Exam Numero Dos Flashcards
What is Naturalistic Observation?
Observing your participants in their natural environment, without controlling or manipulating variables.
-the natural environment is called the field
What are the two kinds of Naturalistic Observation?
Global Naturalistic Observation
Systematic Observation
What are some issues to consider when it comes to concealing or not concealing the researcher in naturalistic observation?
-if the researcher is not concealed, subjects may change their behaviour (reactivity) which affects the validity of your observations
- if the researcher is not concealed, you will want your subjects to habituate to their presence
What are some advantages and disadvantages of Naturalistic observation?
Disadvantages:
-people may behave differently if they know that they are being observed
-the experimenter has no control (they are descriptive methods; you can only speculate as to the causes of behaviour, you cannot make casual statements)
-inner states can only be inferred from behaviour, they are not actually seen
-data interpretation can be difficult
Advantages:
-we can observe behaiour as is occurs in the natural environment
-properly conducted, it has extremely high external validity
-it is exploratory in nature, so it often provides ideas for future research projects
What is Systematic Observation?
-involves observing a few specific behaviours in specific settings
-more structured and less global than naturalistic observation
-observations are typically QUANTITATIVE in nature
example:
Timing how long drivers wait at a stop sign in the country versus in the city
What is Global Naturalistic Observation?
-involves describing in detail that behaviours of your subjects in their natural environment
-the observations are typically QUALITATIVE
example:
Describing in detail the interactions between primary school teachers and their pupils over a one-month period
What is a case study?
One subject is studied in depth in the hopes of revealing results universally true of the population
-descriptive technique
-typically conducted when the individual possesses a rare or unusual condition (so that more can be learned about their condition
-they allow us to learn about certain physiological concepts that we would not have been able to otherwise given ethical concerns
what is random assignment?
when every participant has an equal likelihood of being assigned to either the experimental or control groups
what effect does random assignment have?
it neutralizes individual differences, making the two groups essentially the same
What is ‘Control’?
holding all other factors in an experiment consistent so that an experimenter can tell if the IV had an effect on the DV
make the groups exactly the same, treat them exactly the same = control
what is a control group?
the group that is not exposed to the treatment.
what is the experimental group?
the group that is exposed to the treatment / manipulation
If she asks, “is this a true experiment” on the exam, what do you look for?
look for random assignment
(not the presence of a control group)
what is a confound variable?
a variable that varies along with the IV (due to a lack of control), so its difficult to say which variable causes the change in the DV
–> can serve as an alternative explanation for the change in DV
- discredits the whole thing
if an experimenter successfully controls confound variables, then what are they said to have achieved?
internal validity in the experiment
what is internal validity?
it means that you can say that the IV causes the DV
- it is the ability of your design to test the hypothesis it was designed to test.
(the extent to which the experimenter controls confound variables, they are said to have internal validity in the experiment)
What is a threat to internal validity?
confound variables!
what are confound variables?
variables that co-vary with the IV; they are alternative explanations for the results
what are 4 examples of threats to internal validity
Skeletons Intervene even months (e)after birth
1) Selection
2) Instrumentation
3) Experimental Mortality
4) Experimental Bias
threats to internal validity:
what is Selection?
if the two groups are different somehow before the experiment begins.
it is an unequal distribution of subject-related variables across the two groups
threats to internal validity:
What is Instrumentation?
confounds may be introduced by changes in the criteria used by observers or changes in the mechanical measuring devices
e.g. two different ways of measuring something (no operational definitions) or like your machine breaks between experiments
threats to internal validity:
what is experimental mortality?
loss of subjects in an experiment, if loss is different across groups then the study will lack internal validity
e.g. a bunch of people drop out / die… then its no longer an equal representation of a population of people (because certain people didn’t want to participate … died)
threats to internal validity:
what is experimenter bias?
expectations of an outcome by persons running an experiment may significantly influence that outcome
- when an experimenter is working with both groups and treats them differently
What are the advantages and disadvantages of a case study?
Advantages:
-Can provide us with a great deal if info on a particular condition
Disadvantages:
-Time consuming
-Descriptive, so you cannot determine the causes of the behaviour observed
What is archival research?
The researcher analyzes existing data (they do not collect the original data themselves)
-manuscripts, letters, photos, videos, books, diaries, music, artwork, computer databases, statistical records, etc.
