-
Social Desirability
How to prevent this in the future.
A socially desirable response which has nothing to do with the question.
Prevention: Control privacy & confidentiality
-
Simplicity wording
Avoid jargon terms and technical terms; define complex terms before asking about them.
-
Negative wording
- Avoid negative wording.
- Ex: Do u think that they should not.....
-
-
Questionnaires & Interviews are part of _________ research.
Pros & Cons
- Surveys Pros: Easy, cheap, fast, Lg groups
- Survey Cons: Hard 2 identify age/gender & poor response rate.
-
Double barreled Question
- Two questions with one response.
- Ex: should you forgive a friends gay rights or life choices?
-
Loaded Question
A question that might have strong wording and forces people to answer in only one way.
Ex: do you beat ur kids?
-
Closed Ended Question
- Fixed number of responses
- Rating scales are often used 5-7 points
- (strongly agree to strongly disagree)
-
Open Ended Questions
Free response
-
Interviews can be done.......
Pros & Cons
face to face, on telephone, or focus groups.
- Pros: Unexpected / Unthinkable answers
- Cons: Lose anonymity & Interviewer Bias
-
Questionnaires:
Pros / Cons
Pros: Can mail, use on internet, not very costly, respondents = anonymous.
-
It is often impossible to include all of the population that's why a ________ is used to represent it.
sample
-
How to measure confidence interval.
-
What is probability sampling?
All population is included and it is rare & hard to conduct research this way.
Ex; Simple random sampling: in this way every member of the population has an equal chance to be selected.
Often times hard to include everyone so that's why non probability sampling is more commonly used.
-
Even if random sampling is used, what can be a two ways for bias to occur?
- The sampling frame used (ex white pages)
- Poor response rates.
-
Non probability sampling:
Sampling in which one cannot specify the probability that any member of the population will be included in the sample.
-
Internal Validity:
Certainty that IV manipulation caused the changes in the DV. AND NOT confounding variables
-
External Validity:
Ability 2 generalize results
-
What is confidence interval?
range that tells you how accurate you think you are.
-
Confounding
variable =
a variable that is not controlled in a research investigation.
- e.g.,
- If I had married couples complete a difficult puzzle in either a hot room (experimental cond.) or a normal temperature room (control cond.), but did not
- control the size of the room,
- confounding may occur
-
Basic Simple Experimental design
2 Variables ( IV & DV)
& IV only has two levels
-
Post-test - Only Design =
a true experimental design in which the DV (post-test) is measured only once, after manipulation of the IV.
- e.g.,
- an experimental group that receives the treatment and a control group that does not
- e.g., use two different amounts of the IV
- or
-
Pretest-Posttest Design =
Disadvantage?
a true experimental design in which the DV is measured both before (pretest) and after (posttest) the manipulation of the IV.
Disadvantage:
*Disadvantages: time consuming & pretests can sensitize participants to what you are studying
-
Solomon four-group design
combines both the posttest-only and the pretest- posttest design. (This lets you assess the impact of the pretest.)
-
Three ways of assigning participants to experimental conditions:
- Independent Groups Design (AKA between subjects deign)
- Repeated Measures Design (AKA within subjects design)
- Matched Pairs Design (match participants 1st then randomly assign)
-
Independent Groups Design (AKA between subjects deign)
Repeated Measures Design (AKA within subjects design)
Matched Pairs Design (match participants 1st then randomly assign)
- Independent Groups Design (AKA between subjects deign)
- -Researchers randomly assign participants to ONE level of IV
- Repeated Measures Design (AKA within subjects design)
- -Researchers assign participants to ALL levels of the IV.
- ADV: Fewer research participants needed because running twice.
- DISadv: Practice effect, fatigue effect, or carryover effect
- Matched Pairs Design (match participants 1st then randomly assign)
- -Ex: B4 experiment give IQ test - List in order - pair accordingly with one in exp group and one in control group.
-
To deal with problems, researchers can employ ____________ ___________ or increase the time interval between treatments.
Counterbalancing techniques
-
Setting the stage involves:
- Informed consent
- Explain why experiment is being conducted.
- Must introduce IV & DV within context of the experimental setting.
