Docstoc

Data Analysis

Document Sample
Data Analysis Powered By Docstoc
					Data Analysis
              Using Statistics
• There are several reasons researchers use
  statistics in their research.
   –   To describe
   –   To identify relationships
   –   To determine if there are differences
   –   To identify other variables that may be
       impacting on the research
              To Describe
• Descriptive statistics are used to provide an
  overview of the data
• Typically, the population being studied is
  described statistically. This helps the reader
  of the research to see if the research can be
  generalized to other groups.
       Descriptive Statistics
• The Mean – This is simply the average.
• The Median – This is the halfway number.
  Half of the scores are above this number and
  half are below. When one or two extreme
  scores may skew the mean, the median is a
  better descriptor.
• The Mode – This is the score found most
  often in the data.
       Descriptive Statistics
• The Range – This is the “distance” between
  the highest score and the lowest score.
• Maximum – The highest score in a range.
• Minimum – The lowest score in a range.
     Descriptive Statistics
• Standard Deviation – A number that
  tells how close together or how spread
  out the scores are. The smaller the
  number the more closely grouped the
  scores are.
• Typically, we like to see closely
  grouped scores .
        Standard Deviation
• Assume we have three groups of scores
   – Group 1 – 1,2,3,4,5,6,7,8,9,10
   – Group 2 – 5,5,5,5,5,5,5,5,5,5
   – Group 3 – 4,5,6,4,5,6,4,5,6,5
• The mean score for each of the three groups
  is 5.0, but the Standard Deviation is 3.0, 0,
  0.8 respectively. Without looking at the raw
  scores the SD tell us there is great variation
  in scores for the first group, none in the
  second, and a little variation in the third.
        Standard Deviation
• It is accepted practice to also report the
  Standard Deviation when you are reporting
  the Mean in a research report.
• The Standard Deviation and the Variance
  (the Standard Deviation is the square root of
  the variance) are important in calculating
  statistical tests.
          Statistics Identify
            Relationships
• In an earlier lesson we discussed the various
  types of correlations and how to interpret
  them. Correlational statistics are used to
  identify relationships between variables.
   – You might want to review this information.
        Statistics are Used
      to Identify Differences
• Whenever we conduct research, we often
  want to know if the difference between two
  groups is a “real” difference or did it just
  happen by chance.
• There are several statistics used to
  accomplish this.
                    t-test
• The t-test is used to determine if there
  are differences between two groups
  when the dependent variable is interval
  or ratio.
  – Test scores, job satisfaction scores, salary,
    academic achievement, gain scores, etc.
                      t-test
• There are two types of t-tests
   – Independent or a two-sample t-test – is used for
     comparing two separate groups of individuals. Is
     group A different than group B?
   – Paired t-test - is used for comparing the same
     group of individuals on two scores (such as a
     pretest score and a posttest score).
                          t-test
• The result of a t-test is a value for t,
  such as t=4.61
• Unlike correlations the value of t means nothing,
  it cannot be interpreted.
   – You must look at the P value (Probability) associated
     with the t-test. If the value of P is equal to or less than
     .05, we can conclude the two groups are not the same.
     In other words, our findings are statistically
     significant.
      Analysis of Variance
          (ANOVA)
• ANOVA is used to determine if there
  are differences among three or more
  groups when the dependent variable is
  interval or ratio.
  – Test scores, job satisfaction scores, salary,
    academic achievement, gain scores, etc.
                       ANOVA
• The result of an ANOVA is reported as a value for
  F such as F=4.61.
• Unlike correlations the value of F means nothing,
  it cannot be interpreted.
   – You must look at the P value (Probability) associated
     with the F-value. If the value of P is equal to or less
     than .05, we can conclude the three (or more) groups
     are not the same. In other words, our findings are
     statistically significant.
              ANOVA
• One of the problems with ANOVA is
  when we have statistically significant
  results. The problem is which group is
  different (because we are dealing with
  three or more groups)? ANOVA
  doesn’t identify the difference.
                 ANOVA
• In order to determine where the differences
  are, we have to perform a procedure called
  post hoc analysis. This technique allows us
  to identify which groups are different from
  the other groups.
• There are a variety of post hoc techniques
  that can be used. It all depends upon the
  characteristics of the data.
                Chi Square

• Sometimes our dependent variable is
  categorical in nature.
   – Such as honor roll status; member or not;
     socioeconomic status; rural, suburban or urban;
     obese or not; etc.
• We use the Chi Square test. This can be used
  with two groups, three groups, or more.
                 Chi Square
• The result of a chi square test is a value for
  c2, such as c2 =4.61
• Unlike correlations the value of c2 means
  nothing, it cannot be interpreted.
   – You must look at the P value (Probability)
     associated with the c2. If the value of P is equal
     to or less than .05, we can conclude the groups
     are not the same. In other words, our findings are
     statistically significant.
               A Problem
• One of the deficiencies of research in
  agricultural and extension education is that it
  tends to be simplistic. We may think one
  variable is causing the effect and focus solely
  on that variable, when in fact several
  different variables may be combining to
  cause the effect.
              An Example
• Attendance at the annual conference of the
  Association for Career and Technical
  Education has been steadily declining. Why?
   – Could it be the membership numbers have
     declined, so we should expect a decline in
     attendance?
   – Or has the rising registration cost resulted in
     declining attendance?
   – Or is the location of the conference?
   – Or is it something else?
• One major factor could be the problem or it
  could be a combination of factors.
             The Solution
• There is a statistical technique called
  Multiple Regression that examines a number
  of independent variables then identifies the
  ones causing the change in the dependent
  variable.
• This procedure can even identify the
  contribution of each independent variable on
  the dependent variable.
       Multiple Regression
• It can also tell us how much of the
  variance (difference) can be explained
  by the independent variables we have
  selected. There may be other variables
  at work that we have not yet identified.
              The Tools
• How do we go about calculating all of
  these statistical tests?
  – In the old days these were hand calculated.
    It took several hours, even days.
  – Today we use a computer.
         Statistical Analysis
• Excel has a statistical module. If you do a
  “standard” install of Excel, this module is not
  loaded. You have to do a custom install and
  select to load the statistical module. (installs
  as standard on 2007)
• Excel can perform a number of the tests we
  have discussed.
           Statistical Analysis
• The primary statistical tool used by researchers in
  agricultural and extension education is SPSS-
  (Statistical Package for Social Scientists)
• This is an extremely powerful software program
  and it is easy to use.
• This software is one of the installed applications
  on your Novell launcher in the computer labs.
             Statistical Analysis
• There are several web sites where you can paste
  data and perform statistical analyses.

   – Webstats http://www.webstatsoftware.com/

   – Vassarstats http://faculty.vassar.edu/lowry/VassarStats.html

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:3
posted:7/24/2012
language:
pages:27