Lecture 4 (final) Flashcards
why is it difficult for psychologists to be objective
Doing science is part of human behaviour. When psychologists do science, they engage often behaviours that they are studying. Psychologists are part of their own subject matter, and therefore it is difficult to be objective
what does it mean by All science is a social enterprise
There are conventions, traditions, shared assumptions, and such things as peer review, which ensures that too controversial positions are moderated and even filtered out (see Kuhn’s idea of a paradigm; “Science progresses funeral by funeral”)
is there a “pure” science and a “subjective” science?
Science is a mixture of both, with the hope that all parties engaged in to their best to further progress towards the truth, or whatever aspect of truth humans are fit to comprehend.
what are the 2 basic assumptions underlying experimental approach in psychology
Researchers only influence the participant’s behaviour to the extent that they decide what hypothesis to test, how the variables are to be operationalised, what design to use, etc
The only factors influencing the participant’s behaviour are the objectively defined variables manipulated by the experimenter
what experimental problems undermine the 2 beliefs
experimenter bias, demand characteristics, representativeness, artificiality, and internal vs external validity
what is experimenter bias
????
what was significant about the mice experiment and the observations of their pain
Exposure of mice and rats to male but not female experimenters produces pain inhibition
this happened because when the mice could smell testosterone (of an alfa male) they attempted to conceal their pain in order to look stronger
so this was a factor that they did not even know existed until the experiment happened
what are demand characteristics
The person being studied is not only a passive responder, but might engage in the experiment actively, e.g., trying to solve the problem what the experiment is actually about
how might demand characteristics cause a problem in the experiment
This can lead participants to respond in a way to confirm the assumed hypothesis, in order to please the experimenter
The sum total of cues of the experimental situation that convey the experimental hypothesis to participants are called _________
The sum total of cues of the experimental situation that convey the experimental hypothesis to participants are called the demand characteristics
how is representativeness important in experiments
The data obtained do not represent humanity in general
It is doubtful, whether obtained effects can be found in other populations
give an example of representativeness
Psychological research is done predominantly on white North-American undergraduate students enrolled in psychology courses at colleges and universities
The bias is Anglocentric, Eurocentric, Androcentric, and, used to be Masculinist
How is artificiality a problem in experiments
Psychological research usually unfolds inside research laboratories located at research institutes and university departments
Participants are subjected to often bizarre tasks, which they are asked to perform in the name of science
Often these tasks are the result of a reductionist approach, aimed at identifying mechanisms of cognition/behaviour
It is unclear, however, to what extent the observed behaviour reflects the normal operation of the brain in natural situations and under natural conditions. Animal research (in psychology) faces the same problem
what is a solution to artificiality
One solution are field studies in which the experimenter tries to observe natural behaviour in the wild without being noticed by the observed population
what is FIELD STUDY/NATURALISTIC OBSERVATION
Observe natural behaviour without attempt to control or manipulate it
why is FIELD STUDY/NATURALISTIC OBSERVATION difficult
There is no control over the behaviour of subjects, and therefore it is difficult to determine the cause of the behaviour
despite its difficulties, how might researchers used field study/naturalistic observation to conduct an experiment
Some research strategies involve first field research to describe and identify behaviours, which then can be followed up in other forms of experimental approaches with increased control over behaviour
Participants are observed in an arranged situation.
Experimenter interacts directly or indirectly with participants
what extent do factors influence the behaviour of the participants and to what extent this is moderated by indirect or direct interaction with experimenter when it comes to field study/naturalistic observation
Unclear to what extent situational factors influence the behaviour of the participants and to what extent this is moderated by indirect or direct interaction with experimenter
what are the things to consider when doing observational studies
coding, reactivity, observer bias, experimenter expectancy effect
what is coding
Observational techniques involve the systematic assessment and coding of overt behaviour. Coding involves generating of behavioural categories and noting when and how often behaviour of a certain category was observed. The coded behaviour can then be used to compute indices to quantify behaviours
what is reactivity
Observational studies need to consider whether observer shall be visible or not. The presence of the observer can influence behaviour. This alteration is called reactivity
what is observer bias
Systematic errors in observation that can arise from the observer’s expectations. For example, in many societies women are freer to express sadness than men. Observers coding for facial expression of sadness may tend to rather interpret ambiguous facial expressions as sad in females than in males
what is experimenter expectancy effect
Observer expectations can change the behaviour of the observed being. One method to prevent this effect is to blind the experimenter to the experimental conditions and/or hypotheses
what is an example of experimenter expectancy effect
students were told to train rats to go through a maze, one group was told that their rat was bred from stupid rats and the other group was told that their rat was bred from smart rats… the result was that the stupid rat group didn’t even attempt to train their rat well because they had low expectations for their rat…. in reality neither group had anything “special” about their rat (they did not come from a line or smart or stupid rats)
what are case studies
Report of a single individual, group, situation.
