Research methods and statistics 1 (year one) Flashcards
1
Q
What is a hard science?
A
a science that is objective and measurable e.g chemistry
2
Q
What must a scientifically sound experiment consist of?
A
- operational definitions
- suitable sample size
- control
3
Q
What is the empirical approach?
A
science, evidence-based
4
Q
When did Wilhelm Wundt open the first psychology lab?
A
1879
5
Q
What is introspection?
A
Paying attention to and analysing your own thought processes
6
Q
What is the order of the scientific method?
A
- observation-theory-hypothesis-research-research data
7
Q
What is a theory and what must it consist of?
A
- general principals for outlining or understanding
- must include empirical investigation, prediction, explanation
- must be falsifiable
8
Q
Who promoted falsification?
A
- karl popper (1934)
9
Q
What does a good theory consist of?
A
- testable hypothesis
- guiding research and organising empirical evidence
- be supported or refuted
10
Q
What is a hypothesis?
A
- a theory based prediction
11
Q
To be scientifically testable what must a hypothesis be?
A
- clearly defined
- non-circular
- deal with observable phenomena
12
Q
What are examples of famous studies that are not scientifically sound?
A
Asch (1951), Zimbardo (1971)
13
Q
What are some of the methodological flaws of the Stanford prison experiment?
A
- researcher bias, small sample size, not representative sample, most guards were not violent, worst guard based behaviour on “cool hand luke”
14
Q
What do we need to infer causation?
A
- correlation/co-variation
- time-order relationship (cause has to come before effect)
- eliminate other possible causes
15
Q
- What is the independent groups design?
A
Groups that are made up of different people
16
Q
- What do independent groups look at?
A
difference in performance between subjects
17
Q
What is an independent variable?
A
- IV = variable we manipulate
18
Q
What is a dependent variable?
A
- DV = variable we measure
19
Q
What are some advantages to independent groups?
A
- no fatigue or boredom
- ## no carry over learning
20
Q
What is a natural groups design
A
- IV not manipulated as it is already naturally occuring
21
Q
What does the within groups design measure?
A
- repeatedly measure the same people on the same DV
- controls for individual differences
- ppts may do all conditions at the same time or different times
22
Q
What does power refer to?
A
- the probability that you will find a statistically significant difference when it actually exists
23
Q
What is error variance?
A
- variation caused by individual differences
- reducing error variance makes a significant result more likely
24
Q
What are advantages of within-subjects designs?
A
- individual differences not a problem
- more powerful
- fewer ppts
- more convenient
25
What are the 4 levels of data?
1. nominal
2. ordinal
3. ratio
4. interval
26
What is nominal data?
names, categories
27
What is ordinal data?
data is ordered (e.g on a likert scale)
28
What is interval data?
data points have a similar interval between them (e.g height)
29
What is ratio data?
same as interval, but x has a zero value
30
What is qualitative research?
no set hypothesis
explores opinions, experiences etc
just from asking people e.g interviews
31
What is quantitative research?
numerical data analysed with maths based methods
| via surveys, tasks etc
32
Give examples for each data level
```
- Nominal examples
male/ female, smoker/non-smoker
- Ordinal examples
shoe size, position in race, subjective opinion (likert scales)
- Interval examples
voltage, temperature
- Ratio examples (CANNOT go below 0)
Height, weight, test scores
```
33
What are descriptive statistics?
- describing data
- see what data "looks like"
- looks at central tendancy and measures of dispersion
- only tells us about our sample, not population
34
What are inferential statistics?
- use sample to make inferences about the population
| - help us reach conclusions beyond our data
35
1. What does the statistic you use depend on?
| 2. What are the two main types of descriptive stats?
1. - data level
- distribution of data
2. - measures of central tendancy
- measures of dispersion
36
What are measures of central tendancy?
