Psych 2820 Midterm 1
Psychological Research: Main interest is in human and animal behavior and
Six Stages of Psychological Research:
(1)Generate Research Idea (One-page summary)
(2)Review Literature (Annotated bibliography)
What has been done? What does future research need to address?
(3)Frame you Study (Introduction)
What’s the problem/topic of interest? What have past studies done? What is
missing? Present rational for study, state research question and state
(4)Plan you Methods (Methods)
How will you conduct the study? Who are the participants? How will you
measure the variables? List procedures, What statistical analysis will you
conduct? Does the study cause harm or risks to participants? Voluntary
participation and freedom to withdraw, method or obtaining informed
consent and participant confidentiality.
(5)Data Collection and Analysis (Results)
Avoid Selection biases, individual or group testing environment? How will
you evaluate collected data? Present statistical analyses and visual displays
(tables and figures)
(6)Discuss and Interpret Findings (Final Project)
State whether results support or refute study hypothesis (link to intro),
discuss implication of results (relate to lit findings), id limitations/problems
with study (how to improve) and suggest directions for further research.
Continuous Variable: Infinite number of possible values between adjacent scale
values (height, weight, number grade, time). Something that is continuous can be
divided into discrete categories for the sake of ease (time to the nearest second).
Discrete Variable: Has a countable number of possible values, no possible
intermediate values (letter grades, number of people).
Dichotomous Variable: Discrete variable with only 2 possibilities (yes or no).
Population: Some large (entire) set of numbers
Sample: Some subset of a given population (estimation of the population)
Parameter: used to describe various properties of a population
Statistic: used to describe various properties of a sample (estimate of parameter) Types of Statistics:
(1)Descriptive Statistics: used to classify and summarize numerical data so it
can be communicated and interpreted. Could not describe, could not
(2)Inferential Statistics: used to make generalizations (inferences) about a
population based on studying the sample.
Scales of Measurement:
(1)Nominal: Place observations into one K mutually exclusive categories,
differences between categories are qualitative and not quantitative, no
ordering to categories, can’t do math on the numbers, tells about frequency
of observations (Count Data).
(2)Ordinal: Mutually exclusive categories, categories are rank ordered, each
one can be compared to the others in terms of less than or greater than, no
indication of magnitude of these differences and can’t do math on these
numbers (sum, mean). Ex: Likert Scale
(3)Interval: Numbers are in logical order, equal intervals between numbers, no
true zero point (arbitrary), most psychological data is interval data
(4)Ratio: Interval scale with true zero point, zero means absence of what is
measured (number of errors), zero point must exist theoretically but may not
exist in reality (weight of zero), can’t consist of negative numbers since zero
is as low as it gets.
Interval/Ratio: Difficult to tell the difference between the two, so usually just refer
to something as I/R. Reliability and validity increase going up the scale of
Bar Graph: Vertical or horizontal bars showing how large each value is.
Box Plot: The ends mark the minimum and maximum values, the first and third
quartile are the bottom and the top of the box respectively and the median is the
line in the middle of the box.
Stem/Leaf Plot: Each data value is split into the leaf (usually last digit) and a stem
(the other digits).
Mode: “Most”, the most frequently occurring value.
Median: “Middle”, the value in the middle of the data set, when values are arranged
lowest to highest, use for ordinal, interval or ratio data, unaffected by extreme
Mean: “Average”, sum of scores, divided by the number of values, the balance point
of a distribution, use for interval or ratio data. Generally a better estimate of the
population mean than mode or median. Value may not actually exist in data, the a normal distribution the mean, median and mode are equal. The mean isn’t good
with outliers or skewed data.
Measures of Variability:
(1)Range: Measure of distance, difference between the highest and lowest
value. Can be used for ordinal and interval/ratio scale. A single value, the
boxplot displays the range.
(2)Interquartile Range: IQR= 3 Quartile – 1 Quartile. The range of 50% of
the observations. Gets rid of the upper 25% and the lower 25% of the
st th nd
distribution. No extreme scores. 1 quartile= 25 percentile, 2 quartile=
50 percentile (median) and 3 quartile= 75 percentile. To find the 1st
quartile, location= (# of ns below median +1)/2= ___th measurement (count
from lowest value). The find the 3 quartile, location= (# of ns above the
median +1)/2=___nth measurement (count from highest value). To find the
2 quartile (median equation).
