Research Methods Test 1

Front 1

Empirical Research

Back 1

Uses inductive reasoning to obtain evidence from particular cases and make inferences about general principles (bottom up approach); knowledge gained via systematic observations; inductive reasoning-obtain evidence from particular cases and make inferences about general principles.

Front 2

Experimental Research

Back 2

Characterized by control, manipulation, and randomization; involves the application of some form of intervention or treatment; able to infer cause and effect

Front 3

Descriptive Research

Back 3

Characterized by observation without manipulation or control; experiments of nature; observational studies comparing groups or conditions; UNABLE to infer cause and effect.

Front 4

Qualitative research

Back 4

inductive reasoning, purposeful (small) sample, emerging meaning/hypothesis; new topic/closed culture; rich in depth analysis; close E/P interaction; awareness of bias; interpretive themes; interactive; clinical utility

Front 5

Quantitative Research

Back 5

deductive reasoning, large (random) sample; testing hypothesis; uses instruments; a quick snapshot; E/P distance objectivity; try to control bias; use of statistics; group (sample) data; generalization

Front 6

Dependent Variable

Back 6

answers the question: what do I observe?; what you measure in an experiment and what is affected in the experiment; what can change or is dependent based on the independent variables.

Front 7

Independent Variables

Back 7

stands alone; isn't changed by other variables (ex: age)

Front 8

Null Hypothesis

Back 8

statistical statement indicating no difference between two (or more) groups.

Front 9

Measure of effect size

Back 9

refers to the magnitude of a difference between scores expressed as a standard score. (ex. a range of 0-3 large effect size is .8; moderate is .5; small is .2)

Front 10

Population Parameter

Back 10

all members of a defined population

Front 11

P<.05

Back 11

statistically significant; the most liberal level most likely to reject the null hypothesis

Front 12

ANOVA

Back 12

analysis of variance; allowing for testing across 3 or more groups; avoids problems of multiple T tests; (a) one way anova- ex three levels of therapy (b) two way anova- ex grade x gender (3x2); formula: F=variance b/w (treatment + error)/ variance within (error)

Front 13

Four levels of data

Back 13

Nominal- categorical
Ordinal- Categorical and rank
Interval- equal intervals/arbitrary zero
Ratio- established true zero

Front 14

Statistical Significance

Back 14

likelihood or probability; not happening by chance

Front 15

Statistical Power

Back 15

the probability of correctly rejecting the Null Hypothesis (influenced by N)

Front 16

Non-directional experimental hypothesis:

Back 16

there is a difference but there is not enough information to speculate on the direction of the differences

Front 17

Directional experimental hypothesis

Back 17

states that one particular group's average is higher than the other group's average.

Front 18

Confidence interval

Back 18

account for sampling error when generalizing to the population parameter by increasing the likelihood of including the parameter; typically set at 95% (willing to be wrong 5% OTT) and 99% (willing to be wrong 1% OTT) when data are considered to be normally distributed

Front 19

Type I and Type II Error

Back 19

Type I- incorrectly rejecting the Null hypothesis (alpha error)

Type II: failing to reject a false Null Hypothesis (beta error)

Front 20

Negatively skewed distribution

Back 20

the median is higher than the mean; skewed to the left

Front 21

Positively Skewed distribution

Back 21

the mean is higher than the median; skewed to the right

Front 22

Normal, platykurtic, and leptokertic distributions

Back 22

Normal: bell shaped curve; symmetrical, infinity at each tail, mean, median, mode at center.

Leptokurtic- peaked and fat tails (Kur +)

Platykurtic- less peaked and thinner tails (Kur -)

Front 23

Standard error of the mean

Back 23

standard deviation of the mean

Front 24

Tabled Critical Values

Back 24

Critical values on a t and z score are the values the calculated values must exceed to be statistically significant.

Front 25

Cohen's D

Back 25

Widely used measure of effect size; divide the difference in the means of the two groups by the SD of one of the groups (typically control group) or if the SD are not similar use the pooled standard deviation of the two groups.

