Presentation on theme: "Manipulating Variables in SPSS. Changing Numerical Variables to Categorical Variables Often want/need to change variables from an ordinal, interval, or."— Presentation transcript:
Changing Numerical Variables to Categorical Variables Often want/need to change variables from an ordinal, interval, or ratio scale to a nominal (categorical) scale of measurement Why? –To use as an independent variable in an inferential statistical procedure –To combine the information from one or more existing variables
Procedures for Splitting Data Determining the cut-offs used to create categories can be done using a variety of procedures. –Non-Sample Dependent Procedures: Based on researchers previous experiences or hunches Based on information from the literature Based on categories defined by the scale authors –Sample Dependent Procedures: Median Split Tertiary Split Quartile Split Normal Curve Split
Sample Dependent Procedures Median Split: –Use to split data into two categories (e.g. high and low) –Determine the median. Place individuals who score below the median into one category. Place those that score above the median into the other category. When necessary, use own judgment to decide where to place those who score at exactly the median.
Sample Dependent Procedures Tertiary Split: –Used to divide numerical data into three categories of equal number –Determine the scores at the 33 rd and 66 th percentiles. Place those that score in the bottom third in the first category. Place those that score in the middle third in the second category. Place those that score in the top third in the third category.
Sample Dependent Procedures Quartile Split: –Used to divide numerical data into three categories –Determine the scores that correspond to the quartiles. Place those that score in the bottom 25 percent in the low category. Place those that score in the top 25 percent in the high category. Place those that score in the middle 50 percent in the moderate category.
Sample Dependent Procedures Normal Curve Split: –Used to divide numerical data into three categories –Determine the scores that correspond to z-scores of one and negative one. Place individuals at or below the 16 th percentile into the low category. Place individuals above the 84 th percentile into the high category. Place individuals who score above the 16 th percentile and below or equal to the 84 th percentile into the moderate category.
Response Styles Response Styles : tendencies to respond to questions or test items in a specific way, regardless of the content
Response Styles Willingness to Answer : the differences among people in their style of responding to questions they are unsure about Position Preference : when in doubt about answers to multiple-choice questions, some people always select a response in a certain position A B C D
Response Styles Manifest Content : the plain meaning of the words or questions that actually appear on the page Yea-Sayers : people who are apt to agree with a question regardless of its manifest content Nay-Sayers : people who are apt to disagree with a question regardless of its manifest content
The Social Desirability Response Set Latent Content : the “hidden meaning” behind a question Response Set : a tendency to answer questions based on their latent content with the goal of creating a certain impression of ourselves Some subjects tend to give the socially desirable answer
Dealing with Response Styles and Response Sets Ask participants to answer all items. Clarify that there are no right or wrong answers. Simple yes/no and agree/disagree questions make it easy for subjects to respond based on response style. Build more specific content in the questions. Reverse order some of the questions/responses. Ask the same question multiple ways. Measure social desirability.