I have about 80-100 participants and a single dichotomous predictor as fail/pass. My outcome variable is a continuous one and I want to know if my dichotomous predictor predicts the continuous variable and to what extent. I will go for lineer regression as my outcome is continuous, do you think my sample size is enough? Thanks lot. Predictor variable = knowing the meaning of a word or not (1-0) Predicted outcome = how long that word is focused by the participants.

  • 1
    $\begingroup$ Well that depends of the samples (noise and etc.), but that requires also testing and validation, it cannot be answered it is enough of not. $\endgroup$ – badc0re Nov 5 '14 at 10:43

With a predictor that can take two values (0,1) you're basically comparing the mean duration of focus between those who have 1 and those who have 0. That is, the prediction for those who have 1 is the mean of that group, likewise for those with 0.

As indicated by badc0re, whether or not the sample size is sufficient depends on whether the standard assumption of OLS are satisfied, check out this link. The first two assumptions are critical (correct specification and strict exogeneity of the error term), if they are then sample size has more to do with efficiency (variance of the error).

  • $\begingroup$ Thanks a lot badc0re and alanw. Unfortunately i am not a mathematician but a linguist. Is there a practical way to test these assumptions in spss? $\endgroup$ – Emrah Dolgunsöz Nov 5 '14 at 17:12

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.