SEYS 777                                                                                          Dr. Brian Murfin

Discuss in groups of 3:   Warm-up:  “thinking about data analysis”

How should the situations listed below be studied? What data will be useful to answer the implied question(s)?  Which statistical method(s) would you use? Use of percentiles, means, standard deviation, a test of the difference between means, chi-square, analysis of variance, or the calc. of correlation coefficients?  Present reasons for your selection of the statistic.

Part 1

a. A college professor is interested in determining if there is a relationship between passing and failing her calculus course and whether or not the students took a pre-calc. course.

b. What is the relationship between writing ability and critical thinking ability?

c. A seventh-grade teacher wants to know how the students in his class stand in relationship to one another on a standardized math test.

e. The study skills tests is administered to a group of students at the beginning of the school year. They participate in a special instructional program and are re-administered the test at the end of the school year. The teachers want to know if their gain in achievement is significant.

f. The high school supervisor wants to determine the effect of cooperative learning, peer tutoring, and individualized instruction on the reading achievement of the students in the remedial program.

g. Ms. Rodriguez is interested in determining the average achievement of the students in her class on a teacher-made biology test.

Part 2

h. Locate a research article in a research journal.  What was the question or problem being investigated?  What data was used?   How was the data analyzed?  Reported?  What conclusions were reached?  Did you feel the conclusions were warranted?

i. Create a question based on a research interest you have.    What data can be collected to answer the question?   What would you do with the data to find the answer(s) to your question?

Common Statistical Terms

ALPHA LEVEL - A number set in advance of an experimental or correlational study to indicate the level of probability the researcher is willing to accept of mistakenly rejecting the null hypothesis.  For example, an alpha level of .05, written p < .05 indicates that the researcher is willing to accept a 5% chance that a statistically significant finding will be in error.

ANALYSIS OF VARIANCE (ANOVA) - A statistical method that compares two or more group means to determine if differences between the adjusted means are statistically significant.

CHI-SQUARE (X²) – Parametric and non-parametric statistical method that compares number of responses (frequencies) of different groups or cases in different categories.

CORRELATION COEFFICIENT - A statistic, usually designated r, indicating the degree to which two variables are correlated.  May take on values from -1.0 (perfect negative correlation, when variable A is high, variable B is low, and vice versa) to +1.0

(perfect positive correlation, when variable A is high, variable B is high and vice versa). A correlation coefficient of 0 indicates that variable A and B are unrelated.

CORRELATIONAL DESIGN - A non-experimental research design in which the researcher collects data on two or more variables to determine if they are related (if they consistently vary in the same or opposite directions).

DESCRIPTIVE RESEARCH - Research conducted to describe some phenomenon as it exists, rather than finding relationships between variables (correlational research) or varying treatments to observe the outcomes (experimental research). Examples are surveys, assessment and evaluation research, ethnography, and historical research.

DISTRIBUTION - A pattern of scores on some variable.  For example, a normal distribution of scores is one in which most scores are near the mean of the set of scores and other scores cluster around the mean in a bell-shaped pattern.

EXPERIMENTAL RESEARCH - Research designs in which the experimenter decides how and when subjects will receive certain treatments and observes the effect of the treatments on one or more dependent variables.

INDEPENDENT VARIABLE - A variable (such as treatment) hypothesized to cause one or more outcomes (dependent variables).

MULTIPLE REGRESSION - A statistical method that evaluates the effects of one or more

independent variable(s) on a dependent (outcome) variable.

One-Way ANOVA - Analysis of variance with a single factor.

RELIABILITY - Consistency of a measure over time, across subjects, tests, observers, or within a test or scale.

t-TEST - A statistic used to test the difference between two means for statistical significance.  Also used to test correlation coefficients and regression coefficients for statistical significance (for groups < 30).

VALIDITY - The degree to which an instrument (eg. test or questionnaire) actually measures the concept or construct it is supposed to measure. Also, the degree to which the results of a study can be attributed to treatments, rather than to flaws in the research design (internal validity), or the degree to which the results of a study have relevance to subjects and settings other than the ones involved in the research (external validity).

VARIABLE - Anything that can take on more than one value (eg. age, sex, science achievement score).

VARIANCE – A measure of the variation/dispersion of scores in a distribution.

Other:

SCALES OF MEASUREMENT – Scales used to measure variables, based on their degree

of precision.

Nominal:  Classification of objects into categories based on some characteristic e.g. gender

No order implied between or among categories.

Ordinal:  Classification based on a logical order between categories, a ranking.

Interval:  Classification has rank, equal units, arbitrary zero point e.g. temperature scales

Ratio:  Most precise scale.  Interval scale with a known zero point that reflects the absence       of the characteristic being measured e.g. weight and height measures.

CASE STUDY – An in-depth analysis of one or more events, settings, programs, social groups, individuals or other “bounded systems.” Usually an investigation of one entity, carefully defined and characterized by time and place. Could be a single study or multiple cases.

NORMAL CURVE – A bell-shaped curve based on laws of probability concerning random deviations from a population mean. From it, predications can be made about the distributions of naturally occurring phenomena in sample populations.  Parametric statistics based on normal curve distributions.