Collection of highly detailed descriptions that other research settings not readily offer
what are the downfalls of case studies
Causal relationships often hard to establish, and results may be hard to generalise to others (because case studies happen so rarely)
what are the potentials of case studies
Can provide strong motivation for future research efforts to test hypothesis derived from case study
do the same case studies happen often
nope! or it wouldn’t be a case study
what is an example of a famous case study
Henry Molaison (Patient HM)
what was the Performance Profile for Case HM
Impaired: episodic memory recognition memory for recent objects/places explicit memory task Semantic memory : priming procedural memories conditioning implicit memory tasks
why was case HM a Highly influential case study
This case study marks the beginning of a research field studying the role of hippocampus in memory
what are the different approaches to research
OBSERVATIONAL STUDY: FIELD STUDY/NATURALISTIC OBSERVATION
Case Studies
SELF REPORTS AND (STRUCTURED) INTERVIEWS
what are SELF REPORTS AND (STRUCTURED) INTERVIEWS
Participants are directly asked about aspects of their existences, such as living conditions, thoughts, feelings, actions etc
how are self-reports and (structures) interviews conducted
These questions can be little and highly structured
what are highly structured questions
highly structured questions are given in a determined sequence with determined answer options
what are less structured questions
Less structured questions allow participant to respond in free form
what for do self-reports and (structures) interviews take
These research efforts use interviews, surveys, questionnaires
what is a problem that comes from self-reports and (structures) interviews
Interviews need to deal with biases of many kinds. In general, participants have the wish to appear good or in a light that they deem positive. Questionnaire designers have to consider to what extent the questions produce socially desirable responding, or faking good
how do researchers attempt to avoid the problem with self-reports and (structures) interviews
Often, interviews ask for the same information repeatedly in different forms to try to capture (more or less successfully) response biases
what are Correlational studies
Correlational studies explore how variables are naturally related, describing and predicting relationships between the variables
what can’t Correlational studies detect
Correlational studies cannot detect causal relationships between the variables
give an example of how Correlational studies cannot detect causal relationships between the variables
Example: most colleges require SAT scores in their student applications because it has been found that SAT scores and academic performance positively correlate – the higher the SAT score, the better academic performance
Scoring high on the test does not cause better academic performance. High academic performance also does not cause higher test scores. These variables are correlated, but not related in a causal way (A causes B)
what do Correlational studies allow for
Correlational studies allow making predictions, and these predictions can be tested in controlled experiments to search for causal relationships between identified variables and observational outcomes
what are the directions of correlation
Positive correlation
Negative correlation
Zero correlation
what is Positive correlation:
both variables move in the same direction (the more X, the more Y; the less X the less Y)
what is Negative correlation:
the variables move in opposite directions (the more X, the less Y; the less X the more Y)
what is Zero correlation: the
variables are not predictably related
what are some challenges of correlational studies
Directionality problem
Third variable problem
what is Directionality problem
the direction of the relationship between variables can appear ambiguous. Causations cannot be determined, so it remains unclear whether a positive or negative correlation results from the increase in one or the other measured variable
what is an example of Directionality problem
For example, if you sleep less, your stress levels increase. But it is possible that increased stress levels make you sleep less
what is Third variable problem
this is a basic problem of all correlational studies. The relationship between the two measured correlated variables might be dependent on a third, not measured, variable.
give an example of Third variable problem
Example, texting while driving (A) is correlated with driving dangerously (B):
Risk taking (C) causes some people to text while driving: C→A …AND;
Risk taking (C) causes some people to drive dangerously: C→B
what is experimental method
An experiment is a research method that tests causal hypothesis by manipulating and measuring variables
what are The manipulated variables called
independent variables (IV)
what are the variables to measure the effect of the manipulations called
dependent variables (DV)
what are the components to the experimental method
Independent variable (IV) Dependent variable (DV). Experimental group(s) Control group(s)
what is Experimental group(s)
Participants who receive a treatment.