Front 26

A priori and post hoc testing

Back 26

Priori: done when prior theory leads to a specific hypothesis about where significance differences might be found. Revised version of T test.
Post hoc: specific comparisons of groups testing with a variety of multiple comparison tests. Analyzes all possible comparisons.

Front 27

Sampling Error

Back 27

random error (created by random selection)

Front 28

Multivariant Experiment

Back 28

MANOVA: examines the effects of one or more independent variables on multiple dependent variables. Helps control overall type I error.

Front 29

Parametric Experiment

Back 29

interval and ratio data, homogeneity of groups, used for comparing variance

Front 30

Norm referenced test

Back 30

compares an individual's performance with that of a norm group

Front 31

Content of major sections of a manuscript

Back 31

1. Title: key independent/dependent variables
2. Authorship
3. Abstract: key results; concise and efficient
4. Literature review: Justification and logic for study; research question
5. Methods: participants; procedure; research design
6. Results: brief statement of questions and general findings. Then detailed graphs and results. No interpretations just facts
7. discussion: interpret resultsonly if significant, suggestions for future studies
8. acknowledgement
9. References:
10. Appendices: explain procedures, selection criteria, measurements, algorithms, tests, etc.

Front 32

The peer review process

Back 32

1.
2.
3.
4.

Front 33

Publication lag

Back 33

2-3 years

Front 34

Z and T tests

Back 34

1. z scores: provide precise indications of the area under the curve from the mean of zero to points on the baseline that define (in both +/- directions) the total area under the curve.
2. T tests- small sample sizes and the shape of distribution varies with the number of scores/participants. values increase with smaller samples b/c of greater variability.

Front 35

Measures of central tendency and dispersion

Back 35

standard deviation, variance, range

Front 36

Sagan suggests that science involves a balance between what two concepts?

Back 36

Research/asking any question and defending idea (expert criticism)

Front 37

Siegel suggests that it's not enough to demonstrate that a therapy protocol works but that we also need to:

Back 37

explain how it fits into the overall model and how it will benefit client.

Front 38

Clinical Significance has to do with:

Back 38

how the client feels (asking the client their opinion) a functional issue; meaningful or clinical change; clinical utility and effect size.

Front 39

Statistical significance has to do with:

Back 39

probability

Front 40

In Michael Chial's example of the competitors in the 100 M foot race, what are the examples of:
Nominal data:
Ordinal data:
Interval Data:

Back 40

Nominal: categories (gender, race, etc).
Ordinal: category and rank (order of finish)
Interval: precise measurement of the time difference b/w individuals (arbitrary zero)

Front 41

List the characteristics of interval-level data

Back 41

arbitrary zero; mutually exclusive categories; can be used with parametric statistics; equal occurring intervals/quantities; rank, order, differences

Front 42

The best indicator of central tendency for ordinal data scores is ______. This would be especially true when the data is _______.

Back 42

Median
Skewed.

Front 43

The two basic purposes of inferential statistics:

Back 43

1. accept/reject null hypothesis
2. infer sample to a population

Front 44

A class is titled statistics and not parameters because:

Back 44

You cannot measure parameters because this is an entire population (greek symbol). You can only test a sample which would be a statistic (roman letter) that you then infer to a parameter.

Front 45

To be worthwhile a theory must be:

Back 45

concrete enough to be testable and specific

Front 46

Typical measures of variability or dispersion are:

Back 46

standard deviation; variation; range

Front 47

When investigating a phenomenon that is not well understood the test approach would likely be a: quantitative or qualitative approach?

Back 47

qualitative approach: you need to find out important variables and factors.

Front 48

What are the purposes/advantages of using standard scores? What are some examples?

Back 48

You are able to compare one score to a group of scores.
Example: z score, t test, effect size

Front 49

What proportion of the total area under the normal curve is defined by:
-1 sd to +1 sd
-2 sd to +2 sd

Back 49

-1/+1: 68%
-2/+2: 95%

Front 50

What is the purpose of developing a confidence interval when making inferences from sample statistics to the population parameters? What are the two confidence intervals which are calculated?