-
Straightforward manipulation:
Staged Manipulations:
- Straightforward manipulation:
- Variables are manipulated with instructions & stimulus presentations
- (Most manipulations of IVS in research are like this)
- Staged Manipulations:
- necessary to stage events in order to manipulate the IV successfully
-
- *They often employ a confederate
-
Best way to get the strength of the manipulation?
Past research while keeping ethics in mind.
-
DV in most experiments is one of three general types:
- Self Report Measures
- Behavioral Measures
- Physiological Measures
-
1) Self-Report Measures
Advantages/Disadvantages
Ppl provide info about their own attitudes, beliefs, and behaviors
Advantages: you get info it straight from them.
Disadvantages: They may lie, social desirability (make themselves appear in + light)
-
Behavioral Measures
Advantages / Disadvantages
measures in which direct observation of behavior occur
Advantages:
- Disadvantages: May not always catch
- everything. Need to use IOA reliabilty.
-
3) Physiological Measures
… measures of people’s body responses (e.g.,heart rate, temperature, muscle tension)
Advantages:
Disadvantages:
-
*The DV should be __________
enough to detect differences between
groups
Sensitive
- Ex:
- Men's measurement of how attractive they view their wife. Watch 30 min pornography,
- and control group does nothing, then ask the men after video if wife is
- attractive.
Is this sensitive enough? No!
-
Ceiling Effect =
- Ceiling effect = failure of a measure to detect a
- difference because it was too easy
-
Floor Effect =
failure of a measure to detect a difference because it was too hard
So in depth that there is no correlation
-
Demand Characteristics =
How can we eliminate demand characteristics?
- cues
- that inform the subject how he or she is expected to behave
- Ex: asking them 50x after taking a
- caffeine pill…… if they are nervous.
- We want to eliminate demand
- characteristics, you can use deception to control for this, use filler
- questions, use a control group a group that doesn't get the treatment.
-
Expectancy effects =
any intentional or unintentional influence that the experimenter exerts on subjects to confirm the hypothesis under investigation
-
Blind & Double Blind Study
Blind Study: Participants do not know which group they are in
Double Blind Study: The experimenter also do not know which group participants are in.
-
To control for expectancy effects, researchers can:
- -
- train extensively & practice behaving consistently
- - run all conditions simultaneously or
- make the procedures automated
- - make the experiment double blind (hire
- uninformed people)
-
Pilot studies:
Purpose?
SiZe?
- You
- run your study doing a trial run with a smaller group.
- Point
- is to see if there is any flaws in the study and then can fix them before runing
- actual experiment.
- Ask
- questions after trial run to see if they knew what your hypothesis was….. If
- so…. U messed up somewhere.
-
What are factorial designs?
Designs with more than one IV
-
5x5 factorial design
How many:
Groups
Levels
IVS
Main Effects
- Groups: 25 groups
- Levels: 10 Levels
- IVS: 2 IVS
- Main Effects: 5 Main effects
-
Factorial design yields two things:
- Simple main effects: effects of 1 IV at a particular level of another IV.
- Main effects = Effects of IV by itself.5IV = 5main efects
-
-
Factorial designs yield two kinds of information
- The main effect of an IV
- Whether there is an interaction between variables
-
You run a nova to if the information is ______ ________.
This measures if it was just a fluke or is it significant.
statistically significant
-
What experimental design is run to see the effect on ONE participant.
Also has no control group.
Reversal Design: Treatment is introduced after a baseline (control) period. ABA
- A = Baseline B= Treatment
- Baseline acts as control group / Comparison
Single case design is when there is no control group present.
-
Multiple Baseline design
Why might this be used?
Observing behavior before & after a manipulation under multiple circumstances.
Used in certain situations, it may be unethical to remove a treatment in order to test its effect
Ex: instead of using 1 person to test drug. Use 3 people to see if yield same results. Gets rid of extraneous variables and flukes.
- P1 Drug Day: 10
- P2 Drug Day: 20
- P3 Drug Day: 30
-
Are Quasi experiments true experiments?
What are quasi Experiments & why are they not true experiments?
No
Quasi-experimental designs allow us to study the effect of an IV on a DV, but lack the control features of true experimental designs
-
One-Group Post test-Only Design
One-Group Posttest-Only Design (Partici -> IV -> DV)
- Ex:
- park bench, ur accomplice sits next to stranger and you time how long until they get up &
- leave.
