Brainscape's Knowledge GenomeTM


Top Flashcards (Certified)
All Flashcards

SEM 2 > Research Methods > Flashcards

Research Methods Flashcards

1
Q

What are the 6 rules of thumb for constructing questionnaires

A 
Make items clear (unambiguous)
Avoid presuming questions
Use simple language - short items are best
Avoid double-barrelled Qs
Avoid double negatives
Avoid v mild or extreme Qs
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

What are the advantages / disadvantages of using a Visual Analogue Scale

A

Advantage: Very sensitive - useful for measuring before + after responses
- Difficult for ps to remember previous responses

Disadvantage: False precision - how meaningful are the small differences along the scale

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

What are the advantages / disadvantages of using a 0-100 scale

A

Disadvantage: Ps may respond in multiples of 5 / 10, reducing the number of options
- False precision: how meaningful is a difference of 2 digits

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

What are the advantages / disadvantages of using worded response categories (not helpful - very helpful)
Alternative?

A

Disadvantages: Difficult to assess where the points lie in the psychological continuum / interval between categories- points may not be equal
Alternative: Likert

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

What is Thurstone Scaling and what are advantages / disadvantages

A

Present statements, some of which show favouritism for one variable / another
Get around 50 judges to rate the items
Advtanges: Easy for ps to indicate agreement, not strength
Easy to develop alternate forms of the scale
Disadvantages: V labour intensive - need 50 judges
- Judges’ ratings may be different from potential ps’

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

What is skewness a measure of

A

Skewness is a measure of symmetry (indicates an issue with the items - too many low / high answers)

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

What is kurtosis a measure of

A

Kurtosis is a measure of flatness of distributions

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

What are z scores

A

Z scores / standard scores indicate the number of SD away from the mean a data point is
Skewness statistic / Std. Error
Kurtosis statistic / Std. Error

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

What must the z score be greater than to indicate a problem

A

If the z score is greater than 3.08, this indicates there is significant skewness / kurtosis and the item should be removed

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

What is the definition of a correlation

A

The degree to which 2 variables co-relate

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

What is the definition of correlation coefficient

A

An index of the extent to which two variables are linearly correlated (strength of linear relationship)

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

What are the assumptions of Pearson’s Correlation Coefficient and how does it measure scores

A

Pearson’s correlation uses the size of the scores to measure

  1. Variables must be measured using interval data
  2. Variables must be linearly related
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

Report Pearson’s Correlation Coefficient

A

There was a sig / non-sig pos / neg correlation between x and y r(df) r = , p = . Description of correlation

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

How does Spearman’s Rank Correlation Coefficient measure data and what are the assumptions it has

A

Spearman’s ranks the scores and correlates the ranks

  1. A monotonic, but non-linear scatterplot
  2. Ordinal data (categories)
  3. When outliers are present
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

What are monotonic relationships

A

When variables move in the same direction but not necessarily at a constant rate

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

Report Spearman’s Correlation Coefficient

A

Due to the presence of outliers / monotonic relationship / ordinal data - strength was measured using Spearman’s Rank Correlation Coefficient
Sig / non-sig pos / neg correlation r(df) r = , p = . Such that…

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

When is a Partial Correlation used and what does it measure

A

Partial Correlation is used to control for a third variable
It measures the strength + direction of linear relationship between two variables when controlling for the effect of a third variable
Takes out the amount of variance explained by the variable

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
18
Q

What should you do with participants’ data

A

Make the questionnaires anonymous
Use participant codes
Store names separately from data
Keep consent forms in a locked drawer

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

When should deception be used / how

A

Avoid deception if possible

If deception is necessary, give a full debrief of the true aims

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

What is a Regression analysis used for and how does it do this

A

Regression analysis tests how well a set of IVs are able to predict a DV

Can test the model as a whole (R₂) overall variance %

Or the the independent contribution of each IV (B) beta weights

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

What are unstandardised coefficients and standardised coefficients (beta weights) a measure of

A

Regression coefficient: amount of change occuring in DV for one unit change in IV with effect of other IVs partialled out (unique contribution)

Beta weights are a standardized measure of the slope converted into SD. For every 1 SD increase in IV, the DV will go up (beta weight)

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

what is the equation for Regression analysis and what does it mean?