- this is how "most" people behave
| - measures : mean, median, mode
37
1. What is the mode?
2. What is the median?
3. How do you calculate the median?
4. What is the mean?
1. - most common value or score
- mostly used with nominal data
2. - central value in a data set ordered from lowest to highest
- mostly used with ordinal level data/ skewed data
3. - add up the two scores in the middle and divide by
2
4. - add up all the scores and divide by the number of values
38
1. What does a histogram show?
2. What is normal distribution?
3. What is skewed data?
1. - the frequency of the data
2. - scores average around the middle, very few extreme scores
- bell-shaped curve
- median
3. - mean is not central
- extreme scores affect the mean
- not normally distribution
39
1. How do we get around skewed data?
| 2. What is the 5% trimmed mean?
1. - remove extreme results
2. - take off 5% of scores from each end
equation = %required x no. scores
100
40
What is variance and standard deviation?
- average distance of scores from the mean
41
What are the measures of dispersion?
- range, interquartile range, variance and standard deviation
42
What is the range?
- difference between largest and smallest score
43
What is the IQ range?
- difference between middle 50% of scores
44
How do we calculate the IQ range?
- want to find the range of the middle 50 % of scores (second and third quartile)
- % required x no. scores
100
- answer rounded gives how many scores to take from top and bottom, which you find the range from
45
What is variance?
- deviation of scores from the meaan
- subtract the mean from each score, then find the average
- add up and square all scores, take them away from sum of scores squared divided by n, divided by n-1
- n = how many values there are
46
What is the standard deviation?
- square root of the variance
47
What descriptive statistics should I consider for each data level?
Nominal = frequencies/ %/ mode
Ordinal = median (+range)
Interval/ ratio skewed = median (+range)
ND = mean (+/-SD)
48
What data is chi-square used for and what does it assess for?
nominal data level
| - association between categorical variables
49
what is the equation for expected frequencies?
(values taken from observed frequencies) (row total)x(column total)
N
50
what are p values?
evaluate how well the data in your sample supports the null hypothesis
51
what do high and low p values mean?
o High p value : data are likely with a true null
| o Low p value : data are unlikely with a true null
52
at what value is the p-value said to be significant?
below .05
53
what is the alpha level?
level at which we accept result to be significant
54
what is the effect size for 2x2?
phi
55
what do different phi values represent?
- Φ .1 = small
- Φ .3 = medium
- Φ .5 = large
56
what are type 1 and type 2 errors?
- Type 1 : rejection of true null hypothesis (false positive)
- Type 2 : accepting a null hypothesis (false negative)
57
what is a Bonferroni correction?
- Change the alpha level to prevent type 1 errors
| o Divide alpha level by number of tests that will be conducted
58
what is p hacking and how is it done?
: method of manipulating data to achieve significant results
Multiple analysis
Omitting other info
Controlling for variables
Analyse part way through then gather more data until a significant result is found
Changing DV
59
what is the null hypothesis and alternative hypothesis?
- Null hypothesis = statement of no difference
o True until there is evidence against it
- Alternative hypothesis = statement of difference or association
60
What are one-tailed and two-tailed hypothesis?
- One tailed hypotheses = state which direction the effect will be in (e.g those that subscribe to Zoella will be more likely to choose the unhealthy snack
- Two tailed hypotheses = no direction stated
61
what do box plots show?
show medians, ranges, IQ ranges, skewness etc
Range = upper adjacent value – lower adjacent value
Uneven whiskers = skewness
62
how do you find the IQ range from a box plot?
IQ Range = upper hinge – lower hinge (whole box)
63
What is positive/negative skew?
- Deviation from symmetry
- Show a big difference between means, medians, and mode
- Extreme scores affecting the mean
- Positively skewed : scores greater than the mean skewing
- Negatively skewed : scores lower than the mean skewing
64
what is leptokurtic/paltykurtic distribution?
- Refers to extent to which scores cluster at the tails of the distribution – changes pointiness
- Positive kurtosis : leptokurtic distribution
- Negative kurtosis : platykurtic
Flatter than normal
65
what is the boundary for skewness?
more than twice the standard error
66
what is indicated when there is no overlap between two confidence intervals?
: difference between parameters is significant
67
what are some disadvantages of the between-subjects design?