what are control groups
Participants who receive no treatment or who receive a treatment that is unrelated to the independent variable being investigated
what are the different levels to the independent variable
independent variables have at least two levels: a) treatment level; b) comparison level.
explain how test the hypothesis that cell phone use affects driving performance (give all the components and explain them)
Independent variable (IV). In order to test the hypothesis that cell phone use affects driving performance, the independent variable would be the type of cell phone use
Levels: independent variables have at least two levels: a) treatment level; b) comparison level. In case of the cell phone use study, the independent variable could have four levels: (texting) vs (calling) vs (only holding) vs (no use)
Dependent variable (DV). The measured/observed behaviour. For example, in the cell phone study the DV could be reaction time to red lights
Experimental group(s). Participants who receive a treatment. For example, in the cell phone study you will have participants in the two experimental groups “Texting” and “Calling”
Control group(s). Participants who receive no treatment or who receive a treatment that is unrelated to the independent variable being investigated. For example, in the cell phone study the control group would not use the cell phone. The study could also have a control condition in which participants are asked to talk to listen to the radio or to talk to a passenger in the car This allows to study causal relationships between the IV (e.g., type of cell phone use) and the DV (reaction time to red lights). If the IV consistently influences the DV, then the IV is assumed to cause changes in the DV, all other things being equal
in order to allow causal inferences, experiments depend on what
rigorous control
what are all the parts to establishing causality
control confound Population sampling Random sampling Convenience sampling Random assignment
what is control
All steps taken by the experimenter to minimise the possibility that other variables besides the IVs influence the DV
what is confound
Anything that affects a DV and that varies unintentionally between the IV levels. For example, in the cell phone study a type of car could be used that most participants have little experience operating (e.g., in the US instead of automatic transmission, a car with manual transmission is used). This could affect performance and aggravate or generate effects of cell phone use on driving performance. Thus, the reaction time to red lights could be confounded with the driver’s competence to operate the car. Other confounds could be time of day the experiment is conducted, etc.
what is population sampling
The selection of the participants (i.e., the sample from the population) for a behavioural study is critical to determine the domain of applicability of the results
define population
Population refers to everyone in the group in which the experimenter is interested (e.g., college students; retired bartenders).
define a sample
A sample is a group of people chosen from the population
what is a representative sample
The sampling process should lead to a true representation of the population
what is random sampling
The best way to ensure that the sample is representative is to randomly select a sufficient number of participants from the population. Larger samples lead to more accurate results of the experiment. However, large samples can be too costly in light of available resources
what is sample size estimation
Using a mathematical technique called a priori power analysis, it can be estimated how many participants need to be sampled in order to be able to detect meaningful effects with the experiment
what is convenience sampling
Most psychological research at academic institutions samples from the most conveniently available population, usually undergraduate psychology students
what is random assignment
Participants in the sample are randomly assigned to the experimental conditions (i.e., the levels of the IV). Each participant in the sample has the same chance being assigned to any treatment level.
