CRP phase 4Analyzing and Interpreting Quantitative Data

Describing Quantitative Data

Quantitative data Data that can be quantified and verified, and is amenable to statistical manipulation. Quantitative data defines whereas qualitative data describes.

I.MAKING SENSE OF NUMBERS: STATISTICAL DATA ANALYSIS

Data analysis is the process of examining what data mean to researchers. Statistical data analysis refers to the process of examining what quantitative data mean to researchers. Statistics refers to any numerical indicator of a set of data.

Descriptive Statistical Data Analysis: - method used to describe the population we are studying.
Frequency distributions, measures of central tendency (mean, median, and mode), and graphs like pie charts and bar charts that describe the data are all examples of descriptive statistics.

Inferential Data Analysis: - making predictions or inferences about a population from observations and analyses of a sample.

2 PURPOSES
Estimates the characteristics of a population from data gathered on a sample. Tests for significant DIFFERENCES between groups and significant RELATIONSHIPS between variables.

II. DESCRIBING DATA THROUGH SUMMARY STATISTICS

Summary statistics provide an efficient way to describe an entire set of quantitative data.

A. Measures of CENTRAL TENDENCY

describe the central point of distribution (representative values)

The measure of central tendency describes the one score that best represents the entire distribution, the most CHARACTERISTIC SCORE. The most characteristic score is that which describes the center point of a distribution of scores.

B. Measures of DISPERSION

Describe how scores differ; how scores are SPREAD (measures of VARIABILITY) Do not exist for nominal data because categories are used rather than meaningful numbers.

RANGE (or span) - number that reports the distance

between the highest and lowest scores in a distribution.

VARIANCE - number that represents the mathematical

index of the average distance of the scores on an interval or ratio scale from the mean in squared units. a summary statistic of how much scores vary from the mean - expressed in the original units of measurement.

STANDARD DEVIATION- number that represents

Standard scores provide a common unit of measurement indicating how far away any particular score is from the mean. Used to compare numbers(scores) from different distributions. Z score- most popular standard score used by researchers. The z score is calculated by dividing the deviation score by the standard deviation (X - X bar) divided by the standard deviation. Each z score indicates how many standard deviations that score is from the mean of the distribution.

C. DESCRIBING DATA IN STANDARD SCORES

B. FREQUENCY Distributions

Counting and reporting HOW OFTEN different categories or points on a measurement scale occur. A list of the frequency of responses for each category or measurement point is called a frequency table.

 Show

changes over time

Frequency Tables - a total number of times particular values on a measurement scale occur in a data set. Pie Charts - illustrate the frequency counts of categories. Bar Charts - visually illustrate frequency counts for a nominal and ordinal variables. Line Graphs - use a single point to represent the frequency count on a dependent variable for each of the groups.
Interval

and Ratio level variable frequencies are illustrated using...

Frequency Histograms - distribution of frequencies where blocks touch Frequency Polygons - similar to line graphs except a line connects the points representing the frequency count for each point on the measurement scale rather than each category.

CRP Phase 4- Inferring from Data: Estimation and Significance Testing

INFERENTIAL STATISTICS :
The set of statistical procedures that allows a researcher to go beyond the group that has been measured, and make statements about the characteristics of a much larger group.

Inferential statistics:
accomplish ESTIMATION and SIGNIFICANCE TESTING.

I I. ESTIMATION (Estimation Theory) A. ESTIMATION

Estimates of Population PARAMETERS from Sample STATISTICS are possible as long as 2 ASSUMPTIONS met: Normal Distribution and Random Sample.

1. Assumes that the charactetristic of interest is measured on an INTERVAL or RATIO scale and that the characteristics are distributed NORMALLY 2. Assumes that the SAMPLE has been selected RANDOMLY so it reflects the population.

Estimation procedures are often referred to as PARAMETRIC STATISTICS

B.The Normal Distribution Normal Curve (symmetrical bell shaped curve)

Distributions NOT in the shape of a normal curve can be described using two characteristics:
1. Skewness- the majority of the scores are toward one end of a distribution, which makes the curve tail off in that direction.

a. Positive - tail runs to right side of the curve; b. Negative - tail runs to the left side of the curve.

2. Kurtosis- refers to how pointed or flat the shape of a distribution is when it is not a normal curve.

a. Peaked - scores are clusterd around the middle; LEPTOKURTIC distribution b. Flat - scores are dispersed evenly across the distribution; PLATYKURTIC distribution

C. Use of Random Sampling - each population member has an equal chance of being selected for the sample.

1. Central Limits Theorem - for samples of a sufficiently large size, the real distribution of means is almost always approximately normal. 2. Large samples give more accurate results than do smaller samples. 3. Random samples allow researchers to operate safely under the normal curve. 4. Sample must be representative of population. KISS

D.Inferring from a Random Sample to a Population

1. Sampling Distribution a. Sampling distribution mean b. Standard ERROR (SE) of the mean tells researchers how much these random sample means are likely to differ from the mean of the sampling distribution. 2. Confidence Level - the degree of assurance that a sample statistic can be used to infer a population parameter. 3. Confidence Interval - the range of scores associated with this confidence level. Be cautious!

II. SIGNIFICANCE TESTING

The process of analyzing quantitative data for the puprose of testing whether a nulll hypothesis is probably either correct or false. Significance testing = hypothesis testing.

A. The Logic of Significance Testing

1. Testing a Null Hypothesis -- every research hypothesis has a corresponding null hypothesis. 2. Rejecting a Null Hypothesis -- We do not "PROVE" a research hypothesis. 3. Deciding on the Probability Level (p value) a. The 95% confidence level is referred to as the .05 significance level. b. A significant difference or relationship found at the .05 level of probability means that the researchers are 95% confident that the difference or relationship is NOT DUE TO CHANCE OR ERROR. c. The result is only likely to happen BY CHANCE 5 out of 100 times.

B. The Practice of Significance Testing

1. Step One: Posing a Research Question or Hypothesis and a Null Hypothesis 2. Step Two: Conducting the Study 3. Step Three: Testing the Null Hypothesis

C. Type I Error and Type II Error

1. Type I error a. ALPHA error b. False Positive c. Researcher rejects the null and accepts a research hypothesis d. Much more serious 2. Type II error a. BETA error b. False Negative c. Researcher accepts the null and rejects a research hypothesis d. Not as serious

D. Statistical Power

1. POWER Analysis a. The ability to detect effects of a specific size, given the particular variance and sample size of the study b. Equal to 1 minus the probability of committing a Type II error (failing to reject a false null hypothesis) c. Minimum statistical power is 0 and maximum is 1.0 d. Statistical Power should be .80 at a minimum 2. Increasing Statistical Power a. Employ the most sensitive (and appropriate) statistical test b. Test one-tailed hypotheses and research questions c. Increase the size of the random sample studied 3. Effect Size a. An estimate of the degree to which a phenomenon is present in a population and/or b. The extent to which the null hypothesis is false c. Three Types of Effect Sizes 1) Small n= 393 2) Medium n= 64 3) Large n= 26

III. IMPORTANT CONCEPTUAL DEFINITIONS

Parameters - a characteristic of a population or a universe. Statistic - the measurement of a sample with respect to a variable. Nonparametric Statistics -statistics used only to describe the characteristics of a sample, without being able to generalize back to its population. Parametric Statistics - statistics used to estimate the characteristics (parameters) of a population based on the characteristics (statistics) of a sample. Data Analysis - the methods researchers use to infer meaning from data, to determine what conclusions are justifed.