Chapter 2: Research Methodology Flashcards Preview

CLPS0010 > Chapter 2: Research Methodology > Flashcards

Flashcards in Chapter 2: Research Methodology Deck (51):
1

Nature of Science

-combine logic with research and experimentation
-involve common sense and observation while recognizing the inherent limitations of both

2

Limits of Observation

Senses can be Fooled
-way brain organizes and interprets sensory info varies from person to person
Generalization
-ppl tend to generalize from their observations and assume that it they witnessed it in one situation it applies to all similar ones

3

What is Science?

-an attitude and process, not a product
-question authority
-open skepticism
-intellectual honesty

4

Goals of Science

1) Describe
2) Predict
3) Control
4) Explain

5

Scientific Method

-systematic and dynamic procedure of observing and measuring phenomena
OPTIC
-observe
-predict
-test
-interpret
-communicate

6

Theories

-set of related assumptions from which testable predictions can be made
-can be used to explain observations or make predictions
-constructed after hypotheses are made and tested
*specific, informed, and testable!!

Good Theory:
-falsifiable (need to define what should or should not observe if it is true)
-testable hypothesis
-SIMPLE

7

Replication

-science has no value without replication
-possible that a single finding can be caused by chance
-science is cumulative (building on past knowledge)

8

Pseudoscience

Claims presented that lack several qualities of science
-advances no true knowledge
-disregards facts
-logic is unsound or absent
-vague or incomplete explanations and conclusions

9

Research Ethics
Humans

-has to meet a set of criteria by panel of experts (IRB) before can be conducted

1. Informed Consent
-minimized costs, maximized benefits
-can stop at anytime with no penalty
-deception must be justified to IRB
2. Respect for Persons
-autonomy of ind must be protected
-risk/benefit ration must be protected
3. Debriefing
4. Privacy and Confidentiality
5. Justice
-benefits and costs must be equally distributed across participants

10

Ethical Research with Animals

-entitled to be treated in a humane way to greatest extent possible
-pain and suffering must be minimized
-euthanasia when req must be done painlessly as possible

11

Research Steps
Scientific Method

1-form hypothesis
2-conduct literature review
3-design a study
4-conduct study
5-analyze data
6-report results

12

Belmont Report

-no one can force participation
-have right to discontinue study at any time
-legally and ethically entitled to know what you are agreeing to so that you can make informed decision

13

Stanford Prison Study

-subjects were not fully notified extents of cost of study ( informed consent)

14

Reliability of Research Results

Consistency
-same answer each time you ask question?
-temporal stability
Degree of Agreement bw Investigators
-same answer when dif ppl as same question?

15

External Validity

-extent to which data collected actually tells you something about phenomena you are measuring
-depends on dependent variable

16

Naturalistic Observations

-as natural as possible so as not to induce certain behavior (Hawthorne Effect)
-can use cover story so subjects unaware of true nature of study

17

Blind Conditions

-participants not told if they are given test or control conditions

18

Answering Questions

-ppl may not answer sensitive questions honestly bc want to look good in presence of researcher

19

Observer Bias

-easy to see what we want to see
-perceive data in a certain light

20

Observer Bias
Rosenthal

-"smart mice" vs "dull mice"
-teacher bias: students who's teachers thought they has a high IQ did better in school than those whose teachers thought had an average IQ (regardless of actual IQ)

21

Observer Bias
Double Blind

-researcher administering task does not know which group subject is in or the diagnosis (minimize bias)

22

Experimental Control

-random assignment
-do not want self selection--need to make sure difference between groups is only due to manipulated variable (not extraneous)

23

Representative Samples

-craw valid conclusions about whole population of interest (all dif types of ppl in that pop)

24

Types of Research
Descriptive Studies

-researcher defines a problem and variable of interest but makes no prediction and does not control or manipulate anything
-no testing of hypothesis
-exploratory

Advantages:
-early stages of research to determine whether a phenom. exists
-takes place in real world setting (external validity)
Disadvantages:
-errors in observation due to expectations
-observer's presence can effect behavior being witnessed (reactivity)

25

Types of Research
Descriptive Studies
Naturalistic Observation

-allowing subjects behave as normally would if not being observed
-external validity
-but conditions can't be controlled so can't determine cause and effect of dif variables

26

Types of Research
Descriptive Studies
Participant Observations

-interviews, surveys
-can get biased responses
ex: kinsey did not use representative sampling (oversampled his home state) and interviewed face to face about private issues which may have resulted in subjects being dishonesty
-self report bias: ppl tend to describe themselves in positive ways not necessarily true

27

Types of Research
Longitudinal Studies

Advantages:
-provide info about effects of age on same ppl (allow to see developmental stages)
Disadvantages:
-expensive
-long time
-high drop out rate