There are individual differences between the participants. Random assignment ensures that on average these differences are comparable between experimental groups
what is a problem with random assignment
Selection bias
what is Selection bias
The assignment might not be truly random (e.g., participants differ in unexpected ways). For example, in the cell phone study it could by accident happen that many participants assigned to the “texting” group have a lot of experience with texting. Researchers try to anticipate these possible factors and use them to control their effect by pseudo-randomisation
what is Pseudo-randomisation
Participants are randomly chosen from the sample to be assigned to an experimental group. If their assignment would lead to more people with a confounding factor (e.g., experience texting) being assigned to an experimental group, another participant will be selected for assignment, until all participants from the sample have been assigned
what is the design principle of between subjects design
a separate group of participants (ideally equally in size) is assigned to each of the different conditions by random or pseudo-random selection, in order to promote groups that are equal in respect to characteristics influencing the DV
how are the participants in between subjects design separated
by random or pseudo-random selection
what are the Advantages of between subjects designs
each participant only subjects to one treatment condition only. The obtained DV measure is therefore less influenced by (a) practice/experience effects; (b) fatigue/ boredom of participants; (c) sequence effects. This design is versatile and can be used for many research questions
what are the Disadvantages of between subjects designs
(a) relatively large number of participants needed (large N; N stands for number of subjects); (b) measurements obtained from different participants, inducing variance due to different characteristics
what is the design principle of WITHIN SUBJECTS DESIGN
all participants exposed to all treatment conditions. Therefore, the experimental groups are equivalent, which increases the study’s internal validity
what is another name for WITHIN SUBJECTS DESIGN
OR REPEATED MEASURES DESIGN
what are the Advantages of within subjects designs
(a) fewer participants are needed than in a corresponding between-subjects design; (b) reduced measurement errors because individual differences between treatment groups are eliminated
what are the Disadvantages of within subjects designs
repeatedly measuring may promote (a) progressive error, because performance changes between treatment conditions are in addition to treatment influenced by the experience of participating in an experiment (e.g., fatigue, practice effects); (b) carryover effects, treatment in a particular condition might change participants such that it affects performance in a later condition
what are the 5 types of validity
External validity Internal validity Construct validity Face validity Criterion validity
what is External validity:
the degree to which a result obtained in an experiment can be generalised to other (everyday) situations
where can the results from external validity be found
An experiment with high external or ecological validity produced results that can also be found in real nature
what is internal validity
refers to the quality of an experiment, precisely, whether in an experiment the observed effects measured with the DV are due to the independent variables (IVs) and not to confounds
what happens if there is a lot of control over the IVs
The greater the control over the IVs, the more artificial the experimental situation, and the lower the external validity
give an example of an experiment lacking internal validity
Hypothesis: Special Tutoring will improve test performance on final.
Bad research strategy: draw one sample and assign all participants in it to the same treatment condition
give an example with good internal validity
Hypothesis: Special Tutoring will improve test performance on final
get one group
split them into 2 groups
give one group the hypothesis treatment, and leave the other alone
compare the end results
what is contract validity
does the test relate to the theoretical concepts from which the test was derived? E.g., to what extent does an IQ test measure intelligence as defined in a theory of intelligence? Construct validity requires that it is established that the phenomenon to be measured indeed exists (hence dependent on a theory)
what is face validity
does the test measure what it declares to measure? Eg., does an IQ test actually measure intelligence?
what is Criterion validity:
what is the relationship between the test and other correlated measures (the criteria)? E.g., how well does the outcome of an IQ test correlate with GPA/academic performance/SAT/GRE etc?
what are some other possible error sources arising from measurement
random error
systematic error
what is random error
Your method of measurement produces a random artifact in the data. For example, if you stop participant reaction time by hand with a stopwatch, you will not always be equally accurate. The degree or error varies randomly each time a measurement is taken
what us systematic error
Your measurement is inaccurate by a constant amount. For example, the stopwatch you use to measure reaction time is off by 0.25 seconds. Each measurement will be wrong by this constant amount
is metatheoretical structuralist a section of Wundt’s Structuralism
no, absolutely not
what is the metatheoretical structuralist view
theories consist of typically several theory elements. These elements are structures
In addition, theories need a description of the paradigmatic method. This defines the range of intended applications of a theory
according to the metatheoretical structuralist view, what are structures
Structures are similar to systems. They consist of atomic parts and their relations. They are characterized by theoretical definitions and set-theoretical predicates
what branch does the metatheoretical structuralist view go under
EMPIRICAL STRUCTURALISM
what is the official definition of metatheoretical structuralist view
A theory-element is a set-theoretical structure, consisting of two elements: (1) a theory core; (2) a set of intended applications of the theory core. Structuralist theories usually consist of several theory-elements
according to the official definition of the metatheoretical structuralist view, is structuralism universally applicable
According to this definition, in structuralism a theory is not universally applicable. It is only applicable to the range or domain of intended and/or explicitly stated applications: THE IDEA OF EXTERNAL VALIDITY IS IRRELEVANT FOR THE GOODNESS OF A THEORY
how does THE IDEA OF EXTERNAL VALIDITY play into the metatheoretical structuralist view
THE IDEA OF EXTERNAL VALIDITY IS IRRELEVANT FOR THE GOODNESS OF A THEORY
who started Logical empiricism
Bacon and Descartes
what is Logical empiricism
Logical empiricism (starting with Bacon and Descartes) assumed that science can rest on a secure base of pure, objective empirical observations that are independent of the observer and of theory: what is can actually be discovered by human observation
is the Logical empiricism approach tenable
no
what is distribution (in a graph)
the spread of the data points across the range of possible measurements
In order to summarize and understand the data (complete the analysis), what must first be completed
a frequency distribution
what are bins (in data)
categories, or ranges
what are bins used for
bins (categories, or ranges) would be defined to group and summarize the data
when would the histogram be generated
at the end
what are the different types of data distributions (in graphs)
skewed left
skewed right
normal distribution
what is a graph called when it is skewed left
positively skewed
what is a graph called when it is skewed right
negatively skewed
what is a graph called when it is normally distributed
normally distributed
what are the different measures of central tendency
mode
Median
average (or mean)
what is mode
the value that occurs most often in a dataset. In a frequency distribution, this is the category with the highest frequency
what is median
the data point in a dataset for which 50% of all data points are higher in value, and 50% of all data are lower in value.