Back 50

It is a margin of error for estimating the mean of the population parameter. Intervals 95% and 99% confidence interval used often.

Front 51

Why use z score of 1.96 to define the 95% confidence interval?

Back 51

95% is roughly the area b/w -2/+2 sd from the mean and the 1.96 would be about the mean in a normal distribution.

Front 52

Who developed the t-distribution and why were they created?

Back 52

William Gosset in dublin Ireland at the Guiness brewery....created to measure deviations and variance in small sample populations.

Front 53

How does the shape of the t-distribution differ from the normal distribution. Why does the shape of the t-distribution vary according to df?

Back 53

The shape differs in that it is usually a flatter curve with tails slightly upward. With higher df the shape will resemble more of a normal curve b/c of more samples. With lower df it will become flatter b/c of wide range of scores w/few people.

Front 54

The standard error of the mean is the standard deviation of:

Back 54

mean scores

Front 55

What is the purpose of the values in the t-tables?

Back 55

Critical values: Sample data must be higher than critical values to be statistically significant.

Front 56

Why do more scores cluster near the middle of a distribution when you increase the number of subjects?

Back 56

The more scores you have the less variability between all the scores and the mean will be closer to the middle point btween them. More subjects= higher statistical significance

Front 57

The central goal of inferential statistics in the behavioral sciences is to determine whether the observed ________between two groups is great enough to justify the conclusion that they are not due to _______ or variations among individuals.

Back 57

differences; sampling error.

Front 58

In the following sequence of the steps in hypothesis testing which critical step is missing?
-state the null hypothesis
-select a statistical test
-calculate the test statistic
-contrast the calculated value w/critical value
-draw conclusions

Back 58

Most important is what you alpha level will be or probability level. Type I error.

Front 59

Describe the Null Hypothesis:

Back 59

The null hypothesis is the previously held idea about a relationship between two groups where there is no difference between the two.

Front 60

Describe a directional research hypothesis

Back 60

Proposing that one of the group's mean is higher than the other group's mean. One tailed test.

Front 61

Explain critical (tabled) values and calculated values obtained from a statistic

Back 61

The critical values are the value your calculated value needs to be greater than or equal to to be statistically significant. The calculated value is the df, whether it is a two tailed or one tailed test, and the alpha level.

Front 62

Are the critical values for a t-statistic greater for a one-tailed or two-tailed test?

Back 62

They are greater for a one-tailed test because instead of the 5% not accounted for (Type I error) and placing it on one end rather than splitting it makes the statistic greater. All alpha in one tail so critical value is closer to the mean.

Front 63

The standard error of the difference of Sx1-x2 is

Back 63

Standard deviation of mean/different scores

Front 64

The choice of an alpha level and a one or two-tailed statistical test should be guided by the: _____ but not by: ______

Back 64

literature/research question you are

but not by: convenience

Front 65

What are the basic steps in conducting research?

Back 65

-become familiar with the literature
-recognize an issue, problem, formulate your question
-create a research strategy/design
-develop your rationale and methods
-gather your data (run participants)
-analyze your data (results)
-interpret and draw conclusions
-report the results to your peers

Front 66

Serendipity

Back 66

discovering something not originally sought

Front 67

Pilot Studies

Back 67

essential for understanding (saving time and money)

Front 68

The principles of parsimony

Back 68

do not make more assumptions than the minimum needed to explain something. Provide the simplest explanation using the fewest possible assumptions. Choose from a set of otherwise equivalent models of a given phenomenon, the simplest one.

Front 69

Types of research

Back 69

1. basic: search for understanding; how things work; often without concern for application; may have eventual or unintended application at some later time
2. applied- sometimes referred to as clinical research; goal is immediate application in a (clinical) setting

Front 70

Types of research continued

Back 70

1. Descriptive research: observation without manipulation or control; experiments of nature; observational studies comparing groups and conditions; UNABLE to infer cause and effect
2. experimental research: control, manipulate, and randomize; application of some form of intervention or treatment; able to infer cause and effect.