What are the advantages and disadvantages of archival research?
Advantages:
-valuable as a supplement to traditional data collection methods
Disadvantages:
-may have difficulty obtaining records
-cant be sure of the accuracy of info collected by someone else
-completely descriptive so you cannot establish causal relationships
What is content analysis?
A method used to analyze qualitative data. It allows a researcher to take qualitative data and transform it into quantitative data.
What is a survey?
A descriptive technique designed to gather into from many people, usually by administering a questionnaire
-used to evaluate specific attitudes or behaviours
-called a self-report
what is the trade off between internal and external validity?
tight control = high internal validity
(measures what its supposed to measure)
high internal validity from tight control lowers external validity
(the world isn’t controlled, so it won’t represent the real world as well)
What is a Questionnaire?
Asks people questions about themselves, usually using a paper and pencil format of online testing
-called a self-report
what is the aim of the experimental method?
to tell if the IV actually manipulated the DV
what is POWER?
the ability to detect the effects of the IV if they actually exist. (statistical significance)
What do you do when designing a questionnaire?
-clearly define the topic of your study
-collect demographic info from your participants
what are some ways to increase power?
(5)
- control for confounds
- let other variables that are not of interest vary randomly.
- use large sample sizes
- use a sensitive dependant measure (not too hard that everyone fails / too easy that everyone passes)
- use strong manipulation
What are the two types of questions in questionnaires and surveys?
-Open-ended
-Closed-ended
(definitions on different card)
What is generalizability?
one’s ability to say something about a population based on an observed sample
= external validity
what is external validity?
the extent to which the results of an observation generalize to there situations or are represented in real life.
the degree to which the results of a study can be extended beyond the research setting.
What are open-ended questions? (surveys +questionnaires)
-allow the participant to provide a response in their own words
-they may provide more complete info but are difficult to analyze
-analysis is more qualitative in nature
Example: “Comment on whether you believe a student should or should not arrive to class on time:”
what are the 4 threats to external validity
1) lack of random sampling
2) Mortality (loss of participants from a study)
3) data obtained in a tightly controlled lab setting may not generalize to natural settings
4) reactivity on part of the subjects
Why is mortality bad for external validity?
also what is selective attrition
= the loss of participants from a study..
is bad because if subjects who drop off are significantly different than those who remain (SELECTIVE ATTRITION) then the sample may not be representative of the population
selective attrition = when a certain kind of person drops out
What are closed-ended questions? (surveys+questionnaires)(also called fixed alternative)
-Provide alternatives for the participant
-More control over responses and they are easier to analyze, but the info is not a rich
What are rating scales?
-a variation of closed-ended
-provide a graded response to a question
-Likert scales are part of this - used for measuring attitudes
-provide a series of statements and participants indicate degree of agreement/disagreement
Example:
-Scale 1-10 how sad are you
-how much do you agree with this statement? 1=strongly disagree 5=strongly agree
why is not random sampling a threat to external validity?
if your sample is not random it may not be representative of the population you wish to study.
if your sample is not representative then you cannot generalize to that population.
no generalization = no external validity
What are partially open-ended questions?
-they provide an “other” category so the participant can specify an answer
Example:
If a student is late for class, they should: (circle best response)
a) not enter the classroom
b) enter through back doors only
c) enter through any door
d) other (specify)__________
What are combination questions?
-they have a closed-ended portion and an open-ended portion so the participant can comment on their choice
Example:
A student should arrive to classes on time
1=strongly disagree, 5=strongly agree
Comment:_____________
what are 2 explanations if there is a difference between groups in an experiment? (if there are no confounds)
1) the IV had an effect
2) chance
how do we figure out if our results occurred by chance in an experiment?
inferential stats
if our inferential stats reveal that the results were not due to chance, what will be true for the LEVEL OF SIGNIFICANCE?
the level of significance will be less than .05
(the p-value is less than 0.05)
means that it occurred by chance less than 5% of the time
also
level of significance and p-value are the same thing. (just so you know)
experimental method:
what is the null hypothesis?
states that there is no difference between the two groups. (the IV had no effect [didn’t work / results are due to chance])
experimental method:
what is the alternative hypothesis?