A

Y’ = a + bX

a is the intercept in Y axis, when x = 0

b is the slope (amount of change in y for one unit increase in X)

X = score on the variable x

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

Report a multiple regression analysis (4 stages)

A

A multiple regression analysis was performed with .. as IV and .. as DV

Regression was sig / not sig, F(df, residual) = F score, p = .

Explained r2 % of the variance in the IV

Inspection of the beta weights revealed that IVs made a sig / not sig independent contribution to prediction of DV

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

How do you report significant beta weights

A

One star for less than 0.5 [* p < 0.5]

Two stars for less than 0.1 [** p < 0.1]

Three stars for less than 0.01 [*** p < 0.01]

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
25
Explain dichotomous variables and categorical variables
Dichotomous variables: a variable that can take one of two possible values (yes / no) Categorical variables: where variables are a set of categories (political parties) Needs dummy variables - analysed using one way ANOVA
26
What is the direct entry form of multiple regression analysis
All IVs added together as a fixed set
27
What is the hierarchical entry form of multiple regression analysis
Hierarchical entry assesses the effect of one set of predictors over + above that of another set IVs can be added individually or in groups / blocks can assess overall model or unique contribution of each IV / block of IVs
28
What is the stepwise entry form of multiple regression analysis + what are some disadvantages
Stepwise entry is used to find an efficient model with minimum number of significant predictors (any IVs will do) However, it leaves the decision making to SPSS, removing meaning and interpretation An important predictor may be discarded because of correlation with another IV
29
What is forward selection stepwise analysis
Forward selection - Enters IV with highest correlation (if significant) and calculates regression. Assesses the increase in R2 each IV would give, removing IVs making no significant contributions
30
What is backward selection stepwise analysis
Backwards selection - starts with all IVs and calculates regression. Assesses contribution each IV has on R2, deleting IV with smallest contribution until only significant IVs are left
31
What is forward + backward selection (stepwise)
Forward + backward selection (stepwise) - starts with no predictors, adds one if it meets criteria and removes predictors that don't contribute significantly at each stage
32
What is a latent variable
A latent variable is one that is not directly measured but inferred through other observed variables
33
What is reliability a measure of?
A reliable score is one in which variation in scores can be attributed to the latent variable of interest which exerts a casual influence over the items
34
What are the three measure of reliability
1. Split-Half Reliability 2. Internal Consistency 3. Test-Retest Reliability
35
What is Split-Half Reliability
Scores in first half of scale correlated wit scores in second half of scale / odd scores correlated with even scores If the scale is reliable, the two halves should correlate strongly
36
What is Internal Consistency
It is based on correlations between each scale item score and the total score The Coefficient Alpha (α) indicates how much variance in the scale scores that is attributable to the true score < . 60 = unacceptable > .90 = extremely good (maybe remove some items)
37
What are some caveats of internal consistency
``` The Coefficient Alpha (α) isn't dependent only on the magnitude of correlations, but also the number of items in the scale Coefficient Alpha (α) isn't a measure of unidimensionality ```
38
What is Test-Retest Reliability
Scores in test at one point in time correlated with scores in test at another point If scale has high test-retest reliability, this will result in a strong correlation Low T-R reliability may relfect temporal instability (exam stress 4 + 2 weeks before exam)
39
What is validity a measure of
A valid scale is one that tests the latent variable of interest
40
What are the three measures of validity
1. Content Validity 2. Construct Validity 3. Predictive Validity
41
What is content validity and how is it improved
To what extent does the content of scale items reflect the latent variable? Improved through interviews with members of target population to generate items, or asking experts to rate the extent to which items reflect latent variable
42
What is construct validity and how can this be improved
Construct validity is how well a scale relates to other constructs with similar theoretical explanations Improve by administering the scale to participant sample along with a related scale - a strong correlation would indicate construct validity
43
What is predictive validity and how can this be tested
