Analyzing Your Data

Document Sample
Analyzing Your Data Powered By Docstoc
					                       Psych 361: Research Experience in Psychology
                              Analyzing Your Data and
                              Writing a Results Section
These are guidelines. Please discuss these issues with your faculty mentor, as s/he may have different or
additional suggestions, strategies, and means of evaluation.

1. Eyeball your raw data
       Look through questionnaire responses, videotapes, interview transcripts, or whatever other raw data
        you have to get a sense of the actual responses your participants gave.
       Try to get a thorough sense of the types of responses, the range of responses, and the pattern of
        responses across items/measures.

2. Enter and CHECK your data entry and computations
       Check your data entry thoroughly—that means every cell of data! If possible, find a classmate to
        partner up with, to check each other’s data entry, or work together on the checking (one person reads
        from the raw data, the other confirms with the entered data). Some of the suggestions below also
        catch data entry errors, but not everything.
       Check the accuracy of all of your computed variables (average scores, composite scores, reverse-
        coded variables, etc.) – compute a few by hand for each variable and check against the numbers in
        the dataset. Also check over the computation syntax/code carefully to make sure it is correct.
       Run basic descriptive statistics on each variable you will be using in your analyses, both for the
        whole sample and for any groups (sex, age, treatment condition, etc.) .
        -   Do the means make conceptual sense? Are there group differences in means where you
            expected, and no differences where you expected none?
        -   Are the data very narrow or wide in spread (standard deviation, range)? Do groups differ in
        -   Look at the item ranges of each variable in your dataset (not just computed variables you’ll use
            in your analyses – each single variable) to make sure nothing was entered out of range. For
            example, if a scale goes from 1-7 and during data entry someone accidentally typed 77 instead of
            7 (this happens quite commonly!), this could drastically affect your results. A number out of
            range could also indicate that the data entry got shifted down a cell, which could mess up lots of
            your data, not just one variable.
        -   Check the n for each individual variable – for every missing piece of data, find out why.
       Also read Chapter 11 handout: ―Collecting, Managing, and Analyzing the Data,‖ (from
        Dissertations and theses from start to finish; reference below).

3. Eyeball your data in the computer
       Look at the distribution (scatter plot) of each individual variable you will be using in your

                                                                                   Colleen Conley ~ Knox College
       -   Is it normally distributed or skewed? What does it mean for the variable to be normal or to be
           skewed? Is it expected or problematic? Note that non-normally distributed data are not suited
           for certain statistics because these statistics rest on the assumption of normality.
       -   Is the distribution narrow or wide? Again, think about the meaning of this.
       -   Look for outliers and look into each individual outlying subject/case – investigate what is going
           on and determine if the outlier should be thrown out. Discuss this with your faculty mentor.
      Plot each relevant pair of variables (e.g., Education x Income) to get a sense of the relationship
       between them. How strong is the relationship? Is it linear or not? Positive or negative?
      Make a correlation matrix of all of your relevant variables and examine the various relationships
       and patterns.
      Get a good sense of the demographics of your sample, and how demographic information (age, sex,
       race/ethnicity, educational status, region, etc.) relates to your central variables. You may want to
       plot each of your central variables against each major demographic variable (e.g., how does
       depression vary with age, sex, race/ethnicity, and educational status?).

4. Select appropriate statistics
      Start with your research questions – what do you want the data to answer?
      Read Chapter 10, “Selecting the Appropriate Statistics.” It is very helpful!
      Review your text and notes for your statistics and research methods courses to help understand
       which statistics are appropriate and why.
      Discuss appropriate statistics (and methods for doing them) with your faculty mentor.
      Make sure you understand how your statistics address your research questions! This is very
       important, so be sure to discuss this with your faculty mentor.

5. Analyze your data
      You will probably use SPSS or some similar program to analyze your data. If you need a tutorial on
       using your statistics analysis program, ask your faculty mentor. If you are still having trouble, you
       can ask your course instructor or a classmate for help.
      Note that you will likely come up with additional analyses to do as you start analyzing your data. Be
       prepared for this – build in extra time for this and other potential problems/complications that may

6. Check your analyses
      Spot-check your analyses by doing a few computations by hand (if possible) and comparing with
       the results you obtained.
      Do the results make conceptual sense? Don’t just accept them at face value – think about them!
      If you find a significant effect, make sure you know what direction the effect is in – it could be in
       the opposite direction you predicted!

                                                                                  Colleen Conley ~ Knox College
7. Write your Results section
      In the Results section you present the actual findings – you give your conclusions about your
       questions/hypotheses/predictions. You do not interpret these findings until the Discussion.
      Decide the best way to present your findings: In text, in tables (if you have a lot of data that can be
       easily summarized, such as a correlation matrix or a regression or ANOVA table), or by figure (e.g.,
       for depicting the nature of an interaction, or showing group differences).
      As for the actual writing, the general approach is to state what you found and then present the
       data/analyses supporting your conclusions (be very clear what statistical approach you used). There
       is not too much room for creativity, but there are a couple of different approaches:
       -   Write the results as a narrative (explain what you found) and use statistics to support the
           narrative [e.g., ―Eighth graders were significantly more depressed than sixth graders, supported
           by a main effect of age on depression; t(140) = 2.97, p<.05.‖].
       -   State up-front what statistical result you found, then elaborate [e.g., ―There was a main effect of
           age on depression; t(140) = 2.97, p<.05. Eighth graders were more depressed than were sixth
      There are many technical details involved in presenting your results, and we will not go into them
       here. You will be responsible for following APA style for your final thesis and drafts. Faculty
       mentors do not want to spend the bulk of time on your drafts correct APA style errors, so do not
       count on them to do so. You will probably find that you will rely on the manual so much that it is
       worthwhile to buy your own copy. Make sure you use correct APA style as you go; don’t just save
       the formatting for the end (though you should set aside some time in the end to meticulously check
       APA style). APA style errors will be considered in your paper evaluation.
      Read Chapter 12 handout, “Presenting the Results,” for more detailed information.

Other resources:
      American Psychological Association (2001). Publication manual of the American Psychological
       Association (5th ed). Washington, DC: American Psychological Association.
       Also see:
      Bem, D. J. (2003). Writing the empirical journal article. In J. M. Darley, M. P. Zanna, & H. L.
       Roediger III, (Eds.) The compleat academic: A career guide. Washington, DC: American
       Psychological Association. Full text:
      Cone, J. D., & Foster, S. L. (1993). Dissertations and theses from start to finish: Psychology and
       related fields. Washington, DC: American Psychological Association.
      Your Research Methods and Statistics course materials.
   Our expectation: When reviewing your drafts and final theses, we will expect that you have read these
   resources and followed the guidelines they offer. We want to focus our draft-reading on fine-tuning your
   writing, not on guiding you through the kind of information you can get from these sources.

                                                                                   Colleen Conley ~ Knox College

Shared By: