Psych 225 Lecture 8 Outline of Last Lecture 1) Exam prep and questions2) Contrast t and LSD3) HSD4) A priori vs. Post Hoc tests Outline of Current Lecture 1) Impact of Intervention on Depression Study2) Decrease in error variance3) Stroop differences between Study 1 and 2Current LectureSlide: Impact of Intervention on Depression-P’s received actual treatment versus no treatment ~those on waiting list (no treatment)= +1 and all other conditions (4) had -.25 *see Course Packet on pg. 31-32-combo of condition compared to control produced a very small difference -Ps who received pharmacological intervention will be less depressed than those who received psychotherapy: ~those on waiting list (no treatment) =0 and all other conditions: psychotherapy means were -.5 and pharmalogical means were both +.5*for both of studies the degrees of freedom are 35: look at within (entirety) -why aren’t standard errors (denominator of t) the same?Answer: denominator changes depending on how many variables they have ~ex. in first example there were 8 scores (look at N column in course packet output) These notes represent a detailed interpretation of the professor’s lecture. GradeBuddy is best used as a supplement to your own notes, not as a substitute.~ex. made all other means combined: 32 scores (add up 4 conditions in N column in course packet output) -P’s received CBT will be less depressed than those who received psychodynamic psychotherapy -looking at 2 conditions -difference in CBT and psychodynamic mean is 5.75 which= value of contrast *did not receive support for this finding Pg. 24 in Course Packet (based off original Stroop study)-shows how to increase the power of the F test -original df= 2, 12 1st example: to increase degrees of freedom between we added level to the IV (4 levels instead of 3): thus the F crit is lowered; we don’t have to have as large of an F or as big of a difference between means to find a significant effect, it is more powerful ~df: 3, 122nd example: increasing the number of participants give us a less variable sampling distribution (increases degrees of freedom within and decreases F crit) ~df: 2, 153rd example: increasing the strength of the IV increases the Mean Squares Between which increases the F calculated4th example: decrease in sample variance decreases the Mean Squares Within which increases the F calculated To decrease error variance-when experiments are highly automated, you tend to have less error variance -do within subjects design instead of a between because (you get a handle on participants) -increase specificity of eligibility requirements for participants (make them more uniform)Stroop Study 2: Does type of presentation..pg. 25 in Course PacketStroop Studies 1 vs. 2 (comparing female intro psych students (study 1) to female volunteers in Madison of any age (study 2))1) Standard Deviations: standard deviation would be much larger in the second study 2) Mean Square Within: lower in study 2 (SS within divided by df within) 3) Sum of Squares total, between and within: total: vary more in study 2 (larger) between: no changewithin: larger in study 2 4) Average of the sample variance: larger in 25) Standard Deviation: total larger in study 2 6) dft, dfB, dfWdegrees of freedom total: samedegrees of freedom Between: samedegrees of freedom Within: same7) Sampling distribution of F: same8) F critical: same9) Mean Square Between: same (SS between divided by df Between) 10) F calculated: smaller in study 2 because we increased MSW11) p (obtaining F calc that large or larger if H null is true): larger in study 2 12) Standard error (denominator of t) for tests of ~Hypothesis 1: larger for study 2~Hypothesis 2: larger for study 213) t calc for H1 & H2: smaller t in study 2 14) p (mean difference that large or larger if H null is true): larger in study 2Course Packet pg. 11 and 12-pg. 11: Between Subjects-pg. 12: Within Subjects PG. 12SSIV: effectSSPp: for each PpSSE: random dfT: N-1dfIV: # levels-1 dfPp: #Pp-1dfE:dfT-dfIV-dfPp or (dfIVxdfPp?)Pg. 30 -focus on sphericity assumed -despite reduction in degrees of freedom we have more powerful test because of reduction of Mean Square
View Full Document