- Between subjects : independent, looks at performance between subjects/groups
Disadvantages
High sample size
Individual differences
68
what are confounding and situation variables?
- Confounding variable : extraneous variable that influences results
- Situation variables : variables in condition that could confound, e.g environment, temperature, time of day
69
what are expectancy effects?
- Expectancy effects : expecting an effect can cause that effect e.g expecting a substance rather than a placebo may lead to experiencing some of the effects
70
what are the three types of balancing/matching for between subjects?
- Balancing and matching techniques
1. Random allocation
2. Matched group design
3. Natural group design
71
explain random allocation design?
RANDOM ALLOCATION DESIGN
- Participants randomly assigned to groups
- Controls for participant variables
- Sample size should be larger
72
explain matched group design?
MATCHED GROUP DESIGN
| - Matches participants based on a certain characteristic (sometimes DV)
73
explain the within-subjects design and its disadvantages
WITHIN SUBJECTS DESIGN
- Repeatedly measure the same people on the same DV
Disadvantages
o Boredom/ fatigue
o Order/practice effects
o Individual differences
o Time consuming conditions
o Can’t use for experiments where task cannot be repeated (e.g first impressions)
o Can’t be used if there’s differential transfer (effects of one condition affect performance in some conditions (e.g using cannabis then placebo)
74
what is differential transfer?
effects of one condition affect performance in some conditions (e.g using cannabis then placebo)
75
state and explain different order/practice effects
Learning
Fatigue
Habituation (leads to reduced response)
Sensitisation (leads to greater response)
Contrast (may lead to less effort if initially rewarded
Adaptation (e.g low light levels, drug effects)
76
what is incomplete within-subjects design
- Each condition given to each participant once
- Order of administration varied
- Practice effects balanced
77
what are the main counterbalancing methods for incomplete within subjects?
o All possible orders
| o Selected orders
78
describe all possible orders counterbalancing
Have to calculate the factorial based on levels of IV (result of multiplying number by all numbers less than it)
Used on 3-4 conditions or less
79
describe selected orders counterbalancing
Based on Latin square
Used for more than 3 conditions
Each condition occurs once in each position
Each condition precedes/ follows each other condition only once
80
what are the main counterbalancing methods for complete within subjects?
- Each condition administers several times (different orders each time)
- Practice effects balanced for each participant
- 2 main counter balancing methods
o Block randomisation
o The ABBA design
81
describe block randomisation and ABBA design
1. BLOCK RANDOMISATION
Consists of all conditions
Participants complete the conditions several times, each time in a different order
2. ABBA DESIGN
Presents one random sequence of conditions, then the opposite sequence
82
explain observation without intervention and advantages/diadvantages
```
o Naturalistic observation
o Behaviour occurs naturally, experimenter is a passive recorder
ADVANTAGES
High external validity
Can investigate complex social situations
Useful for developing theories
DISADVANTAGES
Time consuming/ expensive
Description, not causation
Not useful for specific hypotheses
```
83
explain participant observation and advantages/disadvantages
```
- PARTICIPANT OBSERVATION
o Undisguised
• Researcher part of group
• In depth interviews/ observations
Advantages
No ethical problems
Natural setting
Openly record data
Disadvantages
Behaviour may change due to presence
o Disguised
• Those observed are unaware
• Prevents observer influence
Advantages
Access to particular groups
Natural setting
Disadvantages
Ethical issues
Problems recording data
Researcher bias
Interaction : researcher may change the observeds behaviour
```
84
explain structured observations
- Cause an event or set up a situation
- Observe specific behaviour in a particular setting
- No attempt to control for other variables
- Uses behavioural checklist or code using mutually exclusive categories
- Same procedures across other observers
85
explain field experiments
- Well controlled in natural setting
| - Manipulate IV to observe effect on behaviour
86
explain interobserver reliability
- Consistency in measuring between observers
| - Correlations can be used to check reliability
87
explain observer influence
- Reactivity : participant modifies behaviour when they know they are being observed
Socially normative behaviour to gain approval
Demand characteristics : change behaviour depending on what the expected objective of the research is
- Controlling reactivity : unobtrusive measurement
Disguised participant observation
Adaptation : habituation, desensitisation
Indirect measurement : physical traces, archival data
- Expectancy effects with observer bias : knowledge of hypothesis/ previous research
Can be controlled by blind observers
88
explain observation with intervention
Precipitate an uncommon/ difficult to observe event
Gain access to closed event/situation
Establish comparison by adding/manipulating IVs
Control antecedent events/ behaviour
Vary qualities of a stimulus event
3 kinds : participant observation, structured observation, field experiments
89
Define correlational analysis
- Assess relationships between variables
90
define correlation coefficients
- tells us about the strength of the association : range from -1 to +1
- negative values : negative correlation (-1 = positive)
- positive value : positive correlation (+1 = positive)
91
define positive and negative correlations
positive correlation
- as one variable increases so does the other
negative correlation
- as one variable increases the other decreases
92
what are other explanations for correlations?
