TAMU PSYC 330 - Research Methods Cont'd (2 pages)

Unformatted text preview:

PSYC 330 1st Edition Lecture 3 Outline of Last Lecture I Characteristics of a theory Outline of Current Lecture II Sampling Strategies III Interaction Effect IV Reliability and Correlation a Methods of estimating reliability i Test retest ii Internal consistency split half Current Lecture Sampling Strategies Sample a Population o Want to be able to generalize conclusions so the sample should represent the population Control person and situation variables o Planned uniformity You can only generalize something to one group o Systematic variation o Randomization Interaction effect when two variable combine to produce an effect they have to be considered in combination to predict an outcome Interaction the relationship between one variable and another variable depends on a third variable means it depends A classic interaction effect will look like an X Moderator Variable another way of saying there is an interaction present its another way of describing an interaction Reliability and Correlation Reliability ability of a measure to yield consistent results o Consistency doesn t mean accurate Ex Playing darts you consistently hit the board with the dart but you fail to hit the bulls eye Thus you re consistent but not accurate Consistency is measured by correlation If the measure is reliable the correlation will be high with a low reliability the measure or correlation will be zero Reliability is measured form 0 1 you should never have a negative number This consistency typically calculating the correspondence between 2 measurements All involve some form of correlation coefficient Methods of estimating reliability Test retest Give a measure once then give it a second time in the future Calculate the correlation between the 2 Internal consistency split half Instead of giving the measure twice correlate the two halves of the measure Tells you that the different items are consistency measuring the same thing Scorer The two lists of numbers being correlated are listed proved