(3)Average Deviation: If we want to measure how scores deviate from the
mean x- x. To summarize these deviation we would find the average
deviation ∑(x- x)/n (which equals 0)
(4)Sample Variance: Sum of the squared deviations about the mean divided my
N-1. Deviations= ∑(x- x) then divide by N-1 and you get s 2 ∑(x- x)2/N-1
(conceptual formula). The computational formula is
2 å x - N
s = N-1
N-1 (degrees of freedom) allows for an unbiased (average
sample of estimates is equal to the average population value) estimate of
(5)Standard Deviation: Square root of the variance
The larger the standard deviation the more spread the
* Variance and standard deviation are most common measures or variability we use
in this class, if data isn’t mound shaped and asymmetrical (Normal), we would use
the median and IQR. Variance and standard deviation are only for I/R.
Normal Distribution: Any variable that is the sum or average of many small
independent effects will have a distribution that’s bell-shaped. Symmetric, unimodel
distribution. Describes how random variables cluster around the mean. Common
natural occurrence, most statistical test assumes normal distribution. Can be used to
calculate probability and to make a number of inferences.
Z-Scores: “Standard Scores”, number of deviations above or below the mean, tell
you how far away from the mean of the distribution a value is. Z-score for a sample
x - x
z = z = x -m
is: s and for a population s . A value equal to the mean has a z score
of 0, if the value is close to the mean it will have a small positive or small negative
value, if the value is far from the mean it will be a large positive or negative value.
* We can compare the participants’ individual scores relative to the mean and can
use quantitative descriptions of how likely we are to observe a certain value, by
using Z table.
Percentile: Measurement value below which p% of the measures fall.
Types of Psychology Sources:
(1)Theoretical: Researcher describes a theory or conceptual framework of a
construct or process.
(2)Systematic Review/Meta-analysis: Researcher systematically (follows pre-
established procedure) reviews previous studies in the area.
(3)Empirical: Researcher asks a question and present data to answer the
(4)Scholarly: Academic researchers who report results of own experiments
(articles from PsychInfo)
(5)Popular: Journalist reporting on research findings (Time magazine).
Pseudoscience: prone to logical fallacies.
(1)Emotionally Loaded: Appeals to reader’s emotions without using logic.
(2)Bandwagon Fallacy: Because everyone else thinks a certain way, the reader
(3)Faulty Cause-Effect: No real causal relation
(4)Either/Or: Presents only two alternatives
(5)Hasty Generalization: Develops conclusion based on individual cases
Structure of an Article:
(1)Abstract: Overview of research, give you sense of the study but does not give
you great details. (2)Introduction: Identifying the issues, tells you the importance of the
topic/theory, introduces you to previous research and states the study
(3)Methods: What was done, tells you about who the participants are, what was
used to measure study variables and how this was done.
(4)Results: What happened, technical part of the study that tells you what was
the outcome of study.
(5)Discussion: What does it mean, authors explain and interpret the results.
Links back to the introduction.
Operational Definitions: Variable must be linked to observable events, tells the
researcher how to measure that event, define you variables and are subject to
research scrutiny. Control over your study by understanding exactly what you are
measuring. To construct, begin with conceptual idea of what you would like to
measure (construct) and define it in terms of how it would be measured. Objectively
measurable and quantifiable (frequency, duration).
Types of Questionnaires:
(1)Open Ended: Respondent to answer in own words, elicit range of responses,
form close-ended questions from a content analysis of an open-ended
questionnaires. Subject to experimenter bias and require more effort.
(2)Close Ended: Preselected set of alternative responses and respondent
chooses one. Easier to score, more objective and reliable, less effort. Yield
limited information not suitable for identifying new variables that might be
relevant to study.
(1)Construct Validity: degree to which measurement reflects hypothetical
construct of interest. Establishment of construct validity involves “validation”
studies. Will be high if others used it as well.
(2)Criterion-related validity: measure correlates highly with some other
Reliability: The degree to which measurements are consistent, consistency and
repeatability of measurements.
Inter-item Reliability: All questions must measure the same thing.
* Higher correlation means less error variance, and less regression to the mean.
Regression to the mean is higher at time 2.