- other explanations for correlations: 3rd variable, chance
93
what are the two main inferential tests for correlation data?
- spearman rank correlation
- ordinal level data
- skewed ratio/ interval level data
- pearson product-moment correlation
- normally distributed interval/ratio level data
94
What data levels are spearman rank correlations used for?
- spearman rank correlation
- ordinal level data
- skewed ratio/ interval level data
95
what data levels are pearson product-moment correlations used for?
- pearson product-moment correlation
| - normally distributed interval/ratio level data
96
explain how spearman's correlations are determined
- uses the ranks of the data, not the actual data
- not influenced by skewed data
- when we have two values that would get the same rank we add together the ranks and divide by how many tied scores there are
97
describe how you would report results for spearman's correlation
- shows a positive/negative/no correlation
- spearman's : rs
- degrees of freedom = n-2
- order = rs (degrees of freedom) = correlation coefficient, p
98
explain one tailed and two tailed tests
- one-tailed : direction is stated
- we can halve the two-tailed p value to find one-tailed
- alpha values : .025
99
define alpha values
- alpha value : level at which effect is significant
| - typically .05, so p values below .05 are significant
100
define degrees of freedom
- degrees of freedom : the number of observations in the data that are free to vary when estimating parameters
101
what is the symbol for pearsons?
-pearson's symbol = r
102
what are correlation matrices?
- present lots of variables in a table - correlation matrix
| - APA format
103
explain surveys
- predetermined questions
- includes questionnaires and structured interviews
- can be given online, mail etc
104
give advantages/disadvantages of mail surveys
- mail : + convenient
| - response rate/bias
105
give advantages/disadvantages of internet surveys
```
- internet : + efficient/cheap
+ convenient
+ large/diverse sample
- representativeness
- ethics
```
106
give advantages/disadvantages of phone surveys
- phone : + some questions easier to ask
+ large, diverse sample
- sample bias
- interviewer bias
107
give advantages/disadvantages of group surveys
- group : + captive audience
+ large amount of data quickly
- privacy/anonymity
- pressure
108
give advantages/disadvantages of interview surveys
- interview + same questions/order
+ quantitative analysis
- interview bias/ social context
109
give advantages/disadvantages of personal surveys
- personally : + convenient/ large sample
+ good response rate
- representativeness/demand characteristics/ questionnaire fatigue
110
explain psychometric tests
- ability/ aptitude (e.g numerical/verbal reasoning
| - personal qualities (personality/attitudes)
111
when were surveys first used?
- 605-1905 : Chinese civil service exams used to recruit officials
112
describe the army alpha and beta tests
- 1917 : army alpha and beta tests developed by Robert yerkes
- evaluated intellectual/ emotional functioning
- tested verbal/ numerical ability, (e.g following directions)
- also tested capability of serving, job classification, leadership potential
- beta test - non verbal equivalent
- allowed intelligence classification as superior, average, inferior
- highest to lowest score : white Americans, north/west European immigrants, south/ east European immigrants, black Americans
- test was very amercio/eurocentric (e.g what is crisco, celebrities) and required cultural knowledge
- actually measured level of education / acculturation
113
describe the woodworth personal data sheet
- woodworth personal data sheet
- world war 1 by US army
- a test of emotional stability (susceptibility to shell shock)
- first personality test
114
describe the stanford-binet IQ test
- Stanford-binet IQ test
- used to assess for learning disabilities
- used today for clinical/ neurological assessment and educational placement
115
what is the procedure for designing a questionnaire?
