STS401_Ch6
SAS
a set of statistical computer routines and a programming language: Statistical Analysis System
antecedent variable
the variable initially leading to change in the dependent variable; p. 174
intervening variable
the variable through which the antecedent variable brings about change in the dependent variable; p. 174
contingency table: interpreting results
1. Examine the increases or decreases; a positive relationship will be indicated by a clustering on the main diagonal (upper left to lower right); an inverse relationship is indicated by off diagonal clustering (lower left to upper right) 2. a positive relationship will show the highest % in the upper left-hand row getting smaller to right; the bottom category row should go lowest to highest. 3. In examining the relationship between two variables, look for differences among categories of the independent variable. We are interested in how the independent variable (IV) affects the dependent variable (DV).
percentaging
1. Treat each cell entry as a percentage of the grand total (ex: if have 7 cell entries in one category and a grand total of 20, take 7/20=.35 or 35%. Not very useful 2. Treat each cell entry as a percentage of its row total (7 in the high-high cell out of a total of 12 countries with high political rights), 7/12=.5833 or 58.33% 3. Treat each cell entry as a percentage of its column total (7/7=1.000 or 100.0%) The convention is to percentage so that the independent variable, or columns of the table, equal 100%. Every table needs a title
testing the variables
1. create a cross-tab that displays the relationship between 2 variables 2. if data in frequencies, change to percents (need percents as a base for comparison) 3. if data is interval, group data into categories
contingency table: tips
1. only need one percentage sign (%) in body of table, in upper left-hand cell; also include % signs by the 100.0% totals at the bottom 2. If one total at bottom is 99.8% it's usually due to rounding error, however if you get 96 or 104% it's a problem 3. always report the number of cases (n=) for each category of the independent variable (under total percents); helps to retrieve data later 4. include a footnote below table citing the sources of the data and (where possible) page references; failure to cite sources is plagiarism
Microsoft Excel
Microsoft's spreadsheet program that also may be used for statistical analysis
Microsoft Windows
Microsoft's widely used operating system
relationship in tables: positive
a pattern in a table that is the most commonly found pattern showing a positive relationship of top row (highest to lowest moving right) and bottom row (lowest to highest % moving right)
spurious relationship
a relationship that appears to be caused by one varible, but is really caused by another; a coincidental statistical correlation between two variables which is shown to be caused by some third variable.beware of these; often the stats you hear in ads or on tv are the results of spuriousness; relationship between two variables that is the product of a common independent variable; ex: agriculture and telephones are functions of a single independent variable-wealth
relationship in tables: inverse
the upper row increases from left to right and the lower row decreases from left to right
SPSS
a set of statistical computer routines: Statistical Package for the Social Sciences; click on Type in Data on opening screen; type in data first or create own variable names; click on variable view bottom left of screen;
control variable
a third variable that may have an influence on the relationship between the first two variables; control groups are usually given no experimentation or restraints; two outcomes occur when entering a control variable: 1) has no impact on initial relationship; 2) the presence of the cv changes the initial relationship or is necessary for there to be a relationship between the independent and dependent variables
cross-tabs: interpreting
don't focus on single categories of the IV, but instead on the relationship between the two columns; always read a cross-tab from left to right, never top to bottom; also look at whether it's positive or negative (increases or decreases in DV); positive? (high to low, low to high) negative? the opposite occurs; sometimes need to ignore the 4th row if a very small percentage; if only a slight difference among DV categories = no relationship; the larger the table the more difficult it is to analyze, keep them small (5x5 or less); collapse interval level data into 3-4 categories with high frequencies for cross-tabs that make sense
partial tables
enable one to investigate the impact of a third variable on the relationship between two other variables; provide a very thorough way of studying the impact of a control variable, but tables are vague
contingency table: title
first, name the dependent variable, followed by a comma and the word by, and then name the independent variable; other supporting info like year and location of the study may also be inserted parenthetically; ex: Political Rights, by Per Capita GDP
contingency table: percentaging
percent means "per 100"; make the individual cell entries add up to a common base, a total of 100%; to get a percent, each frequency is divided by the total in the distribution and multiplied by 100.
contingency table: construction
place the independent variable's categories into columns and the categories of the dependent variable into rows; the independent variable is placed along the top, the dependent variable placed along the left side; only report percentages, not frequencies; consider dividing by 100 or 1000 if dollar amounts are very large; place tally marks for each data listed; if a clear clustering appears along the main diagonal of the table, then a positive relationship exists between the variables; replace tally marks with numerical sums; change the table from frequencies to percentages upon completion before interpreting the data; only columns add up to 100, not rows; we only care about the effect the IV has on the DV, not the other way around.
causal models
schematic diagrams showing the independent, dependent, and control variables and, where appropriate, positing the flow of causation of change in the dependent variable; p. 174
contingency table (cross-tabs/tables)
table that depicts a possible relationship between the independent variable and the dependent variable; designed to test a hypothesized relationship between two grouped variables, usually nominal or ordinal level of measurement; each variable grouped into categories; no more than 5 categories in a cross-tab; ALWAYS change frequencies to percents before comparing
relationship in tables: none
the percentages should stay the same (or nearly) across all rows