There IS a difference between the two groups
(the IV did have an effect)
when can we reject the null hypothesis?
when our inferential stats reveal that the level of significance is less than 0.05
what is a statistical test
a procedure used to determine statistical significance
what is a t-test
a statistical test used to access the likelihood that differences between two means occurred by chance
“is the difference between my two group means statistically significant or due to random error”
what is the level of significance?
a predetermined probability below which the results of a test must be before they are called statistically significant
typically p < .05
what is error variance?
- variability in the scores caused by variables than other than your IV
(e.g. some subjects will be tired / others won’t) - it could be extraneous or subject-related variables
(extraneous = confound) (subject variable = age, intelligence, ect)
what are the two ways that you can measure your IV?
1) quantitatively
2) qualitatively
what is more powerful, between or within designs?
within designs!
what are some ways of reducing error variance?
x3
1) hold extraneous variables constant by treating the groups exactly the same exact for the manipulation
2) match subjects on characteristics that contribute to error variance
e.g. age
3) use a within-subject design
How to handle error variance you might ask?
3 main things
1) reduce error variance
2) increase the effectiveness of IV by using strong manipulations
3) Randomize error variance across groups with random assignment
main difference in ‘between’ and ‘within’ research designs
between = participants are randomly assigned to conditions so there are different subjects in each group.
within = the same participants are tested in all conditions.
(indicates whether the same participants are tested in all of the conditions or not)
What are things to consider when creating questionnaires?
-include positively and negatively keyed items to avoid response set bias (neutral wording)
-precise questions that only elicit the info you are interested in
-wording should not communicate a point of view
-avoid loaded questions (those with biased wording)
-avoid double negatives (ex. not being on time for class is not acceptable)
-give precise time references (ex. not “how many times did you golf last year?”. Say “in the past 12 months” or “in 2019”
-do not ask double-barrelled questions (asking two questions at once)
What are the four methods of administering a survey?
- Face-to-face interviews
- Telephone interviews
- Mail questionnaires
- Online questionnaires
(definitions on separate slide)
What are face-to-face interviews? List advantages + disadvantages
The questions are asked in the presence of the researcher
Advantages:
-can gauge the respondent’s state
-can judge comprehension and clear up misunderstandings
-can probe for additional info
Disadvantages:
-expensive, so limits the number that can be administered
-interviewer can bias the results
-respondent may not be honest, particularly with sensitive issues
What are telephone interviews? List advantages and disadvantages
Interviews administered over the phone lol
Advantages:
-More cost effective than face-to-face interviews
Disadvantages:
-Respondents may get lost in long questions
-Respondents may be suspicious
-You may anger respondents by “catching them at a bad time”
What are mail questionnaires? List advantages and disadvantages
Questionnaires sent by mail, go figure
Advantages:
-Inexpensive and can be administered to large samples over large areas
-Participants may respond more honestly than face-to-face or on the phone
Disadvantages:
-They may not be returned (percent not returned is nonresponse rate)
-With a very low return (a high nonresponse rate) you may have a biased sample (nonresponse bias)
What are online questionnaires? What are the advantages and disadvantages?
You’ll never get this one - questionnaires administered online
Advantages:
-Inexpensive and can be administered to large samples over large areas
-Participants may respond more honestly than face-to-face or on the phone
Disadvantages:
-They may not be returned leading to the potential of a biased sample
-Nonresponse rate - the percentage of surveys that are not returned
-Nonresponse bias - with a very high nonresponse rate you may have a biased sample
What is social desirability?
The tendency for participants to tell the experimenter, or report on a questionnaire, what they think is socially acceptable or desirable rather than what they truly feel or think
What is probability and non-probability sampling?
Probability: a type of sampling procedure in which one is able to specify the probability that any member of the population will be included in the sample
Non-probability: a type of sampling procedure in which one cannot specify, or does not know, the probability that any member of the population will be included in the sample
What is simple random sampling?
Each member of the population has an equal probability of being included in the sample
What is stratified random sampling?
The population is divided into strata (subgroups) followed by random sampling from each stratum (subgroup).
What is cluster sampling?
Researchers divide a population into smaller groups known as clusters. They then randomly select among these clusters to form a sample.
what are between subject designs?
are also called INDEPENDENT GROUP DESIGN because they are different participants in the group (the groups are independent)
for these designs participants are randomly assigned to the groups
what are within- subject designs?
are also called REPEATED MEASURES because the same group of participants is exposed to all levels of the IV (they are tested repeatedly)
for these designs, a single group of participants are exposed to all levels of the IV
what are some similarities between Between and within subject designs?