- need a topic, then draft, then reexamine/ revise, then do pilot study, then edit and specify procedures for administering
116
what should questionnaire questions be?
- questions must be simple
- dont use double barralled questions
- avoid using loaded/ guiding questions
- avoid negative wording (e.g do you think students shouldnt pay tuition fees
117
describe open ended questions
- open ended questions
+ detailed answers
+/- quick to design, long analysis time
+ participant led
- subjective interpretation
- partially open-ended (multiple answers given
118
describe close ended questions
- closed questions (e.g likert scales, true/false)
- guessing
- unsubtle
- complex to design, quick to mark
- theory led
- questionnaire fallacy : people will find a box to tick, even if their opinion is not represented
119
describe rating scales
- e.g yes or no, agree or disagree, graphic rating scale
- likert scale : labelled statements of a varying strength (e.g strongly agree to strongly disagree
- each measure given a score (positive question : strongly agree = 5, negatie question : strongly agree = 1
- semantic differential scale : connotative meaning between bipolar adjectives, and rating is placed on a scale inbetween
120
explain order effects in questionnaire bias
- order effects/ priming : detailed questions at first may influecne later general questions
- thinking about how the answer to one question while answering another
- counter balance questions and randomising can help
121
explain demand characteristics in questionnaire bias
- demand characteristics : answer in a certain way to sabotage/ give "beneficial" answers/ look more desirable
122
explain acquiescence in questionnaire bias
acquiescent : always agreeing/ disagreeing, even if it contradicts previous answers
- use a mix of positive/negative questions to overcome
123
explain extreme/ neutral responses in questionnaire bias
- extreme/ neutral responses : may not be concentrating, sabotaging etc
- raw data may need to be disregarded
124
explain cultural bias in questionnaire bias
cultural bias : language could be misunderstood, multiple interpretations of words, differing opinions between cultures, social desirability differs
125
explain attitude questionnaires
- assumptions : attitudes can be verbally expressed
- statements will have the same meaning for all participants
- attitudes can be quantified
- problems : consistency, social desirability, ambivalence, normative response bias (use a lie scale)
- implicity : do the statements express what they should clearly or not?
126
explain word association tests
- used in a clinical setting to determine complexes/ deficiencies used to predict things such as drug use
- advantages : quick/ easy to administer
- predict prospective drug use
- self scoring improves validity
- disadvantages : colloquialism - cant make standardised procedures
- tests may not be implicit
127
explain implicit tests
- implicit cognitive tasks : infer attitude/ beliefs from performance on different tasks
- often use reaction times
- e.g IAT
- automatic association between concepts, used for attitudes towards age/gender/ race etc
- now computer based
- categorised target concepts with an attitude as quickly as possible
- faster association = stronger correlation
- disadvantages : cultural values vs beliefs/ attitudes
- ecological validity
- may not act that way
128
explain reliability in psychometric tests
- reliability : internal (all items measure the same thing)
external (consistent across time and setting)
- test-retest reliability and split-half reliability
129
explain classical test analysis in psychometric tests
- classical test analysis : assumes observed score (X) is made up of true score (T) and random error score (E) : X=T+E
- random error : reading errors, social desirability bias, mood, tiredness
- systematic error : characteristic of test e.g "how often do you go to the cinema" would be influenced by factors like wealth
130
explain validity in psychometric tests
- validity : content validity (covers all behaviour/ aspects)
construct validity (measures theoretical construct)
criterion- orientated validity (correlates with establishes measure)
131
explain standardisation in psychometric tests
- standardisation: standardised instructions and procedures
132
explain established population norms in psychometric tests
established population norms : should be able to compare results to an appropriate established tests/theories
133
describe the two types of questionnaires
- knowledge based : ability, aptitude, achievement e.g intelligence tests, clinical assessment instruments
