Review list of Terms and Questions that  are "Fair Game" for Final Exam


 

In general, approximately 60% will be devoted to various past quiz items (highlighted below) and 40% of the quiz will be devoted to the last 3 chapters covered in the text (Chapters App A, ).  The general structure of the quiz will be of a variety of questions format including some multiple choice, brief description, short essay variety similar to questions posed in the prior exams. 

 

Scientific Understanding of Behavior (Chapter 1)

 

            Why is a Course in Research Methods Important?

 

Methods of acquiring Knowledge

- Several methods

- Intuition, Authority (Aristotle), Rationalism (Story of Horses),

- Empiricism, Scientific Method

 

Types of Questions – Are all topics scientifically testable? If not, why not?

            Criterion of Falsifiability-

            Skepticism –

 

ROT test (Repeatable, Observable, Testable)

            Scientific Method (broadly defined)      

1.      Describe Behavior

2.      Prediction (Hypotheses).

3.      Determine Cause (Test).

4.      Explain (Revise or Retest)

 

Multiple Hypothesis Testing (Alternative Explanations) Type of “Strong Test” vs. A “Weak Test” of a Hypothesis.

 

Where to Start (Chapter 2)

 

            Researchable Questions Come from Many Sources.

Hypotheses vs. Predictions

 

Variables -

Predictions -

Sources of Ideas: Many Exist

1. Common Sense - “I-Knew-It-All-Along Phenomenon” or “Hindsight Bias”

2. Observations of the World Around Us.

3. Theory –

4. Past Research - Rich Source

5. Practical Problems – Drug Use, Sex, Violence,

 

Library Research - Often Best Beginning Place – Why?

 

What are Journals? Why Should One Seek these Sources Out vs. Other Sources?

 

What are Psychological Abstracts? E.g.,

 

PsychINFO

Social Science Citation Index

Reference Librarians


Studying Behavior (Chapter 4)

 

Variables -

Operational Definitions -

 

Measurement of Variables

 

Reliability –

True Scores and Measurement Error -

A Number of Forms of Reliability have Been Developed.

Test-Retest Reliability - 

Alternate Forms Reliability-

Internal Consistency

Forms of Validity:

Validity –

Construct Validity

Convergent Validity -

Reactivity?

Relationships Between Variables: Common Types.

Positive linear Relationship

Negative Linear Relationship

Curvilinear Relationship

 

Correlational Vs. Experimental Methods.

 

Active Manipulation ; Independent Variable vs. the Dependent Variable.

 

Correlational Research is “Non-Manipulative”.

 

2 Problems with Correlational Research :

(1) Direction of Cause & Effect and

(2) Third Variable Problems.

 

Experiments - Our Most Powerful Means to Identify Cause and Effect Relationships. 

 

            A True Experiment Involves Randomization of Subjects to Groups.

Field Experiments –

 

External Validity (Generalization) vs. Internal Validity

 

Measurement Concepts (Chapter 5)

 

            Description –

            Reactivity –

 

            Scales of Measurement – Why necessary? Dictates appropriate statistical test.

                        “NOIR” – Nominal, ordinal, interval, ratio scales or data.

 

                        Coding System –

 

                        Pros and Cons of Video tape –

 

                        Reliability issues & Intra-rater reliability–

 

                        Affect vs. Effect – “to influence” vs. “outcome”.

 

                        Case Study –

                       

                        Archival Research – (Content Analysis) –

                       

                        Sampling Techniques –: Nonprobability Sampling and

                        Probability Sampling.

 

                        Haphazard sampling, Quota Sampling

 

                        Stratified Random Sampling –

 

                        Developmental Research – Longitudinal, Cross-sectional, and Cross-

                                    Sequential methods (designs). 

 

                                    Cohort – Cohort Effect –

 

Observing Behavior (Chapter 6)

 

Coding System –

Case Study & limitations – e.g., generalizability

Archival Study –Selective Deposit and Selective Return

Psychobiography

Naturalististic Observation

Participant Observer Study      

 

Survey Research (Chapter 7)

 

Surveys – Questionnaires or (sometimes Interviews) both strengths and weaknesses

 

Self-report techniques (Survey / Interview)

 

Variety of Survey Types: Mailed, Group-Administered, Household Drop-off

 

Interviews (often face-to-face “surveys”)

 

Response Rate -

 

Closed vs. Open-ended

 

Response Sets

 

Social Desirability –

 

Interviewer Distortion

 

Dichotomous responses or Nominal Response Format

 

Interval Level Response Formats (often referred to as Likert Scaling). 

 

Avoid Restriction of Range

 

Likert Scales (or Likert-type scales) –

 

Filter or Contingency Questions

 

Double-barreled questions

 

Interviewer Bias

 

Remember with surveys (especially): garbage in…garbage out…(i.e., the quality of the data is generally proportional to the time and energy devoted to making the instrument).


Appendix A - Writing Research Reports


Review Basics of APA formating, underlying assumpitions of this style (i.e., what is the writer attempting to do and for whom?).