- ^Problem:
Problems? … no control group to aid in interpreting the results.
-
What do you need for an experiment to be a true experiment?
- Comparison group
- &
- Random assignment!
-
One-Group Pretest-Posttest Design
(participants -> DV -> IV (treatment) -> DV)
Problem:
Ex?
- Problem:
- no control group to rule out history effects, maturation effects, instrument decay,etc.
- EX:
- Program: Trying to help ppl quit smoking
- Pretest measures how much they smoke (self report)
- Treatment Given (program)
- Post test measures again. (self report)
-
History Effects:
Ex:
- any event that occurs between 1st
- & 2nd measure & Is not part of the manipulation
- If Obama diagnosed with stage 4 terminal lung cancer between measures it may
- effect data.
-
Maturation Effect:
Ex:
Maturation Effect: Change due to time and not the independent variable.
- Ex:
- Jealousy (longitudinal study) comparing year 1 to year 10. IT can be naturally occurring.
-
Instrumental Decay:
Your measurement doesn't measure accurately over time.
-
Nonequivalent control group design
Problem?
Ex:
Like a true experiment ( has control group) BUT no random assignment is used.
Problem: selection differences may affect the results.
- Ex:
- smoking ex: Get 100 volunteers à
- creates bias -> two groups but not = groups
-
Nonequivalent Control Group Pretest-Posttest Design
How is it similar to true experiment?
*This is one of the most useful quasi-experimental designs.
NO RANDOM ASSIGNMENT but... pretest scores act as a comparison group. Can compare with post test.
- *No
- control group but because of pretest you do have a comparison group.
-
Cross-sectional design:
Usage?
Pros:
Problem?
cross-sectional designs, compare people at different stages or ages in a developmental process)
- Pros: yields immediate results
- Problem: May cause cohort effects & confound with age differences.
-
Longitudinal Designs
Pros:
Cons:
With longitudinal designs, participants of the same age are followed with repeated measurements over a period of time.
- Pros: yields valuable info
- Cons: expensive, many years, & ppl dropping out of study (participant attriation)
-
participant attrition
Participants dropping out of the study
-
Retrospective designs
Problem:
This design relies on participants recall of past events.
Peoples memories can be inaccurate (bias, forgetting)
-
Sequential Designs
Cross sectional & longitudinal methods are combined.
- Compare two sections of population & then when they are older they are measured again & again & again and see how they
- change over time.
-
When analyzing results when is it best to compare group percentages?
Ex:
If I’m studying whether males and females differ in their interest in sex, and ask males and females whether they like or dislike (yes or no) sex (a nominal
variable), I can compare group percentages
When studying NOMINAL VARIABLES.
-
When is it useful to correlate individual scores?
Ex:
when individuals are measured on 2 variables that have a range of numerical values
ex:
- e.g.,
- If I’m studying location in classroom
- and grades in the class, I can determine
- if there’s a relationship between the two
- variables.
- or
- Ex: height & income level
-
When is it useful to compare group means?
Ex:
Useful when studying participants in 2 or more groups.
Ex:
- If I’m studying the effect of exposure to
- violent video games, I can compare the average # of aggressive acts demonstrated in my experimental and control conditions after the manipulation
-
What is frequency distribution?
Ex:
Frequency Distribution = an arrangement of a set of scores that indicates the number of times each score was obtained
Ex: grades (ABCDF) It will show us how many of each was earned in the class can be charted using graphs, pie charts, and more.
-
What are descriptive statistics?
- Descriptive Statistics = statistical measures that describe the results of a study; descriptive
- statistics include measures of central tendency(mean, median, mode), variability, and correlation.
-
Central Tendency Statistic:
Tells us?
What are they?
Central Tendency Statistic tells us what the sample as a whole is like.
Mean, Median, Mode.
-
Variability
What is it?
What are the 2 kinds of variability?
Is a number that characterizes the amount of spread in a distribution of scores.
- 2 kinds:
- Standard Deviation (avg dev. from mean) & Range(Highest - lowest)
-
The SD is small when....
most ppl have similar scores close to the mean.
-
SD is large when most ppl......
have scores that lie further from the mean.
-
Pearson product-moment correlation
coefficient
Represented by?