1) there is data collected for each group (or condition), which is averaged and analyzed
2) the difference among means are tested statistically to determine the probability that the differences could have arisen by chance
3) if this probability is acceptably low, one concludes that the differences are reliable and attributes these differences to the IV
what are the two kinds of between-subject designs, and their subfields?
1) Single Factor randomized group designs
- randomized two group design
- randomized multi group design
2) Matched Group design
- matched pair design
- matched multi group design
Between Subject Design:
what is randomized two-groups design?
- randomly assign participants to two groups
(two levels of the IV)
Between subject design:
what is randomized multi group design?
- 3 or more levels of the IV
- subjects are randomly assigned to the groups
what is ‘Factor’ synonymous with?
Independant variable
Word of the day!
FACTOR
means the same thing as the IV
they are interchangeable
e.g. “the manipulated factor in this experiment”
What is a Matched group design?
- matched participants are assigned to groups
- used if you want to match groups on a certain variable that is important to the experiment
what is matched pair design?
- only two groups are tested
what is matched multi group design?
multiple groups are tested.
What are the steps for matching a group?
1) get each subjects score on the variable of interest
2) rank order of the scores
3) match the top 2 scores, then the next 2, ect
4) for each matched pair, randomly assign one to each condition
Single factor:
What is are the benefits and disadvantages of a randomized two group design?
advantage:
- the experiment is easy to conduct because you don’t need to take a lot of steps
disadvantage:
-provides limited information
Single factor:
What is are the benefits and disadvantages of a randomized multi group design?
advantage:
- you obtain more information
disadvantage:
- greater number of subjects needed
What are the advantages and disadvantages of a Matched design?
advantages:
- can control subject variables that may obscure the effect of the IV
disadvantages:
- more demanding and time-consuming
What are the advantages and disadvantages of WITHIN subject design?
Advantages:
- all subject-related factors are literally identical across conditions so there are no large individual differences to obscure the effect of the IV
- reduced error variance leads to a more powerful design
Disadvantages:
- more demanding on subjects
- there is the risk of carryover effects
what are carryover effect (aka order effects)?
when a previous treatment alters the behaviour in a subsequent treatment
how you behave in the first condition could affect how you behave in the next
what are the sources of carryover effects?
x6
Leonard Faces Happy Smiles Compared (to) Aggression
- Learning
- Fatigue
- Habituation
- Sensitization
- Contrast
- Adaptation
Carryover effect:
what is LEARNING?
learning to perform a task in the first treatment may effect later treatments
Carryover effect:
what is FATIGUE
May cause performance in later treatments to deteriorate
Carryover effect:
what is HABITUATION?
Repeated exposure to a stimulus can lead to reduced responsiveness
Carryover effect:
what is SENSITIZATION
repeated exposure to a stimulus can lead to increased responsiveness
Carryover effect:
what is CONTRAST?
exposure to one condition may alter responses in later conditions
Carryover effect:
what is ADAPTATION?
subjects may adapt to certain manipulations (e.g. drug tolerance)
How do we fix carryover effects?
counterbalancing
What is counterbalancing?
assigning treatments in a different order for different subjects (participants)
e.g. half would get condition ‘A’ first, the other half would get condition ‘B’ first, then the groups would switch
all possible orders have been covered.
What are the two different kinds of counterbalancing?
1) complete counterbalancing
2) partial counterbalancing
what is complete counterbalancing?
provides EVERY POSSIBLE ORDERING of treatments and assigns at least one subject to each order.
if you were full counterbalancing for 3 conditions, there would be 6 possible orders:
ABC, BCA, CAB, CBA, ACB, BAC
if you were complete counterbalancing for 4 conditions, there would be 24 possible orders…
»> groups would also then needed to be tested in groups of 24
What is partial counterbalancing?
- includes only some of the possible treatment orders
- each treatment appears equally as often in each position
(e.g. treatment ‘A’ appears 5 times in the 1st, 2nd, 3rd, 4th position…)
If there were 4 conditions, then for partial counterbalancing you only need to test in groups of 8
What is a Latin square?
a form of partial counterbalancing that ensures each treatment appears an equal number of times in each position…
the first treatment for one participant will be the last participant for the next.
