set of
observations.
Center of the plot = median
Box = middle 50% of observations (aka interquartile range)
Upper and lower quartile are the ends of the box.
Whiskers = top and bottom 25% of scores.

160
Q

what is a bar chart used for

A

graphing means

161
Q

what are scatterplots

A

used for graphing relationships
a graph that plots each persons score on one variable against another

162
Q

what is null hypothesis significant testing

A

A method of assessing scientific theories
We have 2 competing hypothesis - null
hypothesis (no effect) and the alternative
hypothesis (there is an effect).
We compute a test statistic and find out the
likely it is that we would get a value as big as
the one we have if the null hypothesis is true
(ie. by chance).

163
Q

in null hypothesis significance testing, what is a significant effect

A

less than <0.005 = significant effect

164
Q

what is type 1 error

A

saying there is an effect when there isn’t
rejecting the null hypothesis when it is true

165
Q

what is a type 2 error

A

saying there isn’t an effect when there is
accepting the null hypothesis when it is false

166
Q

what is the power of statistical test

A

probability we will find an effect if it exists

167
Q

what is a meta-analysis

A

effect sizes from different studies testing the same hypothesis are combined to get a better estimate of the size of effect in the population

168
Q

what is an alternative to null hypothesis significance testing

A

bayesian analysis

169
Q

what is an IV

A

The variable that is being manipulated
by the researcher.
It is independent from the other
variables.
IV’s can have different levels.
IV goes on the x axis.

170
Q

what is the DV

A

The variable that is hypothesised to
be affected by the IV.
It depends on the IV.
DV goes on the y axis.

171
Q

what is between group design

A

participants placed into different groups
they can be part of one group for the entire experiment

172
Q

what is a within groups design

A

same participants placed into all levels of the independent variable

173
Q

what is quantitative research

A

research that deals with numerical data
data is analysed to compare groups or make inferences
confirm / test hypotheses using numbers

174
Q

what is qualitative research

A

mainly uses words
data analysed to summarise, categories and interpret themes
explorative, an attempt to understand through words

175
Q

what is descriptive research

A

aims to describe a phenomenon
what, when, where and how
does not lead us to think about causation

176
Q

what is correlational research

A

aims to define a statistical relationship between variables
e.g., is there a relationship between cognition and caffeine

177
Q

what is quasi-experimental design

A

experimenter has no control over the allocation of pots to conditions or the timing of experimental conditions

178
Q

what is experimental design

A

aims to establish causality
randomisation in important to reduce the effect of confounding variables

179
Q

what is ABA design

A

baseline behaviour measured (A)
treatment applied and behaviour measured while treatment present (B),
treatment is removed and the baseline behaviour is recorded again (A)

180
Q

types of sampling in qualitative research

A

purposive sampling
theoretical sampling

181
Q

what is purposive sampling

A

selecting participants according to criteria is important for the research question

182
Q

what is theoretical sampling

A

the people you attempt to recruit will change as a result of the things you are learning

183
Q

what is meant by 2x2 design

A

a research design with 2 independent levels, each with 2 levels

184
Q

how can we examine mediated or moderated relationships

A

path analysis

185
Q

what is factor loading

A

a correlation coefficient between a variables and a factor (cluster of variables)

186
Q

what is a mediating variables

A

explained relationships between independent variables and dependent variables

187
Q

what is a moderating variables

A

alters the relationship between the IV and DV

188
Q

what is the standard error

A

the standard deviation (spread) of the sampling distribution

189
Q

what is standard deviation

A

a measure of how much scores vary around the mean score

190
Q

what is variance

A

how the values are dispersed around the mean

191
Q

what are z-scores

A

Number of units of standard deviation
any one value is above or below the
mean
The larger the z-score the further its
value is away from the group’s mean

192
Q

how to calculate z-scores

A

(raw score - mean) /
standard deviation

193
Q

what does a significant F-ratio tell you

A

the model is a better predictor in comparison to the mean

194
Q

what is the p-value

A

the probability of obtaining results at least as extreme as the observed results of a statistical hypothesis test, assuming that the null hypothesis is correct

195
Q

what are the degrees of freedom for one sample t-test

A

sample size - 1

196
Q

what are the degrees of freedom for one way ANOVA

A

sample size - k

where k is the number of cell means

197
Q

what is the ceiling effect

A

when scores tend to cluster at the upper end of a distribution

198
Q

what is the floor effect

A

when a task is so difficult that all scores are very low

199
Q

what is a pairwise comparison

A

post hoc compares two individual means at a time

200
Q

what is a main effect

A

effect of one IV while ignoring the other IV

201
Q

what is an interaction effect

A

the combined effect of two or more IV’s

DClin Application (3 decks)

Brainscape helps you reach your goals faster, through stronger study habits.
© 2024 Bold Learning Solutions. Terms and Conditions