28

Types of Research
Cross Sectional Studies

Adv:
-faster than longitudinal
-less expensive
Disadv:
-unidentifiable variables may be involved (cohort effect)

29

Types of Research
Correlational Studies

-measure two or more variables and examine relationship between them
-can't determine cause and effect

30

Correlational Studies
Scatterplot

-spread away from line of best fit gives idea as to variability
-correlational coefficient is numerical representation of relationship (range from 1 to -1)
-1 is perfect direct relationship
- (-1) is a perfect inverse relationship
-outliers can give misleading coefficient
-if there is a restricted range of values there may be an association but the measure isn't sensitive enough in order to pick up the correlation

31

Correlational Studies
Third Variable

-can cause both A and B (A doesn't cause B)

Matched Samples
-match 3rd variables across groups
Matched Pairs
-match 3rd variables across people (1 to 1 mapping of variable)

32

Experimental Studies

Confounding Variables
-additional variables whose influences cannot be separated from the independent variable of interest

Advantages
-can demonstrate causal relationships (avoid directionality problem)
Disadvantages
-often take place in artificial setting

33

Experimental Studies
Random Assignment

-each part. in study has same chance of being in experimental or control group
-equalize in terms of other variables

34

Experimental Studies
Self Selection

-problem that occurs when anything about a person determines whether he or she will be included in experimental or control group

35

Experimental Studies
Single Blind

-participant does not know condition to which they have been assigned to avoid behaving in a biased way

36

Experimental Studies
Double Blind

-neither participant or experimenter knows which condition to which the participant has been assigned to avoid bias on both sides
-prevents experimenter expectancy effects
-Robert Rosenthal and discovery of experimenter bias (self fulfilling prophecy)

37

Experimental Studies
Quasi Experimental Design

-makes uses of naturally occurring groups rather than randomly assigning subjects to groups

38

Meta Analysis

-quantitative method for combining all published research results on one question and drawing a conclusions

Requires use of effect sizes
-standardized statistics
-allows different studies to be compared to each other even if they used different methods or produced different stats

-look at average effect size across all studies
-have advantage of huge sample size
-there is a publishing bias for positive effect (don't get published if null effects)

39

Measurement

Operational Definition:
-must define property we wish to measure
Good Measurements:
1) Internal Validity
-does data collected address your question?
2) Reliability
-is there consistency of your measurement? or does data collected vary over time?
3) Power
-is the measure sensitive enough to detect a difference if it exists or not to detect a difference that does not exist?
4) Accuracy
-is the measure error-free?

40

Commonly Used Measures
Self Report

Interviews
-responses are coded into broad categories or summarized in a meaningful way
-questions often open ended

Questionnaires
-responses limited to choices given

Adv:
-easy to code, use, and analyze
Disadv:
-self report bias
-social desirability issues

41

Behavioral Measures

-based on systematic observation of people's actions, either in their normal environment or lab setting
-social desirability and observer bias are reduced with behavioral measures
-researchers quantify perceptual or cognitive processes in response to a specific stimulus

Three Major Types:
-rxn time
-response accuracy
-stimulus judgments

42

Physiological Measures

-brain imaging technologies
-blood pressure
-sweating etc
-all these methods are correlational

43

Statistics

-collection, analysis, interpretation, and presentation of numerical data

44

Descriptive Statistics

-summarizes data
-organizes data
ex. plot a frequency distribution

Ideal Experiment
-zero variance and zero overlap
-not realistic
Next Best Thing
-zero overlap (variability exists)
More Likely
-some overlap between groups on performance measures
-need statistics to determine how much overlap does there have to be to determine that the independent variable had no effect on dependent

45

Descriptive Stats
Sources of Variation

-experimental measurement (within and across observers)
-participant behavior (within and between subject variation)

46

Descriptive Stats
Measures of Central Tendency

Mean
-arithmetic average of data set
-most affected by the skew (mode stays where it is and median is less effected)
Median
-score that separates the upper half from the lower half in a data set
Mode
-most frequently occurring score

Normal Distribution
-all three measures of central tendency are the same

47

Descriptive Stats
Variability

-frequency
-range

Standard deviation
-measure of how much scores in a data set vary around the mean
-two different data sets with identical means could have very different SD

48

Different Types of Distributions
on a Frequency Distribution Graph

Normal
-all three measures of central tendency are the same
-perfectly symmetrical bell curve in which scores cluster around the middle
Positive Skew
-mode is top left
-median is down and to the right
-mean is more down and to the right
Negative Skew
-mode is top right
-median down and to the left
-mean more down and to the left

49

Z Score

-tells you how many standard deviations above or below the score is from the mean

z= (y-Yavg)/SD

50

Inferential Statistics

-related to whether or not you can reject or accept the null hypothesis of your experimental design
-results deemed unlikely to be due to chance are called statistically significant

51

t-test

-compares two means to see if the two samples come from the same population