I think it just refers to the way that its ordered…
looks like this:
Participant 1: A B C D
Participant 2: B C D A
Participant 3: C D A B
Participant 4: D A B C
When do you use a Within subject Design?
- when subject differences contribute heavily to variation in the DV
- when the number of subjects is limited and carryover effects are not an issue
e.g. if you are doing a study on bi-polar and you have limited access to participants.
What are the two types of within subject design?
1) single-factor two-level design
2) single- factor multilevel design
What is chi-square?
the test of the independence between two variables (i.e. it establishes whether variable x is independent of variable y)`
Within Design:
what is single factor two level design?
- only 2 levels of the IV
- All subjects receive both levels of the IV
- half of the participants receive treatment A first, half of the participants receive treatment B first
What type of scales must be used for a chi-square?
nominal scales (numbers have no value)
Within Design:
what is single-factor multilevel design?
a single group of subjects is exposed to three or more levels of a single IV
what is a nominal scale?
the numbers are used to refer to categories (no numerical value)
what is the null hypothesis for a chi-square?
variables x and y are independent
what is the alternative hypothesis for a chi-square?
variables x and y are NOT indpendent
What type of contingency table do we put the data for a chi-square in?
a contingency table called the observed table (the table of our observed values)
see notes for example p 37
After the observed table is made for a chi-square, what table do we make next? What hypothesis do we say is true?
the expected table (expected values) is created as if the null hypothesis was true
see p.37 for example
How many observations in each cell of the expectancy table should there be to conduct a chi-square test?
at least 5
see p.38
If the chi-square value is large (different), what can we say about the null and alternative hypothesis?
reject the null hypothesis and provide support for the alternative hypothesis that says the two variables are not independent
If the chi-square value is large (different), what can we say about the null and alternative hypotheses?
reject the null hypothesis and provide support for the alternative hypothesis that says the two variables are not independent
What can we infer if the expected and observed tables of chi-square are similar?
the smaller the value of the chi-square and the more likely the variables are independent.
What can we infer if the expected and observed tables of the chi-square are different?
the larger the value of the chi-square and the more likely the variables are NOT independent
Why will chi-square likely never be zero?
the observed and expected tables will never be exactly the same; even when the null is true and the two variables are indeed independent
How large does the chi-square value have to be before we reject the null and conclude that the two variables are not independent? (What do we use to determine this?)
by using a critical cutoff value (CV)
How is the critical value determined for chi-square?
alpha (.05)
the degrees of freedom are equal to (# of categories in x - 1) * (# of categories in y - 1)
If the chi-square value is greater than the critical value, what can we conclude about the null and alternative hypotheses?
reject the null, support alternative
X and Y are not independent (the observed values are different than the expected values
If the chi-square value is smaller than the critical value, what can we conclude about the null and alternative hypotheses?
reject the alternative, support null
X and Y are independent (the observed values are similar to the expected values)
How to determine the degrees of freedom?
(# of cells in x - 1) x (# cells in y - 1)
How to determine chi-square value?
x^2 = ∑(O-E)^2/E
What is the correlational method?
a research method that determines whether, and to what extent, two variables are related to one another (if related, will determine the strength and relationship of that relationship)
Are variables manipulated in the correlational method?
no, two variables are measured
Are variables manipulated in the correlational method?
no, two variables are measured
Are correlations experimental?
No they are simply descriptive
What is the correlational coefficient?
A statistical measure of the extent that two variables are related to one another (between -1.0 and +1.0)
What is the most widely used correlation coefficient?
the Pearson product-moment correlation coefficient (r)
What are two ways r can vary?
- The direction of the relationship (can be positive or negative)
- The strength of the relationship (ranges from +1 to -1)
How do you determine the direction of the relationship of the correlation coefficient (r) value?
the sign; negative (-) or positive (+)
How do you determine the strength of the relationship of the correlation coefficient (r) value?
the number of the correlation coefficient
What is a positive correlation?
when two variables increase or decrease together in the same direction
ex: children who watch more violent tv ^ are more aggressive ^
What is a negative correlation?
when one variable increases and the other decreases in the opposite direction
ex: the more optimistic people are ^ the less often they get sick\/
Is global naturalistic observation qualitative or quantitative in nature?
qualitative - observations are words, ideas, concepts, descriptions
is systematic observation qualitative or quantitative in nature?
quantitative - observations reduced to numbers
What is the aim of quantitative research?
reduce behaviour to a quantity, reduce observation to a number
MEASUREMENT
what is the aim of qualitative research?
describe verbally and in detail the behaviour being observed; word, object, description, idea
DESCRIPTION
What type of research (qualitative or quantitative) is typically conducted in later stages of research when previous research has already been conducted? (having a clearer understanding of what is expected to be found)
Quantitative
With what type of research (qualitative or quantitative) do you rarely approach your task with precisely defined hypotheses to be tested? (due to not being sure of what will be uncovered).
