1

I am studying on cognition and behaviors. I am lecturing linguistic courses and teaching linguistic terminology with visuals. In my research, just before each course, I asked participants 20 behavioral yes-no questions like "have you ever been bitten by a dog", or "have your ever watched Seinfel". Then I start my couse and throughout the course, I showed them some pictures which is related with linguistic terminology. These visuals are in accordance with these questions like "a picture of a pittbull" or "a scene from Seinfeld". Then at the end of the course, they are required to place these pictures on certain linguistic terminology they are related. I wonder about their retention, which pictures they remember and place correctly. My hypothesis is that 20 questions that I asked them is effective on their choices of visuals. For instance, watching seinfeld make it easier to remember it again. I prepared my data in SPSS, yes-no questions are 1-0, and their retention test is as wrong and correct, 1-2. How can I measure the cause and effect relationship between these dichotomous variables, which test should I use to measure the effect of 20 behavioral items on correct and wrong retention. I have 16 participants. Please, I really need help, I know I cannot bivariate. Thank you very much.

Emrah Dolgunsöz
  • 85
  • 7

1 Answers1

1

It sounds like your dependent variable is whether they get the placement right, your IV is whether they responded "yes" to the question at the beginning of the term. That would be a simple logistic regression. You could do one of those for each of the 20 questions. But perhaps a better method is use a nonlinear mixed model and include the task as an IV and the person as a random variable.

Peter Flom
  • 119,535
  • 36
  • 175
  • 383
  • Thank you so much Peter. I think I cannot employ nonlinear mixed model in spss. I have only 1 dependen and one independent variable. In this case, can I go through simple logistic regression, is is enough for me? Thank you in advance. – Emrah Dolgunsöz May 19 '14 at 10:58
  • Sure. But you'd have to do one for each question (20 in all). – Peter Flom May 19 '14 at 11:02
  • Peter sorry for bothering you again, I am abit confused today. Some people say that I can use chi square but I still insist on logistic regression. Indeed my DV and IV are both dichotomous values. What do you think? Thank you again. – Emrah Dolgunsöz May 20 '14 at 13:36
  • Chi-square does not treat one variable as DV and the other as IV. Both are treated equally – Peter Flom May 20 '14 at 13:42
  • Now it is much more clear, thanks Peter. One more question: I have 20 questions and all of them are dichotomous. Can I transform them to a single variable or should I analyze each question alone. Analysis for each 20 questions are too long to report; well, can I put all questions and tests one under the other so that I shall have 2 variables but nearly too many cases then. I have 16 participants. What do you think? Thank you in advance. – Emrah Dolgunsöz May 21 '14 at 14:07