Front 71

Rationalism

Back 71

knowledge gained via logical thought; deductive reasoning-apply general principles to make inferences about specific cases.

Front 72

Define Theory:

Back 72

a theory is a set of interrelated constructs, concepts, definitions, and propositions that represents a systematic view of a phenomena by specifying relationships among the variables. The purpose is to explain and predict the phenomena.

Front 73

What are the steps in publishing a manuscript?

Back 73

-read information for authors
-send manuscript to journal editor
-associate editor sends to reviewers
-reviewers select 3-4 options
-associate editor makes recommendation to editor
-editor makes decision and corresponds with authors
-time lag to get published is 2-3 years

Front 74

What are the four choices peer reviewers have?

Back 74

1. accept it
2. revise with minor errors
3. revise with major errors
4. outright reject

Front 75

When reviewing a manuscript what is the internal criteria for a reviewer?

Back 75

information independent of the subject area; the technical aspect of the ms. regarding threats to internal and external validity of the study, organization, and clarity of writing.

Front 76

External Criteria?

Back 76

information related to the subject area; how does the research relate to the field; is it a meaningful and timely contribution (reviewer must have expertise in this area)

Front 77

Define the technician

Back 77

more likely to base decisions on dogmas, an authority. The focus is on the book or manual.

Front 78

Professional:

Back 78

knows many strategies and techniques; considers alternative explanations for behavior and outcomes. The focus is on the person and their goals.

Front 79

What are the 4 basic strategies for diagnostic decision making?

Back 79

1. Pattern recognition: recognition of visual and/or auditory features
2. multiple branching: flowchart sequencing through diagnostic steps; compensates for less experience but works poorly w/complex and dynamic problems
3. Diagnosis by exhaustion: search for all possible facts following extensive testing
4. Hypothetico-deductive: hypothesis generated based on experience, pattern recognition, and understanding nature of problem. Primary strategy is to disprove one or more working hypothesis.

Front 80

Limitations of group design

Back 80

lack homogeneity of participants; individual response to intervention; averaging of scores; ethical issues

Front 81

What is a good measure of clinical significance?

Back 81

Cohen's D

Front 82

What is another name for non-parametric statistics?

Back 82

distribution free statistics; no assumption of a normally distributed population.

Front 83

Paragraph with Confidence interval

Back 83

Although sample statistics provide a point estimate of teh population parameter (mean) there is always the possibility of sampling error. Therefore confidence intervals are determined based on the estimated standard error of measurement from the sample data. Confidence intervals based on the sample data increase the likelihood of including the parameter and are typically set at 95% (critical values of +/- 1.96) or 99% critical values of (+/- 2.58) when the data is considered to be normally distributed.

Front 84

Meta-analysis

Back 84

studies are the subjects, effect size (like standard scores, can compare data across areas)

Front 85

Central limit theorem

Back 85

1. Means of a multitude of equal-sized samples drawn from a normally distributed population are themselves normally distributed.
2. The means of a multitude of qual sized samples, regardless of the shape of the population distribution approaches a normal distribution as the sample size reaches 30.
3. the means of a multitude of normally distributed sample means is known as the population mean.

Front 86

How are type I and type II errors related?

Back 86

inversely. as one increases the other decreases.

Front 87

What are the methods of analyzing relationships with parametric statistics?

Back 87

Pearson Product moment correlation coefficient and multiple-regression analysis (must know nature of data)

Front 88

What are the two limitations for t tests with 3+ levels of experience?

Back 88

not efficient and using multiple t comparisons increases the likelihood that one or more of them will reach a significant difference by chance. Probability of falsely rejecting the null (a type 1 error) is increased.

Front 89

ANOVA

Back 89

ratio of systematic variance due to the treatment effect regarding the variance due to sampling (between-group) variance and measurement error (within-group). Determines the variability between groups is greater than within groups.