Qualitative
What type of data is collected in qualitative and quantitative research?
Quantitative: quantities, numbers
Qualitative: descriptions, qualities, senses
What type of research is typically done to test clearly stated hypotheses?
Quantitative
When is quantitative research typically conducted in research?
In the later stages of research when previous research has already been conducted
Is sample size more important in qualitative or quantitative research? why?
Quantitative; we are more interested in predicting behaviour and testing hypotheses and thus obtaining statistical significance(power); sample size is essential.
- with qualitative research, we are simply interested in describing behaviour in its context, so the sample size isn’t as crucial as obtaining an ‘information-rich’ sample.
- Resulting data from qualitative research is rich and more detailed but is less able to be generalized
Why do we need qualitative research?
we are greatly influenced by the situations that we are in, and these field studies describe in detail the behaviour in those kinds of settings.
Interpretation = qualitative (speculation)
Why is qualitative research more important in the earlier stages of research?
Research is more exploratory in nature when there has not been a lot of previous research conducted in that area
What are the advantages of qualitative research?
- exploratory in nature
- don’t need a defined hypothesis, only would have an idea of what may be uncovered
- don’t need a large sample size
ex: in a jungle observing a tribe the researcher would not be sure of what they will observe
What are the advantages of quantitative research?
- can be generalized
- able to predict behaviour and test hypotheses
- can obtain statistical significance
What is the strength of a correlation coefficient that is at (-.2 or +.2)? What does that tell us about the correlation of the variables?
weak relationship, low correlation
The closer the r value is to 0, the __________ the relationship.
weaker
The closer the r value is to -1.0 or +1.0, the _________ the relationship.
stronger
What is the strength of a correlation coefficient that is at (-.8 or +.8)? What does that tell us about the correlation of the variables?
strong relationship, high correlation
What is a helpful tool to use when determining the strength of a correlation coefficient?
visualize the value on a number line
What is a variable?
a characteristic or quantity that can take on two or more values (i.e. it varies)
what is an operational definition (operationally defining variables)?
defining variables in terms of how they are measured or manipulated
what are three things that a good operational definition should have?
- describe how the variable is measured or manipulated
- be replicable
- use objective, simple, and concrete language.
What is a scatterplot?
a graph in which paired scores for many subjects are plotted as single points to reveal the direction and strength of their correlation
What is the scatterplot used for? What do the dots represent? What can be estimated from the scatterplot?
- used to visualize the relationship between two variables
- each dot on the graph represents an individuals score on two variables
- both the direction and strength of a relationship
The direction of the relationship on a scatterplot can be _______ or _______.
positive or negative
- positive: /
- negative: \
Scatterplot
What is a positive relationship and what can we identify from these relationships between variables?
when two variables tend to increase or decrease together, in the same direction
- higher scores on one variable are associated with higher scores on the other variable
- lower scores on one variable are associated with lower scores on the other variable
Scatterplot
What is a negative relationship and what can we identify from these relationships between variables?
when two variables tend to move in opposite directions
- higher scores on one variable are associated with lower scores on the other variable and vice versa
What can be identified about a scatterplot by how spread out or close together the dots are?
the strength of the relationship between variables
If a scatterplot has dots that are more spread out, what can be said about the relationship between variables?
the more spread out the dots appear, the weaker the correlation
If a scatterplot has dots that are closer together, what can be said about the relationship between variables?
the closer the dots appear, the stronger the correlation
What does a strong correlation mean? (scatterplot)
the two variables are very closely related and that they are good predictors of each other, in other words, the scores for one variable change in proportion to the scores for the other variable
What is a zero correlation? (scatterplot)
when no relationship exists between two variables. A change in one variable is not related to the change in the other variable. The variables are not good predictors of one another
Why use correlations? (4)
- exploratory research
- they can be done when ethics prohibits experiments
- to challenge a theory that says that two variables should not be correlated
- making predictions
what is the predictor variable?