Reification of a construct: Referring to an unobservable (construct) as a noun.
Evaluating a construct to the level of reality. Constructs are used as adjectives. Ex:
Intelligent note intelligence.
Construct: unobservable phenomena. Analytical tools for organizing and trying to
understand the operations of observable phenomena. Operationalism: Operational definition. Unobservable theoretical constructs can be
logically ties to observable events allowing unobservable to be manipulated and
measured by reference to the observable event.
Falsifiability/Falsifiability Criterion: explanation of behavior must be testable to
be useful. Must be tentative, can be disproved and changed.
Occam’s Razor: No explanation that involves more, or more complex processes
should be favored over one that uses fewer or simpler processes if the two
explanations are equally good in explaining the behavior.
Lloyd Morgan’s Canon: Avoid making more assumptions that absolutely necessary.
Explanations that are narrower (can only explain under special circumstances) not
favored over general ones as long as they are equally good explanations.
Empirical Principles or Laws: assertions accepted as truths on basis of empirical
Hypothetico-deductive Method: way to evaluate theories by making a testable
hypothesis, designing research to get data and finding how well data conforms to
hypothesis and reevaluating theory in light of data.
Theory: analytic structure, principles and laws (from research), explain set of
observations. Lead to formulation of hypotheses.
Hypotheses: specific statements about how some variables should affect or be
related to other variables. Will be confirmed or not by the outcome supporting or
not supporting the theory.
(1)Generating/recognizing an idea: observing or describing a phenomenon.
(2)Refinement: read existing literature. Devise research project to shed light on
a question. Seek to better describe phenomenon, examine relation of it to
some variables. Deductive Reasoning.
(3)How research should be conducted. Methods and design. Determine which
statistical procedures to use.
(4)Actual data collection
(5)Analyze: statistics. Statistical procedures used dictated by type of research
(6)Outcome: Interpretation. New information incorporates into existing finding
and theory. Inductive Reasoning.
Types of Relationships:
(1)Descriptive: Novel behavior, how often it occurs, who performs, average
amount, wide range of values or narrow range? Examine variables in a sample; sample statistics make statements about population parameters.
Confidence interval: has behavior changed over time. Single sample and
confidence interval: use information (from these studies) to make decisions.
Don’t allow conclude anything about what other variables may be related to
or causing any change in measured variable. Purely descriptive.
(2)Correlational: Values of two (or more) variables covary (changes in one
with reliable and consistent change in the other). Just because two variables
covary doesn’t imply that one causes the other. Post hoc ergo propter hoc
fallacy: misidentify correlational relationship as causal one.
(3)Causal: Changes in one variable produce changes in another covariation rule;
variables must be correlated to be causal. Correlation isn’t sufficient
condition to determine causality. Research to identify correlations precedes
research to find causation. If behavior isn’t correlated with a particular
variable, it can’t be caused by this variable.
Regression Analysis: using existence of correlational relationship as basis of
prediction. One variable is used to make predictions about values of other variable
(only works if correlated). Variable being predicted is criterion variable. Variable
from which prediction is made is called predictor variable.
3 Problems with Determining Causality from Correlational:
(1)Third variable problem: They may covary because they’re both related to
the third variable. May lead to spurious correlation, if the third variable
didn’t exist, may not be correlated. Often caused by media suggesting causal
relationship when only spurious.
(2)Directionality problem: Correlation doesn’t provide any information about
potential direction of relationship. Ex: self-esteem and academics. Low self
esteem, low grades. If we increase self-esteem, will grades increase? If given
lower grades, self-esteem decreases, if given high grades, self esteem
(3)Selection Bias: can lead to spurious correlations. Individuals with certain
characteristics may be more likely to choose different activities
environments, etc. May lead to spurious relationship between these
individuals and some other variable.
2 Conditions for Study to be able to Determine Causality:
(1)True Independent Variables: If the researcher can manipulate the different
types of treatment so as to give any of the treatments to any of the
participants then grouping variable is called an independent variable.
(2)Random Assignment: of participants to different levels of the independent
Static Group Variables/Participant characteristics/ Quasi-independent
Variables: Variables, which characterize research participants and which can form
the basis for group identification, but which cannot be manipulated. Ex: sex, ethnic
identity, aggressiveness or other