- person based : personality, mood, attitude to assess differences between people
134
describe the two types of response references
- normative reference testing : scores compared to norm e.g mean/ median split
- criterion reference testing : scores compared to pre-determined criteria e.g determine if someone is at risk
- restrictive : doesnt take in to account individual differences or non-clinical samples
135
define regression
- focuses on predicting variance in an outcome (criterion or response variable) from predictors (IV)
- creates a statistical model to find out whether model is a good fit for data and find whether there is a significant association/ direction of association
136
give the linear relationship formula
- linear relationship formula = Y = bX + a
- Y : criterion/response variable
- b: slope of the line (based on Pearson's r)
- X: predictor variable (years of experience)
- a: constant or intercept
- calculates line of best fit for the observed data which can be used to make predictions for unobserved values
137
give the regression equation
- Yi = (B0 + B1Xi) + ei
138
explain bivariate linear regression
- two variables
- X is predictor variable (IV/explanatory variable)
- Y is criterion variable (response/outcome/criterion/DV)
139
give the assumptions of regression
- normally distributed continuous outcome
- independent data
- interval/ ratio predictors
- nominal predictors with two categories (dichotomous)
140
define R square
- R square/ adjusted R square
- how close data is to fitted regression line
- proportion of variance explained by the model
- presented as a percentage
- coefficient of determination
141
define ANOVA
- ANOVA
| - measure of model fit : tells us how well regression fits the data
142
define beta coefficient
- beta coefficient
| - number of SDs the criterion variable will change as a result of one SD change in the predictor variable
143
what do we need to interpret a regression?
- we need to :
- assess model fit (f value)
- know how effective model is - R squared value
- know whether an association is significant and direction - beta value
144
explain how to report regression data
- Example :
- a bivariate regression was conducted to investigate the association between years of experience and salary. The regression model predicted approximately 70% of variance in salary, adjusted R^2 = .70, F(1,8) = 22.34, P= .001.
There was a positive association between years of experience and salary 𝜷 = .86, p = .001
145
Explain why multiple regressions are used
- can increase the amount of variance explained by a model by including additional variables
146
Define correlational analysis
- Assess relationships between variables
147
define correlation coefficients
- tells us about the strength of the association : range from -1 to +1
- negative values : negative correlation (-1 = positive)
- positive value : positive correlation (+1 = positive)
148
define positive and negative correlations
positive correlation
- as one variable increases so does the other
negative correlation
- as one variable increases the other decreases
149
what are other explanations for correlations?
- other explanations for correlations : 3rd variable, chance
150
what are the two main inferential tests for correlation data?
pearson product moment correlations
| spearman rank
151
What data levels are spearman rank correlations used for?
- spearman rank correlation
- ordinal level data
- skewed ratio/ interval level data
152
what data levels are pearson product-moment correlations used for?
- pearson product-moment correlation
| - normally distributed interval/ratio level data
153
explain how spearman's correlations are determined
- uses the ranks of the data, not the actual data
- not influenced by skewed data
- when we have two values that would get the same rank we add together the ranks and divide by how many tied scores there are
154
describe how you would report results for spearman's correlation
- shows a positive/negative/no correlation
- spearman's : rs
- degrees of freedom = n-2
- order = rs (degrees of freedom) = correlation coefficient, p
155
explain one tailed and two tailed tests
- one-tailed : direction is stated
- we can halve the two-tailed p value to find one-tailed
- alpha values : .025
156
define alpha values
- alpha value : level at which effect is significant
| - typically .05, so p values below .05 are significant
157
define degrees of freedom
- degrees of freedom : the number of observations in the data that are free to vary when estimating parameters
158
what is the symbol for pearsons?
pearson's symbol = r
159
what are correlation matrices?
- present lots of variables in a table - correlation matrix
| - APA format
