研究法簡介 何明洲 中山醫學大學心理系
Single Factor – Two Levels
Independent groups design: use random assignment –IV, manipulated –Between-subject Matched groups design: use matching procedure –IV, manipulated –Between-subject
Single Factor – Two Levels Nonequivalent groups design –IV, subject –Still need to match participants –Between-subject Repeated-measures design –IV, manipulated –Within-subject
Analyzing single factor and two- level design t test for independent groups and dependent groups
EXAMPLE: THE t AND F TESTS t value is a ratio of two aspects of the data, the difference between the group means and the variability within groups t = group difference within group variability
Multilevel designs 如果只有 2 levels (e.g., no reward and $4) , 資訊量太少, 可能誤判為 linear Add more levels for replication and extension
LINEAR VERSUS POSITIVE MONOTONIC FUNCTIONS
Analyzing Single-Factor, >2 levels F Test (analysis of variance, ANOVA 變異 數分析 ) ≥ 2 conditions (or groups) When 2 conditions, F = t 2 When more than 2 conditions, why not use t test?
EXAMPLE: F TEST 用 t test 作多重比較, 至少出現一個 type I error 機率 1 – (1 – alpha) c (C: # of paired comparisons) 噪音程度 ( 無, 低, 中, 高 ) 對記憶的影響 –C = 4!/(2!2!) = 6 –1-(1-.05) 6 =.26 (=26%!!!) F test 同時比較多組, alpha 控制在.05 H 0 : μ 1 = μ 2 = μ 3 = μ 4 …
EXAMPLE: F TEST Total variance = systematic variance + error variance Systematic variance: deviation of group means from the grand mean between-group variance Grand mean: 全部皆平均, 假設無任何因 IV 所引發的 差異 Error variance: deviation of individual scores in each group from their respective group means within-group variance
EXAMPLE: F TEST Total variance = systematic variance + error variance 變異的程度反應整個實驗情境和總人數所帶有 的變異程度 實驗情境越多, 總人數越多, 則變異程度大 不能只考慮變異程度, 須考慮平均每個實驗情 境和人數, 所帶有的變異程度 F = (systematic variance/df systematic ) / (error variance/df error )
Factorial Designs > 1 IVs
INCREASING THE NUMBER OF INDEPENDENT VARIABLES: FACTORIAL DESIGNS > 1 IVs Factorial Designs 多因子設計 : Designs with more than one independent variable (or factor) 噪音(高 vs. 低)影響雙字詞記憶 噪音(高 vs. 低) x 詞頻(高 vs. 低)
Presentation rate 2-sec4-sec Type of training Imagery Rote Factorial matrix
Factor B B1B2 Factor AA1A1B1A1B2 A2A2B1A2B2 Factorial matrix
INCREASING THE NUMBER OF INDEPENDENT VARIABLES: FACTORIAL DESIGNS Simplest Factorial Design 2 x 3x 4 (two-by-two) factorial design Has two independent variables, each IV has 2 levels 4 conditions Number of levels of first IV x Number of levels of second IV x Number of levels of third IV…
INCREASING THE NUMBER OF INDEPENDENT VARIABLES: FACTORIAL DESIGNS Interpretation of Factorial Designs (A x B) Main effects of an independent variable : effect of A factor ONLY (regardless of B factor, average out B factor) Interaction between the independent variables (how does effect of A factor vary with B factor?) ,條件機率 A (A1, A2) x B (B1, B2) 使用圖表讓讀者瞭解
詞頻 高低 噪音程 度 高 A1B1A1B2A1 低 A2B1A2B2A2 B1B2
詞頻 高低 噪音程度 高 AB 低 CD A - B = C – D 詞頻的效果是否隨者噪音程度改變 A – C = B – D 噪音效果是否隨者詞頻程度改變
INCREASING THE NUMBER OF INDEPENDENT VARIABLES: FACTORIAL DESIGNS 詞頻 高低 噪音程度 高 105 低 Interaction NS Main effect of 噪音 NS Main effect of 難度 *
INCREASING THE NUMBER OF INDEPENDENT VARIABLES: FACTORIAL DESIGNS 詞頻 高低 噪音程度 高 510 低 Interaction * Two main effects NS
INCREASING THE NUMBER OF INDEPENDENT VARIABLES: FACTORIAL DESIGNS Type of question misleadingunbiased Knowledge naive1813 knowledgeable Interaction * Two main effects *
INCREASING THE NUMBER OF INDEPENDENT VARIABLES: FACTORIAL DESIGNS
2 x 3 x 4 factorial design –How many IVs? –How many levels of each IV –How many total conditions –How many DVs?
Varieties of Factorial Design Mixed factorial design: 有 between and within subject variables, 沒有 subject variable P x E factorial designs: 有 subject variable 和 manipulated variable (均為 between) Mixed P x E factorial : 有 subject variable 和 manipulated variable (有 between 和 within)
INCREASING THE NUMBER OF INDEPENDENT VARIABLES: FACTORIAL DESIGNS Interactions and Simple Main Effects( 單純主要效 果 ) 當有 interaction 時,必作的統計分析 Simple main effect: examine mean differences at each level of the independent variable 依研究目的決定要作哪些 simple main effect
INCREASING THE NUMBER OF INDEPENDENT VARIABLES: FACTORIAL DESIGNS Interactions and Simple Main Effects Simple main effect of B1: A1|B1 vs. A2|B1 Simple main effect of A1: B1|A1 vs. B2|A1
Anxiety level LowModerateHigh Task Difficulty Easy4710 Hard741 (Easy vs. Hard)|Low (L vs.M vs.H)|Easy
INCREASING THE NUMBER OF VARIABLES: FACTORIAL DESIGNS
INCREASING THE NUMBER OF INDEPENDENT VARIABLES: FACTORIAL DESIGNS Increasing the Number of Independent Variables in a Factorial Design 2 x 2 x 2 噪音高低 x 作業難易 x 性別
男 女 作業難易 噪音高低噪音高低 噪音高低噪音高低 噪音高低 x 作業難易 x 性別 3-way interaction 2-way interaction ( 男生中, 噪音高 低 x 作業難易 ) simple simple main effect
Presenting data Text Table Figure
Discrete and continuous variable
Continuous variable
圖表 運用之法,存乎一心,沒有絕對對錯, 重要的是「好理解」且「點出重點」
點出文章重點
圖表尺度的影響 1. 注意尺度 2. 加上信賴區間