Tells us?
Can range from?
How to read the scale.
- Represented by: r
- Tells us: how strongly 2 variables r related & direction of relationship.
Can range from: 0 -- +1 & 0 -- -1
- A correlation of 0.00 means that the two variables are not related
- When the correlation coefficient is positive, that means that there is a positive linear relationship
- - When the correlation coefficient is
- negative, that means that there is a negative linear relationship
-
-
How to determine (from number) the strength of Pearson correlation coefficient.
r = +.70 or Very strong positive relationship
r = +.40 - +.69 Strong Positive Relationship
r = +.30 - +.39 Moderate positive Relationship
r = +.20 - +.29 Weak Positive Relationship
r = +.01 - +.19 No / Little Relationship
-
What are inferential statistics?
Ex of inferential statistics
- statistics designed to determine whether results based on sample data are generalizable
- to a population.
Ex: t-test & F tests
-
Statistical inference begins with.......
Two goals of an experiment?
- statement of the null hypothesis and a research hypothesis.
- -
- The Null Hypothesis states that the observed difference in group means is caused by chance, and not the IV (that the IV had no effect)
- The Research Hypothesis states that the observed difference in group means is due to the IV (that the IV did have an effect)
*The goal of an experiment is to be able to reject the null and accept the research hypothesis!
-
When is a result considered statistically significant?
What is it for psych?
What is alpha level?
If there is a low probability that the difference between the obtained sample means was due to random error.
- Psych .05
- Alpha level: probability required for significance.
-
What are t tests and F tests
When are they used?
What do they determine?
t test = a statistical significance test used to compare differences between 2 group means
- … the t value is a ratio of two aspects of the
- data, the difference between the group
- means & the variability within groups
- Will tell u if difference is statistically difference or not.
- F test (AKA analysis of variance or ANOVA) =
- a statistical significance test for determining whether two or more means
- are
- significantly different
… F is the ratio of systematic variance to error variance
-
What are Type I and Type II errors?
Type I error = an incorrect decision to reject the null hypothesis when it is true
- The wrong decision of rejecting the null.
- (ex: u are saying drug works when in reality it doesn't
Determined by Alpha Level (.05)
- *The probability of making a Type I error is
- determined by the alpha level chosen
- e.g., If the alpha level chosen is .05 and the
- null hypothesis is rejected,
- there are 5 chances out of 100 that the
- decision is wrong
Type II error = an incorrect decision to accept the null hypothesis when it is false
- (ex: that the drug did not work but the
- drug actually did work)
- *Making a Type II error is related to 3
- factors:
- -
- alpha level ( if alpha level is set too low you
- become more likely to make type II eror)
- - sample size (if
- the sample size is too low )
- - effect size (strength
- of relation ship between variables – Ex: half pill – future research full pill)
- Type II error is worse because if a drug does actually work and we indicate it does
- not work it wont get published.
-
Non significant results can occur for a number of reasons:
Non significant results do not necessarily indicate that the null hypothesis is correct.
- 1) A true relationship between the variables
- does not exist
2) A true relationship exists, but wasn’t detected because:
- - the study was poorly designed
- (incomprehensible instructions, a weak manipulation of the IV, a dependent measure that is unreliable)
- the alpha level was set too low
- the sample size was too small
- - the relationship is weak so the effect size
- (difference between groups) is too small
-
Convenience Sampling:
Representative samples:
******************************
An incorrect representation of the population affects the external validity.
-
Volunteer & Gender & Locale Bias
- *Volunteers:
- are more highly educated, wealthier, more in need of approval, and more social than nonvolunteers
- *Males and females sometimes respond
- differently too; therefore, it’s a good idea to include a proportionate ratio of males to females
*Locale bias can affect external validity too
- There should be cultural considerations as
- well
-
Two types of Replication
- 1) Exact Replication – replication of research
- using the precise procedures of a study to see whether the same results are obtained
- 2) Conceptual Replication –
- replication of research using different procedures for manipulating or measuring the
- variables
-
Possible reasons for repeated failures:
-the original results reflect a Type I error
-the original researcher misrepresented the results
- -the original results could only be obtained
- under certain limited circumstances
- -the characteristics of the original
- experimenter biased the results
|
|