the variable being used to predict
eg. if you know that SAT scores are positively correlated with college GPA, then you can use a student’s SAT score to predict (within limits) the GPA the student is likely to achieve
eg. SAT
what is the criterion variable?
the variable being predicted
eg. if you know that SAT scores are positively correlated with college GPA, then you can use a student’s SAT score to predict (within limits) the GPA the student is likely to achieve
eg. GPA
what is the criterion?
word of the day
in a correlation, the criterion is the variable that is predicted
What is criterion validity?
the goal is to determine how well the measuring instrument can predict some concurrent or future behaviour
eg. predictive and concurrent validity
Where does the predictor variable go on a scatterplot graph?
x-axis
where does the criterion variable go on a scatterplot graph?
y-axis
What are correlations typically predicting?
general statements of probability, not predictions about specific individuals
i.e. you can’t say Suzanne received 98% on SAT she will get a GPA of 3.8 in grad school
CAN ONLY MAKE general statements about probability…
i.e., in general, students that do well on the SAT tend to do well in graduate school
What are some limitations of correlations?
cannot be used to make causation statements, only to describe relationships between variables
Why can’t we make causation statements with correlations? (3)
- x has a causal influence on y (problem of directionality)
- y has a causal influence on x (problem of directionality)
- or some third variable, z, has a causal influence on both x and y. (third variable problem)
What is the problem of directionality?
even if a causal relationship existed between the two variables, the direction of causality may be difficult to determine
ex: a positive correlation exists between level of aggression in children and the amount of violent tv watched.
It could be that watching violent tv causes aggression (x causing y), or it could be that children with aggressive tendencies like to watch violent TV making it the aggression in children causing the violent TV watching (y causing x)
what is the third variable problem?
if to variables are correlated it could be a third variable (z) that is causing the relationship
ex: optimism is correlated with speed of recovery after surgery. Those who tend to be more optimistic take less time to recover. A possible explanation for this could be that a strong family unit (z) = more optimistic = also helps recovery by giving support at the hospital.
Are correlations designed to detect linear or curvilinear relationships? why?
LINEAR only. the correlational coefficient will not indicate the presence of a relationship, even if it is clear to us that there is a relationship.
what is hypothesis testing (correlational coefficients)?
the correlation coefficient is a descriptive statistic; it describes the direction and strength of a relationship between variables.
- used to learn whether the obtained correlation coefficient is large enough to infer that there is a relationship between the variables in the population.
- set up two hypotheses (null and alternative)
- if results indicate a statistically significant relationship,, null can be refuted, alternative supported.
what is a null hypothesis (correlational coefficient)?
the assertion that the correlation between two variables in the population is zero (i.e., that there is no correlation between the two variables)
what is an alternative hypothesis (correlational coefficient)?
the assertion that there is a relationship between the two variables in the population (that is, the value that was obtained in correlation was not by chance)
What are the two factors that determine whether our correlation coefficient is statistically significant?
sample size - the larger the sample, the more likely the results will be statistically significant
strength of the relationship - the closer it is to -1 or +1 the more likely the results will be statistically significant
definition: population (hypothesis testing)
the complete set of events in which you are interested
definition: sample (hypothesis testing)
your actual set of observations, it is a subset of population
definition: random sample (hypothesis testing)
every person in the population has an equal chance of being chosen to be in your sample
Why is random sampling important? (2)
if your sample is not random, it may not be representative of the population you wish to study, and you cannot GENERALIZE to that population
definition: generalizability (hypothesis testing)
ones ability to say something about a population based on the results from their sample
definition: law of large numbers
large samples are more likely to be representative of the population from which they are drawn than are small samples
definition: replication
when an experiment or study can be copied
helps verify that the results are not due to chance
What is the level of significance?
alpha or a
- predetermined probability below which the results are called statistically significant
- if there is less than a 5% chance that the results are due to random error, the experimenter will reject the null hypothesis, providing support for the. alternative hypothesis.
What is a Quasi Experiment?
what is the difference between a ‘true experiment’ and a Quasi experiment
a type of research design that attempts to establish a cause-and-effect relationship.
The main difference between this and a true experiment is that the groups are not randomly assigned.