what is average (or mean)
the average value of a set of data, computed by adding all values and dividing their sum by the number of data points that were added up
what are the central tendency measures for normal distributions
In a perfectly normally distributed dataset, mode, median, mean are the same value.
In empirical data sets that satisfy criteria for normal distributions, the values are usually not identical, but very close to each other
are the central tendency measurements in skewed distributions the same
In a negatively skewed distribution mode, median, and mean are different. For these distributions, it is best to use the median to characterize the central tendency. The mean is too much affected by outliers (extreme values), and does not well represent the distribution
where does variability come from
Variability results from various sources, such as measurement errors, differences between the subjects participating in the study, etc
when is Measures of central tendency only meaningful
Measures of central tendency are only meaningful if accompanied by a measure of variability
what is Standard deviation
Standard deviation: is an index of the amount of variability in a data set. Standard deviation is the average distance of the data points from the mean. A large standard deviation means the data in the distribution are spread wide around the mean, a small one that they are closely scattered around the mean
what the hell does this even mean
In a normal distribution, about 68% of the values are within ±1 standard deviation from the mean, while 95% fall within +- standard deviations from the mean
In an experiment, independent variables are systematically varied…. how is the effect of this manipulation measured
the effect of this manipulation is measured with the dependent variable.
In an experiment, independent variables are systematically varied, and the effect of this manipulation is measured with the dependent variable.
This leads to data obtained under different treatment conditions. The data points form distributions
The question now is what
if these distributions are different, is the difference meaningful, i.e., big enough to attribute it to the treatment, or too small
how do we know if the change is chance or effect
what is expected if the data between groups significantly bigger
If the group difference in the DV are sufficiently bigger than what we would expect simply by chance, then we believe this difference to be statistically significant
what is Null hypothesis (H0)
The differences between the groups are due to chance
what is Experimental hypothesis (H1).
The differences are due to manipulating the IV
what is the cut-off post for significance
this probability that the difference between the treatment groups is due to chance is assessed with the p-value. This is also called the alpa-error (of type 1 error) — it is the probability to falsely reject the H0 (hypothesis that the results are due to chance) in light of the results
what is the p-value also called
This is also called the alpa-error (of type 1 error)
what is Type II error
The probability to falsely reject the H1, i.e., to conclude that the differences in DV are due to chance, while in fact they were caused by the treatment
by convention, what is p always set to
By convention, p is set to p=0.05
what does p=0.05 it mean
(i.e., 5% chance to falsely accept the H1, namely that the difference is due to the treatment).
SOME CRITICAL ISSUE REGARDING SIGNIFICANCE TESTING (p values)
Level of significance
Sample size effects
Effect size
explain SOME CRITICAL ISSUE REGARDING SIGNIFICANCE TESTING:
Level of significance.
p=0.05 considered often too low a criterion because 5% of all tests performed will produce by chance a significant result. Therefore, p=0.01 is often used or proposed. Lower p-values are assumed to be more conservative.
explain SOME CRITICAL ISSUE REGARDING SIGNIFICANCE TESTING: Sample size effects
The bigger the sample, the more likely it becomes that even very small differences between the groups become statistically significant
explain effect size in relation to explain SOME CRITICAL ISSUE REGARDING SIGNIFICANCE TESTING:
The effect size gives a measure of how big the group differences are for statistically significant results (Cohen’s d is a measure of effect size).
For example, with d=0.0, your probability would be 50% (chance) that you could guess, based on a test score alone, whether the score comes from a control or treatment condition.
For example, with d=3.0, your probability to make this guess correctly would be 99.9%