Predictive validity - does the scale have an association with other external criteria? Correlate the scores with external criterion / measures
44
What are the 7 guidelines for scale development (DeVellis, 1991)
1. Define latent variable of interest 2. Generate item pool 3. Review items for content 4. Administer measure to a development scale 5. Evaluate the items (reverse coding, mean / SD, skewness + kurtosis) 6. Compute coefficient alpha (validity in scale) 7. Validate the scale
45
What is the main aim of factor analysis
The aim of factor analysis is to analyse patterns of correlations between variables (items) in order to reduce these variables to a smaller set of "factors"/"components"
46
Explain how to interpret factor loadings
Each loading is the correlation between a variable and a factor (loading² = proportion of variance in variable accounted for by factor) Loadings > .30 are 'salient' and interpreted, while those under .30 are dismissed The aim is to have items high in one component and low in another
47
What are Eigenvalues and what value is necessary for the variable to be selected
Eigenvalues (within factor) are the sum of squared loadings within a factor An Eigenvalue > 1.0 suggests it should be selected
48
What are communalities and what value suggests a variable should be removed
Communalities (within variables) are the sum of squared loadings within a variable, across selected factors A communality < .30 suggests the variable is unreliable and should be removed
49
How can we figure out how many components / factors should be "extracted"
1. Kaiser's criterion (checking for eigenvalues > 1.0) 2. Cattells' scree test - this plots eigenvalues against component numbers in order of size (retains those above the elbow/ "debris")
50
What are the forms of rotation and when are they needed
Orthogonal rotation (axes are rotated on right angles) Oblique rotation (moving each axis independently) The rotation is used when one factor explains most of the variance. It is used to redistribute variance among the factors
51
What are the 7 sequences of operation to conduct Exploratory Factor Analysis (EFA) CEFSRCR
1. Compute matrix of correlations (check strength of correlations 2. Extract / choose factors (Kaiser's criterion or Cattell's scree test 3. Examine factor loadings (check communalities are over .30) 4. Is the factor structure simple or complex (are there clear differences in component matrix) 5. Do the factors need rotating (orthogonal or oblique) 6. Check factor loadings again (to see if matrix is more simple) 7. Repeat until satisfied
52
What is a Type 1 error
A Type 1 error occurs when a difference / relationship is accepted when in fact there isn't one
53
What is a Type 2 error
A Type 2 error occurs when it is concluded that there is no difference / relationship when in fact there is one
54
What is statistical significance criterion compared with statistical power
Statistical significance criterion is the probability of making a Type 1 error (usually set at .05) Statistical power is the probability of avoiding making a Type 2 error (usually set at .80)
55
What are the limitations of statistical significance testing
Statistical probability cannot measure the magnitude of the result as it may reflect the sample size or the effect size
56
What are three things that influence statistical power
Effect size Sample size Precision of measures
57
What does Effect Size measure
Effect size is a measure of the magnitude of a result, independent of the sample size (measured in d) small = .20, large = .80 A larger sample size can result in a Type 1 error
58
What can a large sample size result in
A large sample size increases the chance of making a Type 1 error, and decreases the chance of making a Type 2 error Larger samples result in smaller 'standard errors', which result in greater power as effects are easier to detect
59
Explain the impact of precision of measures
More reliable measures result in more precise estimates of latent variable (less error variance and smaller standard errors)
60
How should you report a Power Analysis
You must state all inputs determining the result - Effect size (Cohen's d) - Significance level (alpha value usually .05) - Required power (usually .80 unless it's post-hoc) - Sample size
61
What are some criticisms of qualitative research
It is too subjective and influenced by personal bias It doesn't represent the population It cannot be replicated It is not systematic (can be if done correctly)
62
What are different data collection methods for qualitative research
Observations Interviews Focus groups Questionnaires (open ended questions)
63
What is one method of analysing qualitative data
Content analysis Themes can either be based on a pre-defined topic (deductive) Themes can emerge from the data (inductive)

SEM 2 (6 decks)

Brainscape helps you reach your goals faster, through stronger study habits.
© 2025 Bold Learning Solutions. Terms and Conditions