Basic Parts and characteristics (including function) of the following key parts of a traditional research report:

    Title Page
    Abstract
    Introduction
    Methods
       Particpants
       Apparatus
       Procedure
    Results
    Discussion
    References
    Figures vs. Tables


Hour-glass analogy to a good paper.

Understanding Research Results: Description and Correlation (Chapter 12)


Scales of Measurement (Review) - NOIR

    Nominal
    Ordinal
    Interval
    Ratio

Common Data Analyses:
   
    Comparing Groups Percentages
    Correlations
    Compare Group Means (e.g.., t-tests or ANOVA (Analysis of Variance)


Descriptive Statistics


    Frequency Distribution

    Histograms (Bar Graphs)
    Measures of Central Tendency - Mean, Median, & Mode
    Variability, Standard Deviation (SD), Variance, Range
    Pearson r Correlation (r) - Shows degree of relationship between two conitinuously distributed measures or variables (e.g.., Age and GPA).
       r = variy between -1 to 1; scores near 0 reflect little if any systematic co-relation between the two variables.
    Spearman r (rs) = is a correlation for rank order data, generally similar results to that for Pearson r.

    A Scatterplot is typically used to display graphically the correlation between variables.
      In some cases, a curvilinear relationship may be uncovered.


Restriction of Range - try to avoid
Effect Size - an estimate of the strength of a relationship or finding between two variables, the standard r is one form of "effect size", others exist as well.


Statistics - Divided between Descriptive and Inferential statistics.

Descriptive statistics-
Inferential statistics -

 
Null hypothesis (Ho) vs Research Hypothesis (H1) -


Statistically Significant effect (Statistical Significance) - based on probability that two or more groups differences are unlikely due to chance variability alone but are due to real a systematic (replicable) difference between groups.


Probability -


Alpha Level - The probability required by a statistical procedure and decision to state that the results are unlikely due to simple chance fluctuations.
5 out of 100 times is common (P < 0.05 level; some may be more conservative and use a criterion of stating that something is "significant" when the group differences are unlikely to occur on average less than 1 in 100 times or P < .01 criterion).


Sample size - generally, larger samples improve odds that real, systematic differences between groups will be revealed to be "statistically significant". Invariably, if sample size is too low, then one cannot reject the Null hypothesis and you cannot make a claim that you have numerical evidence the groups differ due to treatment or some other underlying cause. 


t and F tests - t tests are used to determine if two groups differ in means and patterns of scores by more than chance fluctuation.  ANOVA (or Analysis of Variance or "F tests" are used when one compares 3 or more groups together for differences in mean or pattern scores.


df (Degrees of Freedom) - Values that indicate the number of cases that are used to derive significance levels.  In effect they are values that reflect the number of scores that are "free to vary" once the means are determined.,  e.g.., if n = 20, then df = 18 for a t-test (between subjects test) for groups of 10 each, based on the formula N1+N2 -2 = 18.


One tailed vs. Two-tailed Tests - One tailed tests typically are used less frequently than two-tailed tests but may be used when there is a clear reason to predict the specific direction between groups differences, e.g.., Males are expected (based on past study and/or strong supporting evidence) to score higher than do females.  In practice many researcher use a more conservative "two-tailed" test when you do not know or have strong prior evidence/belief that you know which group will score above the other but you are looking for differences between them.



Understanding Research Results: Statistical Inference (Chapter 13) (see terms from chapter 12 above )


Null hypothesis (Ho) vs Research Hypothesis (H1) -


Statistically Significant effect (Statistical Significance) - based on probability that two or more groups differences are unlikely due to chance variability alone but are due to real a systematic (replicable) difference between groups.


Probability -


Alpha Level - The probability required by a statistical procedure and decision to state that the results are unlikely due to simple chance fluctuations.
5 out of 100 times is common (P < 0.05 level; some may be more conservative and use a criterion of stating that something is "significant" when the group differences are unlikely to occur on average less than 1 in 100 times or P < .01 criterion).


Type I vs. Type II Errors - (Obviously, we attempt to not commit these, but there is no absolute way to guard against them entirely).  2 X 2 decision matrix - 2 possible source of errors and two ways to make a correct decision based on data.

Type I - when we reject the null hypothesis, when it is actually true (like a "false positive" response - we claim there is a real effect or difference, when in fact further replication reveals it is not there

Type II - when we fail to detect a true underlying trend or difference, based on data alone, but in fact a real difference or trend exists among your variables (it is analogous to "a miss" or you fail to detect the difference.  A common reason for creating a Type II error is to have an inadequate sample size; often by increasing the sample size (and/or reducing error variability) you will gain sufficient power to detect the underlying difference.  In some cases, simple chance factors caused a Type II error to emerge.

Note that our likelihood of commiting a Type I error decreases as we increase the likelihood of committing a Type II error (they are inversely related) - this could occur if we set the alpha level say from .05 to .01.  Note, by doing so you in effect demand that a difference is greater (or a trend is stronger) than is necessary to declare it significant at the 0.05 level.  On the other hand, by failing to reject the null hypothesis, you increase your chances of failing to detect (or make the decision) that you have a real underlying difference in your data.