Front 90

3 assumptions of ANOVA

Back 90

independence among groups of subjects, sampling distribution of scores follow a normal curve, and variances of the scores in each group are equal (homogeneity of variance)

Front 91

How can the assumption of homogeneity not be violated with ANOVA?

Back 91

as long as each group is constituted by an equal number of subjects of scores.

Front 92

Descriptive Statistics Goals

Back 92

Goals: describing a sample of data (central tendency and dispersion); describing how a single datum relates to an entire sample (standard score); describing the amount and direction of association between two samples of data.

Front 93

What do measures of dispersion describe?

Back 93

dissimilarities in a group of data; the way events scatter. (range, variance, standard deviation, and standard error of the mean)

Front 94

What is the standard error the mean considered "normalized?"

Back 94

it takes sample size into account, therefore it allows direct comparison of the amount of dispersion associated with different sized samples.

Front 95

Standard scores express the relation of a single score to:

Back 95

central tendency and dispersion of a group of scores. (z score)

Front 96

What does a z score tell you?

Back 96

where a particular raw score is placed on a set of scores

Front 97

what does a standard score of zeor indicated? Positive? Negative?

Back 97

Zero:particular raw score equals the mean; positive: occurs when a particular score exceeds the mean; negative: when a particular score is less than the mean. Magnitude tells how far a raw score is from the mean (in units of SD)

Front 98

Z scores are "standardized" how? And when are they valid?

Back 98

they allow direct comparison of distributions having different means, ranges, and dispersions; they are valid only if based upon large distributions of interval or ration scale numbers for which the mean and median are similar

Front 99

What do correlation coefficients describe?

Back 99

degree of similarity between distributions

Front 100

What are the commonly used correlation coefficients?

Back 100

interval/ration- pearson product moment correlation coefficient; ordinal- spearman rank-order; nominal- contingency coefficient

Front 101

What does a correlation of 0 indicated? Positive? Negative? (+1/-1)

Back 101

zero means the two sets of data are unrelated. Positive indicates a direct relationship and negative indicates an inverse relationship.

Front 102

What is the magnitude of correlation coefficient?

Back 102

<.2 "negligible", .2-.4 weak, .4-.7 moderate, .7-.9 strong, .9-1 very strong

Front 103

What does the square of the correlation coefficient (coefficient of determination) indicate?

Back 103

strength of association between two sets of data expressed as a percentage of total variation in scores attributable to whatever the distributions have in common.

Front 104

Hypothesis testing

Back 104

two or more sets of sampled data can be compared to determine whether they differ with respect to some specific index or with respect to the entire distribution.

Front 105

What is the central goal of inferential statistics?

Back 105

determine whether observed differences between groups are great enough to justify the conclusion that they are real and not due simple to random errors of measurement, or variations among individuals.

Front 106

Describe goal of inferential statistics

Back 106

ascertain whether data drawn from two or more groups differ with respect to some specified characteristic such that differences may be considered "statistically significant."

Front 107

Some inferential tests are designed to compare what?

Back 107

on empirical group to a population already known; two empirical groups; comparison of 3 or more groups.

Front 108

What do most inferential tests assume?

Back 108

That groups to be compared to each other are statistically independent (subjects have been measured only once); this is b/c procedures involving repeated observations systematically constrain the variance of events.

Front 109

Post hoc

Back 109

Statistical tests of groups not independent of each other (test-retest measures)

Front 110

What are most research hypotheses about?

Back 110

measures of central tendency, dispersion, or entire distributions

Front 111

YOu should select a test for your research hypotheses that conforms to what?

Back 111

a. the scale of measurement implicit in data acquisition b. the nature of what the hypotheses are about c. the number of groups being compared d. whether the groups are statistically independent or related.

Front 112

W/ research hypotheses selection of a statistical test specifies what?

Back 112

statistical model, test statistic, and a sampling distribution for that test statistic.